1 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2 Dominik Vogt <vogt@linux.vnet.ibm.com>
5 * emit-rtl.c (init_emit): Remove ??? comment.
6 * explow.c (get_dynamic_stack_size): Take known alignment of stack
7 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
8 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
9 alignment of 3 virtual registers to BITS_PER_WORD.
11 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
13 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
15 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
16 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
18 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
21 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
23 2017-12-13 Jakub Jelinek <jakub@redhat.com>
25 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
26 with the exception of debug begin stmt markers appear before
30 * final.c (rest_of_handle_final): Call variable_tracking_main only
31 if !flag_var_tracking.
33 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
34 Jakub Jelinek <jakub@redhat.com>
38 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
39 labels debug stmts that can only appear after labels.
41 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
43 PR rtl-optimization/82398
44 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
45 EXPR_USEFULNESS in priority comparison.
47 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
49 PR rtl-optimization/83393
50 * combine.c (move_deaths): If reg_stat points to a too new insn in
51 last_death, do not use it: find the proper insn instead.
53 2017-12-12 Jeff Law <law@redhat.com>
55 PR tree-optimization/83298
56 PR tree-optimization/83362
57 PR tree-optimization/83383
58 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
59 push_value_range a public interface. Add new argument to
60 record_ranges_from_stmt.
61 * gimple-ssa-evrp-analyze.c
62 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
63 Update comments. Handle recording temporary equivalences.
64 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
65 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
66 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
67 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
68 gimple-ssa-evrp-analyze.h.
69 (record_temporary_equivalences_from_phis): Add new argument. When
70 the PHI arg is an SSA_NAME, set the result's range to the range
72 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
74 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
75 Pass it down to children as needed.
76 (thread_outgoing_edges): Likewise.
77 (thread_across_edge): Likewise. Push/pop range state as needed.
78 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
80 2017-12-12 Julia Koval <julia.koval@intel.com>
82 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
83 (PTA_CANNONLAKE): Remove PTA_CLWB.
85 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
88 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
89 (vcondv2div2df): Likewise.
90 (vconduv2dfv2di): Likewise.
92 2017-12-12 Jakub Jelinek <jakub@redhat.com>
94 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
95 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
96 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
97 ATTR_MATHFN_FPROUNDING_ERRNO.
99 2017-12-12 Richard Biener <rguenther@suse.de>
101 PR tree-optimization/83385
102 * graphite-scop-detection.c (get_order, order): Remove.
103 (bb_to_rpo): New global.
105 (build_scops): Sort pbbs in RPO order.
107 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
109 * combine.c (simplify_set): Do not transform subregs to zero_extends
110 if the destination mode is a vector mode.
112 2017-12-12 Jakub Jelinek <jakub@redhat.com>
114 PR tree-optimization/83359
115 * tree-cfg.h (fold_loop_internal_call): Declare.
116 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
117 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
118 (find_loop_dist_alias): New function.
119 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
120 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
123 PR tree-optimization/80631
124 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
125 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
126 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
127 hardcoding zero as the value if COND_EXPR is never true. For
128 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
129 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
130 hardcoding MAX_EXPR as the reduction operation.
131 (is_nonwrapping_integer_induction): Allow negative step.
132 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
133 vect_create_epilog_for_reduction, if no value is suitable, don't
134 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
136 2017-12-12 Richard Biener <rguenther@suse.de>
138 PR tree-optimization/81889
139 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
140 range info from the non-wrapping IV instead of just the range
143 2017-12-12 Julia Koval <julia.koval@intel.com>
145 * config.gcc: Add vaesintrin.h.
146 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
148 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
149 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
151 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
152 * config/i386/immintrin.h: Include vaesintrin.h.
153 * config/i386/sse.md (vaesdec_<mode>): New pattern.
154 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
155 _mm_aesdec_epi128): New intrinsics.
157 2017-12-12 Julia Koval <julia.koval@intel.com>
159 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
160 OPTION_MASK_ISA_VAES_UNSET): New.
161 (ix86_handle_option): Handle -mvaes.
162 * config/i386/cpuid.h: Define bit_VAES.
163 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
164 * config/i386/i386-c.c (__VAES__): New.
165 * config/i386/i386.c (ix86_target_string): Add -mvaes.
166 (ix86_valid_target_attribute_inner_p): Ditto.
167 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
168 * config/i386/i386.opt: Add -mvaes.
169 * doc/invoke.texi: Ditto.
171 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
173 * debug.h (gcc_debug_hooks): Add inline_entry.
174 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
175 * debug.c (do_nothing_debug_hooks): Likewise.
176 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
177 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
178 (dwarf2_lineno_debug_hooks): Likewise.
180 * common.opt (gstatement-frontiers): New, setting
181 debug_nonbind_markers_p.
182 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
183 * toplev.c (process_options): Autodetect value for debug statement
185 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
186 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
188 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
189 markers. Integrate source bind into debug stmt expand loop.
190 (pass_expand::execute): Check debug marker limit. Avoid deep
191 TER and expand debug locations for debug bind insns only.
192 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
194 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
195 * final.c (reemit_insn_block_notes): Take current block from
196 nonbind markers. Declare note where it's first set.
197 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
198 begin stmt markers if enabled.
199 (notice_source_line): Handle nonbind markers. Fail if their
200 location is unknown or that of builtins.
201 (rest_of_handle_final): Convert begin stmt markers to notes if
202 var-tracking didn't run.
203 (rest_of_clean_state): Skip begin stmt markers.
204 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
206 * function.c (allocate_struct_function): Set begin_stmt_markers.
207 * function.h (struct function): Add debug_marker_count counter
208 and debug_nonbind_markers flag.
209 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
210 * gimple-low.c (lower_function_body): Adjust
211 debug_nonbind_markers.
212 (lower_stmt): Drop or skip gimple debug stmts.
213 (lower_try_catch): Skip debug stmts.
214 * gimple.c (gimple_build_debug_begin_stmt): New.
215 (gimple_copy): Increment debug_marker_count if copying one.
216 * gimple.h (gimple_build_debug_begin_stmt): Declare.
217 * gimplify.c (rexpr_location): New.
218 (rexpr_has_location): New.
219 (warn_switch_unreachable_r): Handle gimple debug stmts.
220 (shortcut_cond_r): Call expr_location.
222 (find_goto_label): New.
223 (shortcut_cond_expr): Call expr_has_location, expr_location, and
225 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
227 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
228 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
229 (LANG_HOOKS_INITIALIZER): ... this.
230 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
231 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
233 (update_ebb_live_info): Skip debug insn markers.
234 * lra.c (debug_insn_static_data): Rename to...
235 (debug_bind_static_data): ... this.
236 (debug_marker_static_data): New.
237 (lra_set_insn_recog_data): Select one of the above depending
239 (lra_update_isn_regno_info): Don't assume debug insns have
241 (push_insns): Skip debug insns.
242 * lto-streamer-in.c (input_function): Drop debug stmts
243 depending on active options. Adjust debug_nonbind_markers.
244 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
245 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
246 begin stmt marker notes.
247 (print_insn): Likewise.
248 * recog.c (extract_insn): Recognize rtl for debug markers.
249 * rtl.def (DEBUG_MARKER): New.
250 * tree-inline.c: Include params.h.
251 (remap_gimple_stmt): Handle nonbind markers.
252 (maybe_move_debug_stmts_to_successors): Likewise.
253 (copy_debug_stmt): Likewise.
254 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
255 markers regardless of no side effects.
256 (tsi_link_before): Don't update container's side effects when adding
258 (tsi_link_after): Likewise.
259 (expr_first): Skip begin stmt markers.
260 (expr_last): Likewise.
261 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
262 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
263 Disregard nonbind markers.
264 * tree.c (make_node_stat): Don't set side effects for begin stmt
266 (build1_stat): Likewise.
267 * tree.def (DEBUG_BEGIN_STMT): New.
268 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
269 * var-tracking.c (delete_debug_insns): Renamed to...
270 (delete_vta_debug_insns): ... this.
271 (reemit_marker_as_note): New.
272 (vt_initialize): Reemit markers.
273 (delete_vta_debug_insns): Likewise.
274 (vt_debug_insns_local): Reemit or delete markers.
275 (variable_tracking_main_1): Likewise.
276 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
277 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
278 (gimple_debug_nonbind_marker_p): New.
279 (gimple_build_debug_bind): Adjust.
280 (gimple_build_debug_begin_stmt): New.
281 * doc/invoke.texi (max-debug-marker-count): New param.
282 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
283 (debug_parameter_ref, debug_marker): New.
284 (NOTE_INSN_BEGIN_STMT): New.
285 (DEBUG_INSN): Describe begin stmt markers.
287 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
288 without debug insns, we wouldn't, but clean up debug insns
289 after a control flow insn nevertheless.
291 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
292 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
293 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
294 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
295 (rtl_tidy_fallthru_edge): Likewise.
296 (rtl_verify_fallthru): Likewise.
297 (rtl_verify_bb_layout): Likewise.
298 (skip_insns_after_block): Likewise.
299 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
300 * dwarf2out.c: Include print-rtl.h.
301 (dwarf2out_next_real_insn): New.
302 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
303 Dump debug binds in asm comments.
304 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
305 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
306 callers to use gsi_start_nondebug_bb instead.
307 (gsi_after_labels): Skip gimple debug stmts.
308 (gsi_start_nondebug): New.
309 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
310 (proper_loop_form_for_interchange): Adjust.
311 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
312 * gimple.h (gimple_seq_last_nondebug_stmt): New.
313 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
314 (collect_fallthrough_labels): Likewise.
315 (should_warn_for_implicit_fallthrough): Likewise.
316 (warn_implicit_fallthrough_r): Likewise.
317 (expand_FALLTHROUGH_r): Likewise.
318 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
319 (graphite_copy_stmts_from_block): Skip nonbind markers.
320 * haifa-sched.c (sched_extend_bb): Skip debug insns.
321 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
322 * jump.c (clean_barriers): Skip debug insns.
323 * omp-expand.c (expand_parallel_call): Skip debug insns.
324 (expand_task_call): Likewise.
325 (remove_exit_barrier): Likewise.
326 (expand_omp_taskreg): Likewise.
327 (expand_omp_for_init_counts): Likewise.
328 (expand_omp_for_generic): Likewise.
329 (expand_omp_for_static_nochunk): Likewise.
330 (expand_omp_for_static_chunk): Likewise.
331 (expand_omp_simd): Likewise.
332 (expand_omp_taskloop_for_outer): Likewise.
333 (expand_omp_taskloop_for_inner): Likewise.
334 (expand_oacc_for): Likewise.
335 (expand_omp_sections): Likewise.
336 (expand_omp_single): Likewise.
337 (expand_omp_synch): Likewise.
338 (expand_omp_atomic_load): Likewise.
339 (expand_omp_atomic_store): Likewise.
340 (expand_omp_atomic_fetch_op): Likewise.
341 (expand_omp_atomic_pipeline): Likewise.
342 (expand_omp_atomic_mutex): Likewise.
343 (expand_omp_target): Likewise.
344 (grid_expand_omp_for_loop): Likewise.
345 (grid_expand_target_grid_body): Likewise.
346 (build_omp_regions_1): Likewise.
347 * omp-low.c (check_combined_parallel): Skip debug stmts.
348 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
349 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
350 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
352 * tree-cfg.c (make_blobs_1): Skip debug stmts.
353 (make_edges): Likewise.
354 (cleanup_dead_labels): Likewise.
355 (gimple_can_merge_blocks_p): Likewise.
356 (stmt_starts_bb_p): Likewise.
357 (gimple_block_label): Likewise.
358 (gimple_redirect_edge_and_branch): Likewise.
359 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
361 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
363 * tree-pretty-print (print_declaration): Omit initializer in slim
365 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
367 (eliminate_unnecessary_stmts): Stabilize block removal order.
368 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
369 * var-tracking.c (get_first_insn): New.
370 (vt_emit_notes): Call it.
371 (vt_initialize): Walk any insns before the first BB.
372 (delete_debug_insns): Likewise.
374 * gimple.h (enum gimple_debug_subcode): Add
375 GIMPLE_DEBUG_BEGIN_STMT.
376 (gimple_debug_begin_stmt_p): New.
377 (gimple_debug_nonbind_marker_p): New.
378 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
379 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
380 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
381 * insn-notes.def (BEGIN_STMT): New.
382 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
383 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
384 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
385 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
386 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
387 (INSN_DEBUG_MARKER_KIND): New.
388 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
389 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
390 (INSN_VAR_LOCATION_PTR): New.
391 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
393 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
394 deps only if debug bind insns are enabled.
395 (pass_expand::execute): Avoid deep TER and expand
396 debug locations for debug bind insns only.
397 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
398 debug stmts special handling down to debug bind stmts.
399 * combine.c (try_combine): Narrow debug insns special handling
400 down to debug bind insns.
401 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
402 Narrow debug insns preexisting special handling down to debug
404 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
405 handling down to debug bind insns.
406 * function.c (instantiate_virtual_regs): Skip debug markers,
407 adjust handling of debug binds.
408 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
409 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
410 * haifa-sched.c (schedule_insn): Narrow special handling of debug
411 insns to debug bind insns.
412 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
413 special handling of debug stmts to debug bind stmts.
414 * ipa-split.c (split_function): Likewise.
415 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
416 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
418 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
419 * regrename.c (build_def_use): Likewise.
420 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
421 (pass_cprop_hardreg): Narrow special casing of debug insns to
423 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
424 * reload1.c (reload): Likewise.
425 * sese.c (sese_insert_phis_for_liveouts): Narrow special
426 casing of debug stmts to debug bind stmts.
427 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
428 * ssa-iterators.h (num_imm_uses): Likewise.
429 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
430 debug stmts to debug bind stmts.
431 * tree-inline.c (tree_function_versioning): Narrow special casing
432 of debug stmts to debug bind stmts.
433 * tree-loop-distribution.c (generate_loops_for_partition):
434 Narrow special casing of debug stmts to debug bind stmts.
435 * tree-sra.c (analyze_access_subtree): Narrow special casing
436 of debug stmts to debug bind stmts.
437 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
438 stmts to debug bind stmts.
439 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
440 casing of debug stmts to debug bind stmts.
441 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
442 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
443 casing of debug stmts to debug bind stmts.
444 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
446 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
447 debug stmts to debug bind stmts.
448 (gimple_replace_ssa_lhs): Likewise.
449 (insert_debug_temp_for_var_def): Likewise.
450 (insert_debug_temps_for_defs): Likewise.
451 (reset_debug_uses): Likewise.
452 * tree-ssanames.c (release_ssa_name_fn): Likewise.
453 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
454 (adjust_debug_stmts): Likewise.
455 (adjust_phi_and_debug_stmts): Likewise.
456 (vect_do_peeling): Likewise.
457 * tree-vect-loop.c (vect_transform_loop): Likewise.
458 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
459 * var-tracking.c (adjust_mems): Narrow special casing of debug
460 insns to debug bind insns.
461 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
462 (compute_bb_dataflow, vt_find_locations): Likewise.
463 (vt_expand_loc, emit_notes_for_changes): Likewise.
464 (vt_init_cfa_base): Likewise.
465 (vt_emit_notes): Likewise.
466 (vt_initialize): Likewise.
467 (vt_finalize): Likewise.
469 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
470 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
471 (next_nonnote_nondebug_insn_bb): New.
472 (prev_nonnote_nondebug_insn_bb): New.
473 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
474 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
475 (prev_nonnote_nondebug_insn_bb): Declare.
476 (next_nonnote_nondebug_insn_bb): Declare.
477 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
478 * cfgrtl.c (get_last_bb_insn): Likewise.
479 * lra.c (push_insns): Likewise.
481 2017-12-11 David Malcolm <dmalcolm@redhat.com>
484 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
486 * selftest.c (selftest::test_runner::test_runner): New ctor.
487 (selftest::test_runner::~test_runner): New dtor.
488 * selftest.h (class selftest::test_runner): New class.
490 2017-12-11 Carl Love <cel@us.ibm.com>
492 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
493 vec_extract_fp32_from_shortl]): Add #defines.
494 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
495 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
496 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
497 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
498 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
499 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
500 * doc/extend.texi: Add documentation for the added builtins.
502 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
504 PR rtl-optimization/80693
505 PR rtl-optimization/81019
506 PR rtl-optimization/81020
507 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
508 are not mentioned in i3. Place the REG_UNUSED note on i2,
509 possibly modified to REG_DEAD, if it did not originate in i3.
511 2017-12-11 Jakub Jelinek <jakub@redhat.com>
513 * recog.c (store_data_bypass_p_1): New function.
514 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
515 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
517 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
519 PR rtl-optimization/83361
520 * ifcvt.c (if_convert): Call fixup_partitions.
522 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
524 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
525 early folding of splat_u{8,16,32}.
527 2017-12-11 Jakub Jelinek <jakub@redhat.com>
529 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
530 output_operand_lossage first argument with capital letter.
531 (aarch64_override_options): Don't start error and sorry first argument
534 2017-12-11 Andi Kleen <ak@linux.intel.com>
536 PR gcov-profile/83355
537 * auto-profile.c (string_table::get_index_by_decl): Don't
538 recurse when abstract origin points to itself.
540 2017-12-11 Bin Cheng <bin.cheng@arm.com>
542 PR tree-optimization/83320
543 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
544 (prune_datarefs_not_in_loop): Ditto.
546 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
548 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
550 2017-12-10 Jakub Jelinek <jakub@redhat.com>
552 PR tree-optimization/83337
553 * gimple-loop-interchange.cc (compute_access_stride): Handle
554 bitfield DRs properly.
556 2017-12-09 Jakub Jelinek <jakub@redhat.com>
558 PR tree-optimization/83338
559 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
560 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
561 vector of unsigned integers to vector of signed integers.
563 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
565 PR rtl-optimization/83317
566 * lra-constraints.c (process_address_1): Add insn code check.
568 2017-12-08 Michael Matz <matz@suse.de>
570 Fix PR tree-optimization/83323
571 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
572 head-controlled loops and loop BBs.
573 * common.opt (funroll-and-jam): Remove, instead ...
574 (floop-unroll-and-jam): ... reuse this option.
575 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
576 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
577 (-floop-unroll-and-jam): ... this option.
579 2017-12-08 Jakub Jelinek <jakub@redhat.com>
581 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
582 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
583 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
585 2017-12-08 Richard Biener <rguenther@suse.de>
588 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
589 handle accesses outside of zero-sized vars.
591 2017-12-08 Martin Jambor <mjambor@suse.cz>
593 PR tree-optimization/83141
594 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
595 test for MEM_REFs implicitely changing types with padding. Remove
597 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
599 2017-12-08 Jakub Jelinek <jakub@redhat.com>
601 * config/arc/arc.c (arc_attribute_table): Add exclusions to
603 * config/avr/avr.c (avr_attribute_table): Likewise.
604 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
605 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
606 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
607 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
608 of struct attribute_spec.
609 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
611 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
614 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
616 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
619 * profile-count.c (profile_count::from_gcov_type): Move from
620 profile-count.h; handle overflow.
621 * profile-count.h (profile_count::from_gcov_type): Move offline.
623 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
625 PR rtl-optimization/83304
626 * combine.c (move_deaths): If we do not know where a register died,
629 2017-12-08 Richard Biener <rguenther@suse.de>
631 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
632 Provide -fopt-info-loop feedback when we interchange in a nest.
634 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
636 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
637 for armv6 ARM CPU IDs.
639 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
641 * common/config/arm/arm-common.c: Include <algorithm>.
642 (INCLUDE_VECTOR): Define.
643 (compare_opt_names): New function.
644 (arm_rewrite_selected_arch): Only strip out extensions that can be
645 expressed through -mfpu. Sort the remaining extensions
648 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
650 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
651 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
652 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
653 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
655 2017-06-08 Tristan Gingold <gindold@adacore.com>
658 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
659 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
661 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
663 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
664 of struct attribute_spec.
666 2017-12-08 Julia Koval <julia.koval@intel.com>
668 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
669 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
670 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
671 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
672 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
673 _mm_maskz_dpwssds_epi32): Ditto.
675 2017-12-08 Richard Biener <rguenther@suse.de>
677 PR tree-optimization/81303
678 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
679 conditions try to create a comparison vector type matching
680 the data vector type.
681 (vectorizable_condition): Adjust.
682 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
683 Leave invariant conditions alone in case we can vectorize those.
685 2017-12-08 Julia Koval <julia.koval@intel.com>
687 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
688 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
689 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
690 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
691 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
693 2017-12-08 Julia Koval <julia.koval@intel.com>
695 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
696 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
697 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
698 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
699 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
700 _mm_maskz_dpbusds_epi32): New intrinsics.
702 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
704 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
706 (ld<bh>io_signed): Likewise.
707 (st<bhw>io): Likewise.
708 * config/nios2/predicates.md (ldstio_memory_operand): Allow
709 SMALL_INT12 constant integer operand.
711 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
713 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
715 (nios2_large_constant_allowed): ...this. Adjust uses.
716 (nios2_plus_symbolic_constant_p): Rename to...
717 (nios2_plus_large_constant_p): ...this. Adjust uses.
718 (nios2_legitimate_address_p): Correct CONST_INT handling.
719 (nios2_symbolic_memory_operand_p): Rename to...
720 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
721 (nios2_large_constant_p): Check for large constant integers too.
722 (nios2_split_large_constant): Handle constant integers.
723 (nios2_split_symbolic_memory_operand): Rename to...
724 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
725 (nios2_legitimize_constant_address): Handle constant integers.
726 (r0rel_constant_p): Handle small constant integers.
727 (nios2_print_operand_address): Handle r0-relative integer addresses.
728 * config/nios2/nios2-protos.h: Adjust for renamed functions.
729 * config/nios2/nios2.md: Adjust for renamed functions.
731 2017-12-07 Andrew Waterman <andrew@sifive.com>
733 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
734 (TARGET_HAVE_SRODATA_SECTION): New define.
735 (riscv_select_section): New function.
737 2017-12-08 Joseph Myers <joseph@codesourcery.com>
738 Alexander Monakov <amonakov@ispras.ru>
739 Jakub Jelinek <jakub@redhat.com>
742 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
744 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
746 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
747 a bool success value. Don't call output_operand_lossage here.
748 (aarch64_print_ldpstp_address): Return a bool success value.
749 (aarch64_print_operand_address): Call output_addr_const if
750 aarch64_print_address_internal fails.
751 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
752 'y'; call output_operand_lossage instead. Call output_operand_lossage
753 if aarch64_print_ldpstp_address fails.
755 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
757 * tree-vector-builder.h
758 (tree_vector_builder::binary_encoded_nelts): Declare.
759 * tree-vector-builder.c
760 (tree_vector_builder::binary_encoded_nelts): New function.
761 * fold-const.c (negate_expr_p): Likewise.
762 (operand_equal_p, fold_checksum_tree): Likewise.
763 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
764 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
765 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
766 (uniform_vector_p): Likewise.
767 * varasm.c (const_hash_1, compare_constant): Likewise.
768 * tree-ssa-ccp.c: Include tree-vector-builder.h.
769 (valid_lattice_transition): Operate directly on the VECTOR_CST
771 * ipa-icf.c: Include tree-vector-builder.h.
772 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
773 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
775 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
777 * tree.c (build_vector): Delete.
778 * tree.h (build_vector): Make static and move into the self-testing
781 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
783 * vector-builder.h (vector_builder::derived): New const overload.
784 (vector_builder::elt): New function.
785 * tree-vector-builder.h (tree_vector_builder::type): New function.
786 (tree_vector_builder::apply_step): Declare.
787 * tree-vector-builder.c (tree_vector_builder::apply_step): New
789 * gimple-fold.h (tree_vector_builder): Declare.
790 (gimple_build_vector): Take a tree_vector_builder instead of a
791 type and vector of elements.
792 * gimple-fold.c (gimple_build_vector): Likewise.
793 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
795 (get_initial_defs_for_reduction): Likewise.
796 (vectorizable_induction): Likewise.
798 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
800 * tree-vector-builder.h
801 (tree_vector_builder::new_binary_operation): Declare.
802 * tree-vector-builder.c
803 (tree_vector_builder::new_binary_operation): New function.
804 * fold-const.c (fold_relational_const): Use it.
805 (const_binop): Likewise. Check that both input vectors have
806 the same number of elements, thus excluding things like WIDEN_SUM.
807 Check whether it is possible to operate directly on the encodings
810 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
812 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
813 new_unary_operation, operating only on the encoded elements.
814 (const_unop): Likewise.
815 (exact_inverse): Likewise.
816 (distributes_over_addition_p): New function.
817 (const_binop): Use tree_vector_builder and new_unary_operation
818 for combinations of VECTOR_CST and INTEGER_CST. Operate only
819 on the encoded elements unless the encoding is strided and the
820 operation does not distribute over addition.
821 (fold_convert_const): Use tree_vector_builder and
822 new_unary_operation. Operate only on the encoded elements
823 for truncating integer conversions, or for non-stepped encodings.
825 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
827 * config/sparc/sparc.c: Include tree-vector-builder.h.
828 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
829 * expmed.c: Include tree-vector-builder.h.
830 (make_tree): Use tree_vector_builder instead of build_vector.
831 * fold-const.c: Include tree-vector-builder.h.
832 (const_binop): Use tree_vector_builder instead of build_vector.
833 (const_unop): Likewise.
834 (native_interpret_vector): Likewise.
835 (fold_vec_perm): Likewise.
836 (fold_ternary_loc): Likewise.
837 * gimple-fold.c: Include tree-vector-builder.h.
838 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
840 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
841 (simplify_vector_constructor): Use tree_vector_builder instead
843 * tree-vect-generic.c: Include tree-vector-builder.h.
844 (add_rshift): Use tree_vector_builder instead of build_vector.
845 (expand_vector_divmod): Likewise.
846 (optimize_vector_constructor): Likewise.
847 * tree-vect-loop.c: Include tree-vector-builder.h.
848 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
849 of build_vector. Explicitly use a stepped encoding for
851 * tree-vect-slp.c: Include tree-vector-builder.h.
852 (vect_get_constant_vectors): Use tree_vector_builder instead
854 (vect_transform_slp_perm_load): Likewise.
855 (vect_schedule_slp_instance): Likewise.
856 * tree-vect-stmts.c: Include tree-vector-builder.h.
857 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
858 (vect_gen_perm_mask_any): Likewise.
859 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
860 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
862 (build_vector_from_val): Likewise. Explicitly use a duplicate
865 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
867 * doc/generic.texi (VECTOR_CST): Describe new representation of
869 * vector-builder.h: New file.
870 * tree-vector-builder.h: Likewise.
871 * tree-vector-builder.c: Likewise.
872 * Makefile.in (OBJS): Add tree-vector-builder.o.
873 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
874 * tree-core.h (tree_base): Add a vector_cst field to the u union.
875 (tree_vector): Change the number of elements to
876 vector_cst_encoded_nelts.
877 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
878 (VECTOR_CST_ELTS): Delete.
879 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
880 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
881 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
882 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
883 (VECTOR_CST_ENCODED_ELT): Likewise.
884 (vector_cst_encoded_nelts): New function.
885 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
886 VECTOR_CST_NELTS_PER_PATTERN as arguments.
887 (vector_cst_int_elt, vector_cst_elt): Declare.
888 * tree.c: Include tree-vector-builder.h.
889 (tree_code_size): Abort if passed VECTOR_CST.
890 (tree_size): Update for new VECTOR_CST layout.
891 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
892 VECTOR_CST_NELTS_PER_PATTERN as arguments.
893 (build_vector): Use tree_vector_builder.
894 (vector_cst_int_elt, vector_cst_elt): New functions.
895 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
896 encoded elements and then create the vector in the canonical form.
897 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
898 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
899 (tree_c_tests): Call test_vector_cst_patterns.
900 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
902 (hash_tree): Likewise.
903 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
904 (streamer_write_tree_header): Likewise.
905 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
906 (streamer_alloc_tree): Likewise. Update call to make_vector.
907 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
909 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
911 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
912 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
913 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
915 2017-12-07 Bin Cheng <bin.cheng@arm.com>
916 Richard Biener <rguenther@suse.de>
918 PR tree-optimization/81303
919 * Makefile.in (gimple-loop-interchange.o): New object file.
920 * common.opt (floop-interchange): Reuse the option from graphite.
921 * doc/invoke.texi (-floop-interchange): Ditto. New document for
922 -floop-interchange and mention it for -O3.
923 * opts.c (default_options_table): Enable -floop-interchange at -O3.
924 * gimple-loop-interchange.cc: New file.
925 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
926 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
927 * passes.def (pass_linterchange): New pass.
928 * timevar.def (TV_LINTERCHANGE): New time var.
929 * tree-pass.h (make_pass_linterchange): New declaration.
930 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
931 interchange. Record IV before/after increment in new parameters.
932 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
933 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
935 (check_reduction_path): ...New function here.
936 * tree-vectorizer.h (check_reduction_path): New declaration.
938 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
941 PR rtl-optimization/80818
942 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
943 always early clobbered.
944 * lra-lives.c (process_bb_lives): Check input hard regs for early
945 clobbered non-operand hard reg.
947 2017-12-07 Jakub Jelinek <jakub@redhat.com>
950 * tree-cfg.c (verify_gimple_assign_binary): Don't require
951 types_compatible_p, just that TYPE_MODE is the same.
953 2017-12-07 Martin Sebor <msebor@redhat.com>
956 * attribs.c (empty_attribute_table): Initialize new member of
957 struct attribute_spec.
958 (decl_attributes): Add argument. Handle mutually exclusive
959 combinations of attributes.
960 (selftests::test_attribute_exclusions): New function.
961 (selftests::attribute_c_tests): Ditto.
962 * attribs.h (decl_attributes): Add default argument.
963 * selftest.h (attribute_c_tests): Declare.
964 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
965 * tree-core.h (attribute_spec::exclusions, exclude): New type and
967 * doc/extend.texi (Common Function Attributes): Update const and pure.
968 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
969 of struct attribute_spec.
970 * config/arc/arc.c (arc_attribute_table): Same.
971 * config/arm/arm.c (arm_attribute_table): Same.
972 * config/avr/avr.c ( avr_attribute_table): Same.
973 * config/bfin/bfin.c (bfin_attribute_table): Same.
974 * config/cr16/cr16.c (cr16_attribute_table): Same.
975 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
976 * config/h8300/h8300.c (h8300_attribute_table): Same.
977 * config/i386/i386.c (ix86_attribute_table): Same.
978 * config/ia64/ia64.c (ia64_attribute_table): Same.
979 * config/m32c/m32c.c (m32c_attribute_table): Same.
980 * config/m32r/m32r.c (m32r_attribute_table): Same.
981 * config/m68k/m68k.c (m68k_attribute_table): Same.
982 * config/mcore/mcore.c (mcore_attribute_table): Same.
983 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
984 * config/mips/mips.c (mips_attribute_table): Same.
985 * config/msp430/msp430.c (msp430_attribute_table): Same.
986 * config/nds32/nds32.c (nds32_attribute_table): Same.
987 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
988 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
989 * config/rl78/rl78.c (rl78__attribute_table): Same.
990 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
991 * onfig/rx/rx.c (rx_attribute_table): Same.
992 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
993 * config/sh/sh.c (sh_attribute_table): Same.
994 * config/sparc/sparc.c (sparc_attribute_table): Same.
995 * config/spu/spu.c (spu_attribute_table): Same.
996 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
997 * config/v850/v850.c (v850_attribute_table): Same.
998 * config/visium/visium.c (visium_attribute_table): Same.
1000 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1003 * config/arm/arm.c (INCLUDE_STRING): Define.
1004 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1005 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1008 2017-12-07 Michael Matz <matz@suse.de>
1010 Add unroll and jam pass
1012 * gimple-loop-jam.c: New file.
1013 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1014 * common.opt (funroll-and-jam): New option.
1015 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1016 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1017 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1018 * passes.def: Add pass_loop_jam.
1019 * timevar.def (TV_LOOP_JAM): Add.
1020 * tree-pass.h (make_pass_loop_jam): Declare.
1021 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1022 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1023 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1024 to flow_loop_tree_node_add.
1025 (duplicate_subloops, copy_loops_to): Append to sibling list.
1026 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1027 * doc/invoke.texi (-funroll-and-jam): Document new option.
1028 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1030 2017-12-07 Richard Biener <rguenther@suse.de>
1032 PR tree-optimization/83296
1033 PR tree-optimization/67769
1034 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1035 flow sensitive info in an unrelated BB.
1036 (value_replacement): Use reset_flow_sensitive_info.
1037 (minmax_replacement): Reset flow sensitive info on the def
1038 we move. Do not reset flow sensitive info in the whole BB
1039 we move the stmt to.
1040 (abs_replacement): Likewise.
1042 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1045 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1046 rs6000_cpu to the given -mcpu=, or to the default processor.
1048 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1050 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1051 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1052 instead of rs6000_cpu_attr.
1053 (rs6000_adjust_cost): Ditto.
1054 (is_microcoded_insn): Ditto.
1055 (rs6000_adjust_priority): Ditto.
1056 (rs6000_issue_rate): Ditto.
1057 (rs6000_use_sched_lookahead): Ditto.
1058 (rs6000_use_sched_lookahead_guard): Ditto.
1059 (rs6000_sched_reorder): Ditto.
1060 (force_new_group): Ditto.
1061 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1062 (group_ending_nop): Ditto.
1064 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1066 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1067 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1068 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1069 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1071 (rs6000_reassociation_width): Ditto.
1072 (rs6000_emit_epilogue): Ditto.
1073 (rs6000_adjust_cost): Ditto.
1074 (is_microcoded_insn): Ditto.
1075 (is_cracked_insn): Ditto.
1076 (rs6000_adjust_priority): Ditto.
1077 (rs6000_sched_reorder): Ditto.
1078 (rs6000_sched_reorder2): Ditto.
1079 (insn_must_be_first_in_group): Ditto.
1080 (insn_must_be_last_in_group): Ditto.
1081 (rs6000_register_move_cost): Ditto.
1082 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1085 2017-12-07 Julia Koval <julia.koval@intel.com>
1087 * config.gcc: Add vaesintrin.h.
1088 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1090 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1091 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1093 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1094 * config/i386/immintrin.h: Include vaesintrin.h.
1095 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1096 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1097 _mm_aesdec_epi128): New intrinsics.
1099 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1101 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1102 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1103 c-family/c-spellcheck.cc.
1104 (best_macro_match::best_macro_match): Likewise.
1106 (struct edit_distance_traits<cpp_hashnode *>): Move to
1107 c-family/c-spellcheck.h.
1108 (class best_macro_match): Likewise.
1110 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1112 PR tree-optimization/83293
1113 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1114 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1115 might insert into empty bb.
1118 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1119 simplify for plus with :c added, and pointer_plus without that.
1120 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1121 with undefined overflow and the conversion is not widening,
1122 perform negation in utype and only convert to type afterwards.
1123 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1124 simplify for plus with :c added, and pointer_plus without that.
1125 If type is integral with undefined overflow and the conversion is
1126 not widening, perform minus in utype and only convert to type
1127 afterwards. Move the last pointer_diff_expr simplify into the
1130 2017-12-06 Martin Sebor <msebor@redhat.com>
1132 PR tree-optimization/82646
1133 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1134 strncpy, not maxlen.
1136 2017-12-06 Martin Sebor <msebor@redhat.com>
1138 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1141 PR tree-optimization/83075
1142 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1143 strncat/strncpy don't change length of source string.
1145 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1148 2017-11-29 Martin Aberg <maberg@gaisler.com>
1150 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1151 to prevent b2bst errata sequence.
1152 (sqrtdf2_fix): Likewise.
1154 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1156 PR tree-optimization/81945
1157 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1158 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1159 to dest_cfun has orig_loop_num set, either remap it to the new
1160 loop number if the loop got moved too, or clear it.
1162 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1164 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1166 (thunderx2t99_nothing): New insn reservation.
1167 (thunderx2t99_mrs): New insn reservation.
1168 (thunderx2t99_multiple): New insn reservation.
1169 (thunderx2t99_alu_basi): Add bfx to reservation.
1170 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1172 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1175 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1177 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1179 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1180 tree-ssa-pre.c::remove_dead_inserted_code.
1181 * tree-ssa-dce.h: New file.
1182 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1183 (remove_dead_inserted_code): Move and rename to function
1184 tree-ssa-dce.c::simple_dce_from_worklist.
1185 (pass_pre::execute): Update use.
1187 2017-12-05 Richard Biener <rguenther@suse.de>
1189 PR tree-optimization/83277
1190 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1191 to code-gen liveout vars.
1193 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1195 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1196 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1197 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1200 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1202 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1204 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1205 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1208 2017-12-05 Richard Biener <rguenther@suse.de>
1210 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1211 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1212 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1213 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1215 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1217 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1219 2017-12-05 Martin Liska <mliska@suse.cz>
1220 Jakub Jelinek <jakub@redhat.com>
1222 * doc/invoke.texi: Document the options.
1223 * flag-types.h (enum sanitize_code): Add
1224 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1225 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1226 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1227 * opts.c: Define new sanitizer options.
1228 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1229 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1231 2017-12-05 Julia Koval <julia.koval@intel.com>
1233 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1234 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1235 (ix86_handle_option): Handle -mavx512vnni.
1236 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1237 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1238 * config/i386/i386-c (__AVX512VNNI__): New.
1239 * config/i386/i386.c (ix86_target_string): Handle new option.
1240 (ix86_valid_target_attribute_inner_p): Handle new option.
1241 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1242 * config/i386/i386.opt (mavx512vnni): New option.
1244 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1247 * athlon.md: Disable for generic.
1248 * haswell.md: Enable for generic.
1249 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1250 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1252 (ix86_adjust_cost): Move generic to haswell path.
1254 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1256 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1257 instead of MEM_P in a couple more places. Fix formatting issues.
1259 2017-12-04 Jim Wilson <jimw@sifive.com>
1261 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1262 instead of GP_REG_LAST-1.
1263 (riscv_adjust_libcall_cfi_prologue): Likewise.
1264 (riscv_adjust_libcall_cri_epilogue): Likewise.
1265 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1268 2017-12-04 Luis Machado <luis.machado@linaro.org>
1270 * ipa-pure-const.c (check_decl): Add missing newline.
1271 (state_from_flags): Likewise.
1273 2017-12-04 Jeff Law <law@redhat.com>
1275 PR tree-optimizatin/78496
1276 * gimple-ssa-evrp-analyze.h
1277 (evrp_range_analyzer::get_vr_values): Simplify.
1278 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1279 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1280 and gimple-ssa-evrp-analyze.h.
1281 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1282 (simplify_stmt_for_jump_threading): Copy a blob of code from
1283 tree-vrp.c to use ranges to simplify statements.
1284 (dom_opt_dom_walker::before_dom_children): Call
1285 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1286 (dom_opt_dom_walker::after_dom_children): Similarly for
1287 evrp_range_analyzer::leave.
1288 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1291 * gimple-ssa-evrp-analyze.c
1292 (evrp_range_analyzer::extract_range_from_stmt): Always use
1293 vr_values::update_value_range so preexisting range info is
1294 medged with new range info, even if the new range is VR_VARYING.
1296 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1298 * combine.c: Adjust comment.
1299 (use_crosses_set_p): Delete.
1300 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1301 (try_combine): Ditto.
1303 2017-12-04 Richard Biener <rguenther@suse.de>
1305 PR tree-optimization/83255
1306 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1307 Re-add zero-iteration check.
1309 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1311 PR rtl-optimization/83245
1312 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1313 hard registers as earlyclobber, also if not in an asm.
1315 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1319 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1322 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1323 rs6000_cpu based on cpu_index, not tune_index.
1325 2017-12-04 Richard Biener <rguenther@suse.de>
1327 PR tree-optimization/83238
1328 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1329 code match comment, rejecting invalid SESE regions.
1331 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1333 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1334 require base operand is a REG_POINTER prior to reload on targets
1335 with non-equivalent space registers.
1337 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1339 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1340 (update_specialized_profile): Fix updating of counts.
1341 (perhaps_add_new_callers): Likewise.
1343 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1346 * x86-tune.def: Remove obsolette FIXMEs.
1347 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1348 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1349 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1351 (X86_TUNE_PAD_RETURNS): Disable for generic.
1353 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1355 PR tree-optimization/83170
1356 PR tree-optimization/83241
1357 * gimple-ssa-store-merging.c
1358 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1359 gimple_vuse (ins_stmt) in case it has changed.
1360 (imm_store_chain_info::output_merged_store): Likewise.
1362 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1366 * tree-cfg.c (pass_warn_function_return::execute): Handle
1367 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1368 with BUILTINS_LOCATION.
1372 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1373 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1374 instead of DECL_MODE.
1376 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1377 last argument to rtx pointer.
1378 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1379 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1380 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1381 depending on the chosen ISAs.
1382 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1383 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1384 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1386 * config/i386/sse.md (mov<mode>_internal): Likewise.
1387 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1389 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1391 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1393 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1395 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1396 parameters from prototype.
1397 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1398 parameters. Don't print a comment.
1399 (emit_fusion_gpr_load): Adjust.
1400 (emit_fusion_load_store): Adjust.
1401 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1402 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1403 comment on the second line.
1405 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1408 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1409 rs6000_cpu based on cpu_index, not tune_index.
1411 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1413 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1414 which_alternative instead of which_alternative + 1.
1415 (output_asm_insn): Print an extra tab if the template is short.
1417 2017-12-01 Jim Wilson <jimw@sifive.com>
1419 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1421 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1422 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1424 * doc/tm.texi: Regenerate.
1426 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1429 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1430 whether we can allocate pseudos before trying to fix an address.
1431 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1432 memory address is indexed or indirect.
1433 (floatuns_<mode>si2_hw2): Likewise.
1435 2017-12-01 Jason Merrill <jason@redhat.com>
1437 * Makefile.in (TAGS): Add c-family/*.cc.
1439 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1441 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1442 (call_value_insn): Likewise.
1443 (sibcall_insn): Likewise.
1444 (sibcall_value_insn): Likewise.
1445 (movsi_aarch64): Likewise.
1446 (movdi_aarch64): Likewise.
1447 (add_losym_): Likewise.
1448 (ldr_got_small_): Likewise.
1449 (ldr_got_small_sidi): Likewise.
1450 (ldr_got_small_28k_): Likewise.
1451 (ldr_got_small_28k_sidi): Likewise.
1452 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1453 Move output_addr_const to symbolic case. Add error check.
1455 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1457 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1458 * config/s390/s390.md: Disable MVC merging peephole if it would
1459 disable operand forwarding.
1460 (new peephole2): Split MVCs if it would turn them into up to 2
1463 2017-12-01 Richard Biener <rguenther@suse.de>
1465 PR tree-optimization/83232
1466 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1467 detection of same access. Instead of breaking the group here
1468 do not consider the duplicate. Add comment explaining real fix.
1470 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1472 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1474 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1476 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1478 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1480 * function.h (struct function): Remove cilk_frame_decl,
1481 is_cilk_function and calls_cilk_spawn fields.
1482 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1484 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1485 cilk_elemental field.
1486 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1487 * target.def: Adjust comment.
1488 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1489 Don't test cilk_elemental.
1491 PR tree-optimization/83233
1492 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1493 bswap_stat name for the struct.
1496 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1497 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1498 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1499 SWITCH_BREAK_LABEL_P set on the label.
1500 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1501 added for default case if it was missing and not all cases covered.
1502 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1503 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1505 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1506 to avoid -Wimplicit-fallthrough warning.
1507 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1508 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1510 PR tree-optimization/83221
1511 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1513 (init_reassoc): Formatting fix.
1516 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1517 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1521 * tree-cfg.c: Include asan.h.
1522 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1524 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1526 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1528 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1530 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1531 vpcmpeqd instruction.
1533 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1535 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1536 registers type for 128bit mode.
1538 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1540 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1542 * gengtype-state.c (read_state_pair): Likewise.
1543 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1544 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1545 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1546 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1547 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1548 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1549 * builtins.c (fold_builtin_3): Likewise.
1550 * graphite-scop-detection.c
1551 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1552 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1555 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1558 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1559 parameter. Return true if ignore_decl_rtl_set_p is true and other
1560 conditions are satisfied.
1561 * asan.h (asan_protect_global): Add new parameter.
1562 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1563 to asan_protect_global calls.
1565 2017-11-30 Jim Wilson <jimw@sifive.com>
1567 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1568 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1569 -mstrict-align, add info on default value. Delete redundant lines for
1570 -mabi. Add missing -mexplicit-relocs docs.
1572 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1574 * config/arc/arc.md (trap): New pattern.
1576 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1578 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1579 instruction to end into a delay slot.
1580 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1581 can be placed into a delay slot against reg_note.
1583 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1585 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1586 labels number of usages.
1588 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1590 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1592 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1594 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1597 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1598 multiplication by power of 2 constant into two shifts + comparison.
1600 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1603 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1605 2017-11-30 Richard Biener <rguenther@suse.de>
1607 PR tree-optimization/83202
1608 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1609 allow_peel argument and guard peeling.
1610 (canonicalize_loop_induction_variables): Likewise.
1611 (canonicalize_induction_variables): Pass false.
1612 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1613 peeling from cunrolli.
1615 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1617 * combine.c (try_combine): Print a message to dump file whenever
1618 I0, I1, or I2 cannot be combined into I3.
1620 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1622 PR rtl-optimization/83156
1623 PR rtl-optimization/82621
1624 * combine.c (try_combine): Don't split an I2 if one of the dests is
1625 set again before I3. Allow unused dests.
1627 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1629 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1631 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1633 PR rtl-optimization/80818
1634 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1635 recursively. Use insn code for clobber.
1636 (lra_set_insn_recog_data): Pass the new arg to
1637 collect_non_operand_hard_regs.
1638 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1640 (lra_update_insn_regno_info): Pass insn to
1641 add_regs_to_insn_regno_info.
1643 2017-11-29 Jim Wilson <jimw@sifive.com>
1644 Andrew Waterman <andrew@sifive.com>
1646 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1647 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1648 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1649 use SINGLE_SHIFT_COST.
1650 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1651 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1653 2017-11-29 Julia Koval <julia.koval@intel.com>
1655 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1656 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1657 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1658 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1659 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1660 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1661 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1662 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1663 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1664 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1665 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1666 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1667 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1668 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1669 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1670 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1671 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1672 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1673 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1674 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1675 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1676 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1677 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1678 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1679 __builtin_ia32_vpshldv_v2di_mask,
1680 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1681 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1682 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1684 2017-11-29 Julia Koval <julia.koval@intel.com>
1686 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1687 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1688 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1689 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1690 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1691 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1692 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1693 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1694 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1695 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1696 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1697 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1698 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1699 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1700 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1701 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1702 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1703 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1704 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1705 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1706 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1708 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1710 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1711 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1714 2017-11-29 Martin Aberg <maberg@gaisler.com>
1716 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1717 to prevent b2bst errata sequence.
1718 (sqrtdf2_fix): Likewise.
1720 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1722 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1723 (div_sqrt_insn_p): New function.
1724 (sparc_do_work_around_errata): Insert NOP instructions to
1725 prevent sequences that could trigger the TN-0013 errata for
1726 certain LEON3 processors.
1727 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1728 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1729 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1730 (in_branch_delay): Prevent div and sqrt in delay slot if
1732 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1734 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1736 * config/sparc/sparc.c (atomic_insn_p): New function.
1737 (sparc_do_work_around_errata): Insert NOP instructions to
1738 prevent sequences that could trigger the TN-0010 errata for
1740 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1741 instruction referable in atomic_insns_p.
1743 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1745 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1746 (atomic_compare_and_swap_leon3_1): Likewise.
1749 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1751 * config/sparc/sparc.c (fpop_insn_p): New function.
1752 (sparc_do_work_around_errata): Insert NOP instructions to
1753 prevent sequences that could trigger the TN-0012 errata for
1755 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1756 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1757 (in_branch_annul_delay): Prevent floating-point instructions
1758 in delay slot of annulled integer branch.
1760 2017-11-29 Richard Biener <rguenther@suse.de>
1762 PR tree-optimization/83202
1763 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1765 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1766 nodes vectorized to the same stmts multiple times.
1767 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1768 (vect_analyze_slp_instance): Adjust.
1769 (scalar_stmts_to_slp_tree_map_t): New typedef.
1770 (vect_schedule_slp_instance): Add a map recording the SLP node
1771 representing the vectorized stmts for a set of scalar stmts.
1772 Avoid code-generating redundancies.
1773 (vect_schedule_slp): Allocate map and pass it down.
1775 2017-11-29 Nathan Sidwell <nathan@acm.org>
1778 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1781 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1783 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1784 cases for printing LDP/STP memory addresses.
1785 (aarch64_print_address_internal): Renamed from
1786 aarch64_print_operand_address, added parameter, add Pmode check.
1787 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1788 (aarch64_print_operand_address): Indirect to
1789 aarch64_print_address_internal.
1790 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1791 'y' operand output specifier.
1793 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1796 * tree.c (build_simple_mem_ref_loc): Handle
1797 get_addr_base_and_unit_offset returning a MEM_REF.
1800 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1803 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1806 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1808 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1810 (layout::move_to_column): Likewise.
1812 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1815 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1816 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1818 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1819 Martin Jambor <mjambor@suse.cz>
1822 * tree.h (expr_type_first_operand_type_p): Declare
1823 * tree.c (expr_type_first_operand_type_p): New function.
1824 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1825 (ipa_value_from_jfunc): Adjust declaration.
1826 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1827 Use it as result type for arithmetics, unless it is NULL in which case
1828 be more conservative.
1829 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1830 ipa_get_jf_pass_through_result.
1831 (propagate_vals_across_pass_through): Likewise.
1832 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1833 is to propagate_vals_across_pass_through.
1834 (propagate_constants_across_call): Pass PARM_TYPE to
1835 propagate_scalar_across_jump_function.
1836 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1837 ipa_value_from_jfunc.
1838 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1839 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1840 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1841 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1842 target_type, pass it to ipa_value_from_jfunc.
1843 (update_indirect_edges_after_inlining): Pass parameter type to
1844 try_make_edge_direct_simple_call.
1846 2017-11-28 Jeff Law <law@redhat.com>
1848 * gimple-ssa-evrp-analyze.c
1849 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1850 refine ranges if scev_initialized_p returns true.
1851 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1853 2017-11-28 Julia Koval <julia.koval@intel.com>
1855 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1856 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1857 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1858 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1859 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1860 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1861 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1862 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1863 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1864 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1865 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1866 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1867 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1868 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1869 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1870 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1871 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1872 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1873 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1874 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1875 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1876 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1878 2017-11-28 Julia Koval <julia.koval@intel.com>
1880 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1881 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1882 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1883 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1884 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1885 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1886 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1887 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1888 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1889 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1890 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1891 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1892 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1893 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1894 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1895 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1896 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1897 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1898 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1899 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1900 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1901 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1902 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1903 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1904 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1905 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1906 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1907 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1908 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1909 __builtin_ia32_vpshld_v2di_mask): New builtins.
1910 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1911 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1913 2017-11-28 Richard Biener <rguenther@suse.de>
1915 PR tree-optimization/80776
1916 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1918 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1920 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1921 If the incoming edge is an effective fallthru because the other
1922 edge only reaches a __builtin_unreachable () then record ranges
1923 derived from the controlling condition in SSA info.
1924 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1925 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1927 2017-11-28 Olivier Hainque <hainque@adacore.com>
1929 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1930 on mingw build hosts.
1932 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1935 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1937 2017-11-28 Richard Biener <rguenther@suse.de>
1940 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1941 copies generated from memcpy use a character array as reference
1944 2017-11-28 Julia Koval <julia.koval@intel.com>
1945 Sebastian Peryt <sebastian.peryt@intel.com>
1947 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1948 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1949 cilk-common.o, cilk.h, cilk-common.c): Remove.
1951 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1952 * builtins.c (is_builtin_name): Remove cilkplus condition.
1953 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1954 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1955 cilk-builtins.def, cilkplus.def): Remove.
1956 * cif-code.def (CILK_SPAWN): Remove.
1957 * cilk-builtins.def: Delete.
1958 * cilk-common.c: Ditto.
1960 * cilkplus.def: Ditto.
1961 * config/darwin.h (fcilkplus): Delete.
1962 * cppbuiltin.c: Ditto.
1963 * doc/extend.texi: Remove cilkplus doc.
1964 * doc/generic.texi: Ditto.
1965 * doc/invoke.texi: Ditto.
1966 * doc/passes.texi: Ditto.
1967 * gcc.c (fcilkplus): Remove.
1968 * gengtype.c (cilk.h): Remove.
1969 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1971 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1973 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1975 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1976 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1977 cilkplus conditions.
1978 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1979 inline_read_section): Ditto.
1980 * ipa-inline-analysis.c (cilk.h): Remove.
1981 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1982 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1983 append_linker_options): Remove condition for fcilkplus.
1984 * lto/lto-lang.c (cilk.h): Remove.
1985 (lto_init): Remove condition for fcilkplus.
1986 * omp-expand.c (expand_cilk_for_call): Delete.
1987 (expand_omp_taskreg, expand_omp_for_static_chunk,
1988 expand_omp_for): Remove cilkplus
1990 (expand_cilk_for): Delete.
1991 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1992 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1993 execute_lower_omp, diagnose_sb_0): Ditto.
1994 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1995 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1996 * tree-nested.c: Ditto.
1997 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1998 (dump_generic_node): Ditto.
1999 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2000 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2002 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2004 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2006 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2007 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2008 for vector mode and !TARGET_SIMD.
2010 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2012 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2014 * tree.h (SWITCH_LABELS): Remove.
2015 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2016 assert SWITCH_BODY is non-NULL.
2017 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2019 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2021 PR tree-optimization/80788
2022 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2023 has TREE_OVERFLOW set, call drop_tree_overflow.
2025 2017-11-28 Richard Biener <rguenther@suse.de>
2027 PR tree-optimization/83158
2028 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2030 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2033 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2034 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2036 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2038 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2039 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2041 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2044 * varasm.c (compare_constant): Take the mode of the constants into
2045 account when comparing floating point constants.
2047 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2049 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2050 from explicit instantiation of debug_helper.
2051 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2053 2017-11-27 Richard Biener <rguenther@suse.de>
2055 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2058 2017-11-27 Richard Biener <rguenther@suse.de>
2060 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2061 (type_hash_canon): Also clear int_cst_hash_table entry for
2063 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2065 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2067 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2068 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2069 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2071 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2073 * hash-map.h (gt_cleare_cache): Avoid UB.
2075 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2077 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2078 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2079 (decide_unroll_runtime_iterations): Likewise.
2080 (decide_unroll_stupid): Likewise.
2082 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2085 * config/i386/i386.md: Add a loop with incssp.
2087 2017-11-27 Martin Jambor <mjambor@suse.cz>
2089 PR tree-optimization/81248
2090 * tree-sra.c (splice_param_accesses): Remove size check.
2091 (decide_one_param_reduction): Fix size check.
2092 * gimple-pretty-print.c (dump_profile): Silence warning.
2093 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2095 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2098 * dbxout.c (lastlineno): New variable.
2099 (dbx_debug_hooks): Use dbxout_switch_text_section as
2100 switch_text_section debug hook.
2101 (dbxout_function_end): Switch to current_function_section
2102 rather than function_section. If crtl->has_bb_partition,
2103 output just one N_FUN, depending on in_cold_section_p.
2104 (dbxout_source_line): Remember last lineno in lastlineno.
2105 (dbxout_switch_text_section): New function.
2106 (dbxout_function_decl): Adjust dbxout_block caller.
2107 (dbx_block_with_cold_children): New function.
2108 (dbxout_block): Return true if any LBRAC/RBRAC have been
2109 emitted. Use dbx_block_with_cold_children at depth == 0
2110 in second partition. Add PARENT_BLOCKNUM argument, pass
2111 it optionally adjusted to children. Output LBRAC/RBRAC
2112 around recursive call only if the block is in the current
2113 partition, if not and anything was output, emit empty
2115 * final.c (final_scan_insn): Compute cold_function_name
2116 before calling switch_text_section debug hook. Call
2117 that hook even if dwarf2out_do_frame if not emitting
2121 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2122 TREE_READONLY decls.
2124 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2126 PR rtl-optimization/82488
2127 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2129 2017-11-26 Julia Koval <julia.koval@intel.com>
2131 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2133 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2136 2017-11-26 Julia Koval <julia.koval@intel.com>
2138 * config/i386/driver-i386.c (host_detect_local_cpu):
2139 Detect skylake-avx512.
2141 2017-11-26 Julia Koval <julia.koval@intel.com>
2143 * config.gcc: Add -march=cannonlake.
2144 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2145 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2146 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2147 (PTA_CANNONLAKE): New.
2148 (processor_target_table): Add cannonlake.
2149 (ix86_option_override_internal): Ditto.
2150 (fold_builtin_cpu): Ditto.
2151 (get_builtin_code_for_version): Handle cannonlake.
2152 (M_INTEL_COREI7_CANNONLAKE): New.
2153 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2154 * doc/invoke.texi: Add -march=cannonlake.
2156 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2158 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2159 (try_init_one_plugin): Alternative implementation for MinGW.
2160 * Makefile.in (plugin_implib): New.
2161 (gengtype-lex.c): Fix broken AIX workaround.
2162 * configure: Regenerate.
2163 * doc/plugins.texi: Document support for MinGW.
2165 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2167 PR rtl-optimization/81553
2168 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2169 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2170 is a shift where C1 has different mode than the whole shift, use C1's
2171 mode for MULT rather than the shift's mode.
2174 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2175 builtins not enabled in the currently selected ISA.
2177 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2179 PR tree-optimization/71026
2180 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2181 (insert_reciprocals): Change to insert reciprocals before a division
2182 by a square and to insert the square of a reciprocal.
2183 (execute_cse_reciprocals_1): Change to consider division by a square.
2184 (register_division_in): Add importance parameter.
2186 2017-11-24 Richard Biener <rguenther@suse.de>
2188 PR tree-optimization/82402
2189 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2190 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2192 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2194 * match.pd (0-ptr): New transformation.
2196 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2199 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2200 when skipping checking.
2202 2017-11-24 Richard Biener <rguenther@suse.de>
2204 PR tree-optimization/83128
2205 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2206 (vn_reference_lookup_3): Likewise.
2208 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2211 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2212 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2215 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2216 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2217 check for the latter with gimple_call_builtin_p. Do not handle
2218 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2220 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2222 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2225 2017-11-23 Julia Koval <julia.koval@intel.com>
2227 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2228 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2229 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2230 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2231 _mm512_maskz_expandloadu_epi16): New intrinsics.
2232 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2233 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2234 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2235 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2236 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2237 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2238 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2239 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2240 _mm256_maskz_expandloadu_epi8): New intrinsics.
2241 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2242 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2243 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2244 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2245 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2246 * config/i386/sse.md (VI248_VLBW): New iterator.
2247 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2249 2017-11-23 Julia Koval <julia.koval@intel.com>
2251 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2252 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2253 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2254 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2255 _mm512_mask_compressstoreu_epi16): New.
2256 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2257 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2258 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2259 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2260 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2261 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2262 _mm256_mask_compressstoreu_epi8): New.
2263 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2264 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2265 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2266 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2267 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2268 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2269 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2270 __builtin_ia32_compresshi128_mask,
2271 __builtin_ia32_compressstoreuqi512_mask,
2272 __builtin_ia32_compressstoreuhi512_mask,
2273 __builtin_ia32_compressstoreuqi256_mask,
2274 __builtin_ia32_compressstoreuqi128_mask,
2275 __builtin_ia32_compressstoreuhi256_mask,
2276 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2277 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2279 (ix86_expand_special_args_builtin): Handle new types.
2280 (s4fma_expand): Handle new builtin array.
2281 * config/i386/immintrin.h: Include new headers.
2282 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2283 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2285 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2287 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2288 fixes. Declare temp and g variables at the top in order to avoid
2289 {} in most of the cases.
2291 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2293 * match.pd (ptr-0): New transformation.
2295 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2297 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2298 (struct addr_mode_cost_table): New.
2299 (struct tune_params): Add field addr_mode_costs.
2300 * config/arm/arm.c (generic_addr_mode_costs): New.
2301 (arm_slowmul_tune): Initialise addr_mode_costs field.
2302 (arm_fastmul_tune): Likewise.
2303 (arm_strongarm_tune): Likewise.
2304 (arm_xscale_tune): Likewise.
2305 (arm_9e_tune): Likewise.
2306 (arm_marvell_pj4_tune): Likewise.
2307 (arm_v6t2_tune): Likewise.
2308 (arm_cortex_tune): Likewise.
2309 (arm_cortex_a8_tune): Likewise.
2310 (arm_cortex_a7_tune): Likewise.
2311 (arm_cortex_a15_tune): Likewise.
2312 (arm_cortex_a35_tune): Likewise.
2313 (arm_cortex_a53_tune): Likewise.
2314 (arm_cortex_a57_tune): Likewise.
2315 (arm_exynosm1_tune): Likewise.
2316 (arm_xgene1_tune): Likewise.
2317 (arm_cortex_a5_tune): Likewise.
2318 (arm_cortex_a9_tune): Likewise.
2319 (arm_cortex_a12_tune): Likewise.
2320 (arm_cortex_a73_tune): Likewise.
2321 (arm_v7m_tune): Likewise.
2322 (arm_cortex_m7_tune): Likewise.
2323 (arm_v6m_tune): Likewise.
2324 (arm_fa726te_tune): Likewise.
2325 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2327 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2329 * config/arm/arm.c (arm_mem_costs): New function.
2330 (arm_rtx_costs_internal): Use arm_mem_costs.
2332 2017-11-23 Mark Wielaard <mark@klomp.org>
2334 * dwarf2out.c (init_sections_and_labels): Use generation to create
2335 unique ranges_section_label and ranges_base_label. Return generation.
2336 (output_rnglists): Add generation argument. Use generation to create
2337 unique ranges labels.
2338 (dwarf2out_finish): Get generation from init_sections_and_labels
2339 and pass generation to output_rnglists.
2341 2017-11-23 Mike Stump <mikestump@comcast.net>
2342 Eric Botcazou <ebotcazou@adacore.com>
2344 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2345 * cfgloop.h (struct loop): Add unroll field.
2346 * function.h (struct function): Add has_unroll bitfield.
2347 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2348 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2349 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2350 (pass_rtl_unroll_loops::gate): Likewise.
2351 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2352 for which loop->unroll==1.
2353 (decide_unroll_constant_iterations): Use note for consistency and
2354 take loop->unroll into account. Return early if loop->unroll is set.
2355 Fix thinko in existing test.
2356 (decide_unroll_runtime_iterations): Use note for consistency and
2357 take loop->unroll into account.
2358 (decide_unroll_stupid): Likewise.
2359 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2360 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2361 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2363 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2364 (print_loop): Print loop->unroll if set.
2365 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2366 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2367 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2369 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2370 loop->unroll is set and smaller than the trip count. Otherwise bypass
2371 entirely the heuristics if loop->unroll is set. Remove dead note.
2372 Fix off-by-one bug in other note.
2373 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2374 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2376 (tree_unroll_loops_completely): Make static.
2377 (pass_complete_unroll::execute): Use correct type for variable.
2378 (pass_complete_unrolli::execute): Fix formatting.
2379 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2381 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2383 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2384 enable when TARGET_PREFER_AVX128 is set.
2386 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2388 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2390 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2391 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2393 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2395 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2396 frequencies. Fix estimation of aggregate parameters.
2398 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2400 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2403 2017-11-23 Tom de Vries <tom@codesourcery.com>
2405 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2407 2017-11-23 Tom de Vries <tom@codesourcery.com>
2409 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2412 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2415 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2416 sibcall_value_pcrel_fdpic): Use local variable instead of
2418 (calli_tbr_rel): Add missing operand 2.
2419 (call_valuei_tbr_rel): Add missing operand 3.
2421 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2424 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2425 bitpos/bitsize covering the whole destination, use store_expr only if
2426 the complex mode is the same. Otherwise, use expand_normal and if
2427 it returns CONCAT, subreg each part separately instead of trying to
2428 subreg the whole result.
2430 2017-11-23 Richard Biener <rguenther@suse.de>
2432 PR tree-optimization/23094
2433 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2434 come here from walking over backedges in the first iteration.
2435 (vn_reference_lookup_3): Skip clobbers that store the same value.
2437 2017-11-23 Richard Biener <rguenther@suse.de>
2439 PR tree-optimization/81403
2440 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2441 a block we need a leader relative to.
2442 (phi_translate_1): For nary processing require a leader from
2443 get_representative_for given we run expression simplification
2444 using match-and-simplify. Remove previous fix.
2446 2017-11-22 Jeff Law <law@redhat.com>
2448 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2449 Use new method allocate_value_range rather than accessing the
2450 vrp_value_range_pool data member directly.
2451 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2452 to use extract_range_from_stmt method to avoid need for
2453 extract_range_from_assignment method.
2454 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2455 method rather than setting values_propgated data member directly.
2456 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2457 and values propagated data members and extract_range_from_assignment
2458 method. Reorder private data members to conform to standards.
2459 Add new methods set_lattice_propagation_complete and
2460 allocate_value_range.
2462 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2464 PR rtl-optimization/83030
2465 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2466 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2467 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2468 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2469 CROSSING_JUMP_P flag is already set before setting it.
2470 * cfgrtl.c (fixup_partition_crossing): Likewise.
2471 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2474 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2476 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2477 Handle the case where both arguments are using gen_const_vec_series.
2479 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2482 * pretty-print.c (pp_format): Move quoting implementation to
2483 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2484 to pass address of "quote" local.
2485 (pp_begin_quote): New function.
2486 (pp_end_quote): New function.
2487 * pretty-print.h (printer_fn): Convert penultimate param from bool
2489 (pp_begin_quote): New decl.
2490 (pp_end_quote): New decl.
2491 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2492 param from bool to bool *.
2493 * tree-diagnostic.h (default_tree_printer): Likewise.
2495 2017-11-22 Jeff Law <law@redhat.com>
2497 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2498 Initialize vr_values.
2499 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2500 vr_values via vr_values class instance rather than delegators.
2501 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2502 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2503 (evrp_range_analyzer::push_value_range): Likewise.
2504 (evrp_range_analyzer::pop_value_range): Likewise.
2505 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2506 most delegators. Those remaining are exposed as public interfaces.
2507 Make vr_values a pointer and private.
2508 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2510 (evrp_range_analyzer::get_vr_value): New method.
2511 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2512 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2513 (class evrp_dom_walker): Attach evrp_folder class, initialize
2514 it in the ctor. Remove temporary delegators.
2515 (evrp_dom_walker::before_dom_children): Call methods in attached
2516 evrp_range_analyzer class via class instance pointer. Use
2517 free value_range_constant_singleton to remove need for
2518 op_with_constant_singleton_value delegator method. Do not
2519 create a vrp_prop class instance for every call! Narrow
2520 scope of a couple variables.
2521 (evrp_dom_walker::cleanup): Call methods in attached
2522 evrp_range_analyzer class via class instance pointer.
2523 * vr-values.h (class vr_values): Privatize many methods and
2526 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2528 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2529 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2530 * expr.c (expand_expr_real_2): Likewise.
2531 * fold-const.c (const_unop): Likewise.
2532 * optabs-tree.c (optab_for_tree_code): Likewise.
2533 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2534 * tree-inline.c (estimate_operator_cost): Likewise.
2535 * tree-pretty-print.c (dump_generic_node): Likewise.
2536 (op_code_prio): Likewise.
2537 (op_symbol_code): Likewise.
2538 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2539 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2540 * internal-fn.c (direct_internal_fn_optab): New function.
2541 (direct_internal_fn_array, direct_internal_fn_supported_p
2542 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2543 * fold-const-call.c (fold_const_reduction): New function.
2544 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2546 * tree-vect-loop.c: Include internal-fn.h.
2547 (reduction_code_for_scalar_code): Rename to...
2548 (reduction_fn_for_scalar_code): ...this and return an internal
2550 (vect_model_reduction_cost): Take an internal_fn rather than
2552 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2554 (vectorizable_reduction): Use internal functions rather than tree
2555 codes for the reduction operation. Update calls to the functions
2557 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2558 Use calls to internal functions rather than REDUC tree codes.
2559 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2561 2017-11-22 Olivier Hainque <hainque@adacore.com>
2563 * config/vxworks.c (vxworks_override_options): Pick default
2564 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2565 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2566 DWARF_GNAT_ENCODINGS_DEFAULT.
2567 * config/vxworksae.h: Likewise.
2569 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2571 PR tree-optimization/83104
2572 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2575 2017-11-22 Marek Polacek <polacek@redhat.com>
2576 H.J. Lu <hongjiu.lu@intel.com>
2577 Jason Merrill <jason@redhat.com>
2582 * calls.c (initialize_argument_information): Call
2583 warn_parameter_passing_abi target hook.
2584 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2585 argument onto stack.
2586 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2587 * common.opt: Update -fabi-version description.
2588 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2589 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2591 (ix86_is_empty_record): New function.
2592 (ix86_warn_parameter_passing_abi): New function.
2593 (TARGET_EMPTY_RECORD_P): Redefine.
2594 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2595 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2596 * doc/tm.texi: Regenerated.
2597 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2598 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2599 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2600 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2601 instead of int_size_in_bytes.
2602 * expr.c (copy_blkmode_to_reg): Likewise.
2603 * function.c (aggregate_value_p): Return 0 for empty types.
2604 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2606 (locate_and_pad_parm): Call arg size_in_bytes instead
2608 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2609 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2610 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2612 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2613 (std_gimplify_va_arg_expr): Skip empty records. Call
2614 arg_size_in_bytes instead size_in_bytes.
2615 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2616 * tree-core.h (tree_type_common): Add empty_flag.
2617 (tree_decl_common): Update comments.
2618 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2620 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2621 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2623 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2624 * tree.c (default_is_empty_type): New function.
2625 (default_is_empty_record): New function.
2626 (arg_int_size_in_bytes): New function.
2627 (arg_size_in_bytes): New function.
2628 (get_ultimate_context): New function.
2629 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2630 TRANSLATION_UNIT_WARN_EMPTY_P.
2631 (default_is_empty_record, arg_int_size_in_bytes,
2632 arg_size_in_bytes, get_ultimate_context): Declare.
2634 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2636 * config/arm/arm.c (cmse_clear_registers): New function.
2637 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2638 code by call to cmse_clear_registers.
2639 (cmse_nonsecure_entry_clear_before_return): Likewise.
2641 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2643 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2644 (vdot_s32, vdotq_s32): New.
2645 (vdot_lane_u32, vdotq_lane_u32): New.
2646 (vdot_lane_s32, vdotq_lane_s32): New.
2649 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2652 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2653 for unsigned values with fewer HWIs than the precision.
2654 (test_overflow): New function.
2655 (wide_int_cc_tests): Call it.
2657 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2659 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2660 has the same mode class as Pmode.
2662 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2664 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2665 padding_bits_to_clear_ptr.
2666 (cmse_nonsecure_entry_clear_before_return): Likewise.
2668 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2670 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2671 auto_sbitap instead of integer bitfield to control register needing
2674 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2676 PR tree-optimization/83044
2677 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2678 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2679 that need the upper bound. Subtract offset from
2680 get_addr_base_and_unit_offset only if positive and subtract it
2681 before division by eltsize rather than after it.
2684 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2685 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2686 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2690 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2691 expand_binop with *mul_widen_optab, make sure at least one of the
2692 operands doesn't have VOIDmode.
2695 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2697 PR rtl-optimization/82044
2698 PR tree-optimization/82042
2699 * dse.c (record_store): Check for overflow.
2700 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2701 add_wild_read instead of clear_rhs_from_active_local_stores on
2702 overflow and log it into dump_file.
2704 2017-11-22 Richard Biener <rguenther@suse.de>
2706 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2707 fake edges to exit when looking for a place to insert.
2708 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2710 (insert_into_preds_of_block): Commit edge insertion immediately,
2711 assert that doesn't require new BBs.
2712 (fini_pre): Release expressions.
2713 (pass_pre::execute): Shuffle things around a bit, if the fn
2714 is too large do not compute AVAIL either as this is really the
2717 2017-11-22 Richard Biener <rguenther@suse.de>
2719 PR tree-optimization/83089
2720 * tree-if-conv.c (pass_if_conversion::execute): If anything
2721 changed reset SCEV and free the number of iteration estimates.
2723 2017-11-21 Martin Sebor <msebor@redhat.com>
2725 PR tree-optimization/82945
2726 * calls.h (warn_nonstring_bound): Remove unused function.
2728 2017-11-21 Martin Sebor <msebor@redhat.com>
2730 PR tree-optimization/82945
2731 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2732 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2733 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2735 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2736 * calls.h (get_attr_nonstring_decl): Declare new function.
2737 * doc/extend.texi (attribute nonstring): Update.
2738 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2739 get_attr_nonstring_decl and handle it.
2740 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2741 detection of nul-termination.
2742 (strlen_to_stridx): Change to a pointer.
2743 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2744 (pass_strlen::execute): Same.
2746 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2748 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2749 for the new option -mprefer-vector-width=[none|128|256|512].
2750 * config/i386/i386.c (ix86_target_string): remove old style options
2751 -mprefer-avx256 and make -mprefer-avx128 as alias.
2752 (ix86_option_override_internal): Apply defaults for the
2753 -mprefer-vector-width=[128|256] option.
2754 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2755 Implement macros to work with -mprefer-vector-width=.
2756 * config/i386/i386.opt: Implemented option
2757 -mprefer-vector-width=[none|128|256|512].
2758 * doc/invoke.texi: Documentation for
2759 -mprefer-vector-width=[none|128|256|512].
2761 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2763 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2765 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2767 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2769 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2771 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2773 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2774 * expr.c (expand_expr_real_2): Likewise.
2775 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2776 fold_binary_loc): Likewise.
2777 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2778 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2779 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2780 MINUS_EXPR transformations.
2781 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2782 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2783 * tree-inline.c (estimate_operator_cost): Likewise.
2784 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2785 op_symbol_code): Likewise.
2786 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2787 * vr-values.c (extract_range_from_binary_expr): Likewise.
2788 * varasm.c (initializer_constant_valid_p_1): Likewise.
2789 * tree.def: New tree code POINTER_DIFF_EXPR.
2791 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2793 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2794 integer suffix to movbe mnemonic.
2795 (*bswaphi2_movbe): Ditto.
2796 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2798 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2801 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2802 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2803 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2805 2017-11-21 Martin Liska <mliska@suse.cz>
2807 * tree-inline.c (expand_call_inline): Remove not needed
2810 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2811 Jakub Jelinek <jakub@redhat.com>
2814 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2815 Remove static keyword from f variable.
2817 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2819 PR tree-optimization/83086
2820 * gimple-ssa-store-merging.c
2821 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2822 rather than n.base_addr.
2824 2017-11-21 Martin Liska <mliska@suse.cz>
2826 PR rtl-optimization/82044
2827 PR tree-optimization/82042
2828 * dse.c (check_mem_read_rtx): Check for overflow.
2830 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2832 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2835 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2837 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2838 and its corresponding call arg location note.
2840 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2841 Andrew Burgess <andrew.burgess@embecosm.com>
2843 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2845 (arc_return_slot_offset): Likewise.
2846 (arc_eh_return_address_location): New declaration.
2847 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2848 (MUST_SAVE_REGISTER): Add exception handler case.
2849 (MUST_SAVE_RETURN_ADDR): Likewise.
2850 (arc_frame_pointer_required): Likewise.
2851 (arc_frame_pointer_needed): New function.
2852 (arc_compute_frame_size): Changed.
2853 (arc_expand_prologue): Likewise.
2854 (arc_expand_epilogue): Likewise.
2855 (arc_initial_elimination_offset): Likewise.
2856 (arc_return_slot_offset): Delete.
2857 (arc_eh_return_address_location): New function.
2858 (arc_builtin_setjmp_frame_value): Likewise.
2859 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2860 (EH_RETURN_STACKADJ_RTX): Define.
2861 (EH_RETURN_HANDLER_RTX): Likewise.
2862 * config/arc/arc.md (eh_return): Delete.
2864 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2866 * print-tree.h (debug_vec_tree): Remove prototype.
2867 * gdbinit.in (pvt): Remove macro.
2869 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2871 PR tree-optimization/83047
2872 * gimple-ssa-store-merging.c
2873 (imm_store_chain_info::output_merged_store): If the loads with the
2874 same vuse are in different basic blocks, for load_gsi pick a load
2875 location that is dominated by the other loads.
2878 * config/i386/i386.c (ix86_memmodel_check): Start
2879 -Winvalid-memory-model diagnostics with lowercase letter.
2882 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2883 set in_cold_section_p to the partition containing loc_list->first.
2884 When seeing loc_list->last_before_switch node, update secname and
2885 perform range_across_switch second partition handling only after that.
2888 * run-rtl-passes.c: Include debug.h.
2889 (run_rtl_passes): Call debug_hooks->assembly_start.
2890 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2894 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2895 OPTAB_DIRECT in calls to expand_simple_binop.
2897 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2900 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2902 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2905 * spellcheck.h (best_match::blithely_get_best_candidate): New
2908 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2910 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2911 with lower case letter.
2913 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2915 * config/i386/i386.md (bswaphi2): New expander.
2916 (*bswaphi2_movbe): New insn pattern.
2917 (bswaphi -> rorhi pepehole2): New peephole pattern.
2919 2017-11-20 Jeff Law <law@redhat.com>
2921 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2922 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2923 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2924 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2925 gimple-ssa-evrp-analyze.h.
2927 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2928 set BB_VISITED here.
2929 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2931 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2933 * tree-predcom.c: Add general comment on Store-Store chains.
2934 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2936 (get_chain_last_ref_at): Rename into...
2937 (get_chain_last_write_at): ...this.
2938 (get_chain_last_write_before_load): New function.
2939 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2940 CT_STORE_STORE when write reference is added.
2941 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2942 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2943 (initialize_root_vars_store_elim_1): Ditto.
2944 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2945 definition is created.
2946 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2947 chain by replacing it with dominant stored value.
2949 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2951 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2953 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2955 * vr-values.c (extract_range_from_binary_expr): Use a full range
2958 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2960 * config/arm/arm.md (R4_REGNUM): Define constant.
2961 (nonsecure_call_internal): Remove r4 clobber.
2962 (nonsecure_call_value_internal): Likewise.
2963 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2964 clobber and resequence match_operands.
2965 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2966 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2967 (nonsecure_call_value_reg_thumb2): Likewise.
2969 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2971 PR tree-optimization/78821
2972 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2973 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2974 to the address of the base rather than the base itself.
2975 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2976 (find_bswap_or_nop_finalize): New function.
2977 (find_bswap_or_nop): Use it.
2978 (bswap_replace): Return a tree rather than bool, change first
2979 argument from gimple * to gimple_stmt_iterator, allow inserting
2980 into an empty sequence, allow ins_stmt to be NULL - then emit
2981 all stmts into gsi. Fix up MEM_REF address gimplification.
2982 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2983 (struct store_immediate_info): Add N and INS_STMT non-static
2985 (store_immediate_info::store_immediate_info): Initialize them
2986 from newly added ctor args.
2987 (merged_store_group::apply_stores): Formatting fixes. Sort by
2989 (stmts_may_clobber_ref_p): For stores call also
2990 refs_anti_dependent_p.
2991 (gather_bswap_load_refs): New function.
2992 (imm_store_chain_info::try_coalesce_bswap): New method.
2993 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2994 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2995 (imm_store_chain_info::output_merged_store): Fail if number of
2996 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2997 and NOP_EXPR rhs_code.
2998 (pass_store_merging::process_store): Compute n and ins_stmt, if
2999 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3000 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3002 (pass_store_merging::execute): Calculate dominators.
3004 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3005 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3006 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3007 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3008 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3009 class pass_optimize_bswap, bswap_replace,
3010 pass_optimize_bswap::execute): Moved to ...
3011 * gimple-ssa-store-merging.c: ... this file.
3012 Include optabs-tree.h.
3013 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3014 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3015 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3016 anonymous namespace, remove static keywords.
3017 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3018 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3020 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3023 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3025 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3027 * vec.h (debug_helper): New function.
3028 (DEFINE_DEBUG_VEC): New macro.
3029 * hash-set.h (debug_helper): New function.
3030 (DEFINE_DEBUG_HASH_SET): New macro.
3031 * cfg.c (debug_slim (edge)): New function.
3032 Call DEFINE_DEBUG_VEC for edges.
3033 Call DEFINE_DEBUG_HASH_SET for edges.
3034 * cfghooks.c (debug_slim (basic_block)): New function.
3035 Call DEFINE_DEBUG_VEC for basic blocks.
3036 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3037 * print-tree.c (debug_slim): New function to handle trees.
3038 Call DEFINE_DEBUG_VEC for trees.
3039 Call DEFINE_DEBUG_HASH_SET for trees.
3040 (debug (vec<tree, va_gc>) &): Remove.
3041 (debug (<vec<tree, va_gc>) *): Remove.
3042 * print-rtl.c (debug_slim): New function to handle const_rtx.
3043 Call DEFINE_DEBUG_VEC for rtx_def.
3044 Call DEFINE_DEBUG_VEC for rtx_insn.
3045 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3046 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3047 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3048 (debug (vec<rtx_insn *> *ptr): Remove.
3049 (debug_insn_vector): Remove.
3050 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3052 2017-11-20 Tom de Vries <tom@codesourcery.com>
3054 PR rtl-optimization/82020
3055 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3056 IF_THEN_ELSE condition.
3058 2017-11-19 Jeff Law <law@redhat.com>
3060 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3061 of degenerates resulting from ignoring an edge.
3063 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3066 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3068 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3071 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3072 for uninitialied counts.
3074 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3077 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3079 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3082 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3084 2017-11-19 Tom de Vries <tom@codesourcery.com>
3086 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3089 2017-11-19 Tom de Vries <tom@codesourcery.com>
3091 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3094 2017-11-19 Tom de Vries <tom@codesourcery.com>
3096 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3099 2017-11-19 Tom de Vries <tom@codesourcery.com>
3101 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3103 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3105 2017-11-19 Tom de Vries <tom@codesourcery.com>
3107 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3109 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3110 semicolon after MCORE_EXPORT_NAME call.
3112 2017-11-19 Tom de Vries <tom@codesourcery.com>
3115 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3116 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3118 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3120 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3121 (cgraph_edge::clone): Cleanup updating of profile.
3122 * ipa-cp.c (update_profiling_info): Likewise.
3123 * ipa-inline-transform.c (inline_transform): Likewise.
3124 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3125 * ipa-split.c (execute_split_functions): Do not split when function
3127 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3128 * profile-count.c (profile_count::dump): Add global0.
3129 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3131 (profile_count::to_sreal_scale): Likewise.
3132 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3133 (profile_count::combine_with_ipa_count): New function.
3134 * profile-count.h (profile_guessed_global0adjusted): New.
3135 (profile_count::adjusted_zero): New.
3136 (profile_count::global0adjusted): New.
3137 (profile_count::combine_with_ipa_count): New.
3138 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3139 correct profile of return block of split functions.
3140 (copy_cfg_body): Remove unused profile_count.
3141 (copy_body): Likewise.
3142 (expand_call_inline): Update.
3143 (tree_function_versioning): Update.
3145 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3147 * hash-set.h (hash_set::empty): New.
3148 * tree-ssa-threadbackward.h: Delete.
3149 * tree-ssa-threadbackward.c (class thread_jumps): New.
3150 Move max_threaded_paths into class.
3151 (fsm_find_thread_path): Remove arguments that are now in class.
3152 (profitable_jump_thread_path): Rename to...
3153 (thread_jumps::profitable_jump_thread_path): ...this.
3154 (convert_and_register_jump_thread_path): Rename to...
3155 (thread_jumps::convert_and_register_current_path): ...this.
3156 (check_subpath_and_update_thread_path): Rename to...
3157 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3158 (register_jump_thread_path_if_profitable): Rename to...
3159 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3160 (handle_phi): Rename to...
3161 (thread_jumps::handle_phi): ...this.
3162 (handle_assignment): Rename to...
3163 (thread_jumps::handle_assignment): ...this.
3164 (fsm_find_control_statement_thread_paths): Rename to...
3165 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3166 (find_jump_threads_backwards): Rename to...
3167 (thread_jumps::find_jump_threads_backwards): ...this.
3168 Initialize path local data.
3169 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3170 from within thread_jumps class.
3171 (pass_early_thread_jumps::execute): Same.
3173 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3175 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3177 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3179 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3180 to canonical location.
3182 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3185 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3187 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3189 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3191 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3192 rcrt1.o%s/grcrt1.o%s for -static-pie.
3194 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3196 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3197 ix86_shift_rotate_cost): Break out from ...
3198 (ix86_rtx_costs): ... here.
3199 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3202 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3204 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3205 when entry block was promoted unlikely.
3206 (estimate_bb_frequencies): Increase frequency scale.
3207 * profile-count.h (profile_count): Export precision info.
3209 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3211 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3212 disturbin profile of entry block.
3214 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3216 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3219 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3221 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3222 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3224 2017-11-17 Jeff Law <law@redhat.com>
3226 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3227 from evrp_dom_walker class. Various methods moved into new class.
3228 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3229 (evrp_range_analyzer::enter): New method.
3230 (evrp_range_analyzer::leave): New method.
3231 (evrp_dom_walker): Remove delegators no longer needed by this class.
3232 Replace vr_values data member with evrp_range_analyzer
3234 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3235 method extracted from evrp_dom_walker::before_dom_children.
3236 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3237 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3239 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3240 Add private copy constructor and move assignment operators.
3241 Privatize methods and class data where trivially possible.
3242 (evrp_dom_walker::cleanup): New function, extracted from
3243 execute_early_vrp. Simplify access to class data.
3245 * vr-values.h (get_output_for_vrp): Prototype.
3246 * vr-values.c (get_output_for_vrp): New function extracted from
3247 vrp_visit_assignment_or_call and extract_range_from_stmt.
3248 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3250 2017-11-17 Luis Machado <luis.machado@linaro.org>
3252 * config/aarch64/aarch64.c
3253 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3254 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3255 tune_params::AUTOPREFETCHER_WEAK.
3257 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3260 * config/arm/arm.c (arm_valid_target_attribute_rec):
3261 Parse "arch=" and "+<ext>".
3262 (arm_valid_target_attribute_tree): Re-init global options.
3263 (arm_option_override): Make non-static.
3264 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3265 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3266 (__ARM_FEATURE_CRC32): Support undef.
3267 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3268 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3270 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3272 * gdbinit.in (break-on-diagnostic): New command.
3274 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3276 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3277 return to indirect jump for EH return if control-flow
3278 protection is enabled. Change explicit 'false' argument in
3279 pro_epilogue_adjust_stack with a value of
3281 * config/i386/i386.md (simple_return_indirect_internal):
3282 Remove SImode restriction to support 64-bit.
3284 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3286 * combine.c (added_notes_insn): New.
3287 (try_combine): Handle added_notes_insn like added_links_insn.
3288 Rewrite return value code.
3289 (distribute_notes): Set added_notes_insn to the earliest insn we added
3292 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3294 PR rtl-optimization/82621
3295 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3296 dest of one of those SETs is unused.
3298 2017-11-17 Richard Biener <rguenther@suse.de>
3301 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3302 * tree-pretty-print.c (dump_generic_node): Handle
3303 annot_expr_parallel_kind.
3304 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3305 * gimplify.c (gimple_boolify): Likewise.
3307 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3309 * config.gcc (extra_headers): Add cet.h for x86 targets.
3310 * config/i386/cet.h: New file.
3311 * doc/install.texi: Add --enable-cet/--disable-cet.
3313 2017-11-17 Richard Biener <rguenther@suse.de>
3315 PR tree-optimization/83017
3316 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3317 (gen_parallel_loop): Properly count iterations.
3318 (parallelize_loops): Handle loop->can_be_parallel independent
3319 of flag_loop_parallelize_all. Make static profitability test match
3321 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3322 * invoke.texi (parloops-min-per-thread): Document.
3324 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3326 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3327 upstreaming review comments.
3329 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3331 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3332 with fast unaligned access.
3333 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3335 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3337 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3338 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3340 2017-11-17 Richard Biener <rguenther@suse.de>
3342 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3343 folding of references.
3345 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3348 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3349 of replacing call to strncmp with corresponding call to strcmp when
3352 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3354 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3355 option prefer-avx256 for skylake-avx512 configuration.
3356 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3357 (get_builtin_code_for_version): Ditto.
3359 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3360 Monk Chiang <sh.chiang04@gmail.com>
3362 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3363 (FIXED_REGISTERS): Reserve more register numbers.
3364 (CALL_USED_REGISTERS): Likewise.
3365 (REG_ALLOC_ORDER): Likewise.
3366 (REG_CLASS_CONTENTS): Likewise.
3367 (REGISTER_NAMES): Likewise.
3369 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3370 Kito Cheng <kito.cheng@gmail.com>
3372 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3374 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3376 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3378 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3379 parameter list for vec_splats.
3381 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3383 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3385 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3387 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3390 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3391 with the required version of automake 1.11.6.
3393 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3395 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3397 (DU_C2_3_power9): Correct reservation combinations.
3398 (FP_DIV_power9, VEC_DIV_power9): New.
3399 (power9-alu): Split out rotate/shift...
3400 (power9-rot): ...to here, correct dispatch resource.
3401 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3403 (power9-fp): Correct latency.
3404 (power9-sdiv): Add div/sqrt resource.
3405 (power9-ddiv): Correct latency, add div/sqrt resource.
3406 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3407 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3409 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3411 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3413 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3414 switch statement mapping KF built-ins to TF built-ins if we don't
3415 have the proper ISA 3.0 assembler support.
3417 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3419 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3420 (lower_emutls_function_body): Do not compute it.
3422 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3424 * ipa-split.c (split_bb_info): Turn time to sreal.
3425 (split_point): Likewise.
3426 (dump_split_point): Likewise.
3427 (fine_split_points): Likewise.
3428 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3430 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3432 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3435 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3437 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3439 (expensive_function_p): Remove useless assert.
3440 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3442 2017-11-16 Martin Sebor <msebor@redhat.com>
3444 PR tree-optimization/82588
3445 PR tree-optimization/82583
3446 * tree-vrp.c (check_array_ref): Handle flexible array members,
3447 string literals, and inner indices.
3448 (search_for_addr_array): Add detail to diagnostics.
3450 2017-11-16 Nathan Sidwell <nathan@acm.org>
3454 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3455 (SET_DECL_ASSEMBLER_NAME): Forward to
3456 overwrite_decl_assembler_name.
3457 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3458 (overwrite_decl_assembler_name): Declare.
3459 * tree.c (overwrite_decl_assembler_name): New.
3460 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3461 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3462 (LANG_HOOKS_INITIALIZER): Add it.
3463 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3464 * langhooks.c (lhd_set_decl_assembler_name): Use
3465 SET_DECL_ASSEMBLER_NAME.
3466 (lhd_overwrite_decl_assembler_name): Default implementation.
3468 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3469 Jackson Woodruff <jackson.woodruff@arm.com>
3471 PR tree-optimization/71026
3472 * match.pd: Canonicalize constant multiplies in division.
3474 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3476 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3477 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3478 Remove OPT_fomit_frame_pointer entry.
3479 * common/config/arc/arc-common.c: Likewise.
3480 * common/config/arm/arm-common.c: Likewise.
3481 * common/config/avr/avr-common.c: Likewise.
3482 * common/config/c6x/c6x-common.c: Likewise.
3483 * common/config/cr16/cr16-common.c: Likewise.
3484 * common/config/cris/cris-common.c: Likewise.
3485 * common/config/epiphany/epiphany-common.c: Likewise.
3486 * common/config/fr30/fr30-common.c: Likewise.
3487 * common/config/frv/frv-common.c: Likewise.
3488 * common/config/ia64/ia64-common.c: Likewise.
3489 * common/config/iq2000/iq2000-common.c: Likewise.
3490 * common/config/lm32/lm32-common.c: Likewise.
3491 * common/config/m32r/m32r-common.c: Likewise.
3492 * common/config/mcore/mcore-common.c: Likewise.
3493 * common/config/microblaze/microblaze-common.c: Likewise.
3494 * common/config/mips/mips-common.c: Likewise.
3495 * common/config/mmix/mmix-common.c: Likewise.
3496 * common/config/mn10300/mn10300-common.c: Likewise.
3497 * common/config/nios2/nios2-common.c: Likewise.
3498 * common/config/pa/pa-common.c: Likewise.
3499 * common/config/pdp11/pdp11-common.c: Likewise.
3500 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3501 * common/config/riscv/riscv-common.c: Likewise.
3502 * common/config/rs6000/rs6000-common.c: Likewise.
3503 * common/config/rx/rx-common.c: Likewise.
3504 * common/config/s390/s390-common.c: Likewise.
3505 * common/config/sh/sh-common.c: Likewise.
3506 * common/config/sparc/sparc-common.c: Likewise.
3507 * common/config/tilegx/tilegx-common.c: Likewise.
3508 * common/config/tilepro/tilepro-common.c: Likewise.
3509 * common/config/v850/v850-common.c: Likewise.
3510 * common/config/visium/visium-common.c: Likewise.
3511 * common/config/xstormy16/xstormy16-common.c: Likewise.
3512 * common/config/xtensa/xtensa-common.c: Likewise.
3513 * invoke.texi (-fomit-frame-pointer): Update documentation.
3515 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3517 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3519 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3521 * cfg.c (scale_bbs_frequencies_int,
3522 cale_bbs_frequencies_gcov_type): Remove.
3523 * cfg.h (scale_bbs_frequencies_int,
3524 cale_bbs_frequencies_gcov_type): Remove.
3526 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3528 * tree-ssa-loop-manip.c
3529 (scale_dominated_blocks_in_loop): Update to profile counts.
3530 (tree_transform_and_unroll_loop): Likewise.
3532 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3534 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3535 scale_bbs_frequencies_int.
3537 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3539 * final.c (compute_alignments): Use counts rather than frequencies.
3541 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3543 * cfgloopanal.c: Include sreal.h
3544 (average_num_loop_insns): Use counts and sreal for accounting.
3546 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3548 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3551 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3552 Kito Cheng <kito.cheng@gmail.com>
3554 * config/nds32/constraints.md: Provide more constraints.
3555 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3556 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3557 support constraints usage.
3559 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3561 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3563 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3564 Kito Cheng <kito.cheng@gmail.com>
3566 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3567 * config/nds32/nds32.opt: Refine the layout.
3568 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3569 TARGET_EXT_STRING): Support new options.
3570 * config/nds32/nds32.h: Likewise.
3571 * config/nds32/nds32.md: Likewise.
3572 * config/nds32/nds32-predicates.c: Likewise.
3573 * config/nds32/constraints.md: Likewise.
3574 * common/config/nds32/nds32-common.c: Likewise.
3576 2017-11-16 Julia Koval <julia.koval@intel.com>
3579 * config/i386/gfniintrin.h: Add sse check.
3580 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3582 2017-11-16 Julia Koval <julia.koval@intel.com>
3584 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3585 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3586 (ix86_handle_option): Handle -mavx512vbmi2.
3587 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3588 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3589 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3590 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3591 (ix86_valid_target_attribute_inner_p): Ditto.
3592 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3593 * config/i386/i386.opt (mavx512vbmi2): New option.
3594 * doc/invoke.texi: Add new option.
3596 2017-11-16 Julia Koval <julia.koval@intel.com>
3598 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3599 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3600 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3601 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3602 _mm512_gf2p8mul_epi8): New intrinsics.
3603 * config/i386/i386-builtin-types.def
3604 (V64QI_FTYPE_V64QI_V64QI): New type.
3605 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3606 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3607 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3608 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3609 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3610 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3612 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3614 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3615 explicitly as a stream of bytes.
3617 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3619 * config/rs6000/altivec.h (vec_xst_be): New #define.
3620 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3621 and externalize from *altivec_vperm_<mode>_internal.
3622 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3624 (XL_BE_V8HI): Likewise.
3625 (XL_BE_V4SI): Likewise.
3626 (XL_BE_V4SI): Likewise.
3627 (XL_BE_V2DI): Likewise.
3628 (XL_BE_V4SF): Likewise.
3629 (XL_BE_V2DF): Likewise.
3630 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3631 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3632 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3633 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3634 VSX_BUILTIN_VEC_XST_BE.
3635 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3636 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3638 (altivec_init_builtins): Replace conditional calls to def_builtin
3639 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3640 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3641 with unconditional calls. Remove calls to def_builtin for
3642 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3643 __builtin_vec_xst_be.
3644 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3645 to define_expand, and add alternate RTL generation for P8.
3646 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3647 vsx_ld_elemrev_v8hi.
3648 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3649 add alternate RTL generation for P8.
3650 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3651 vsx_ld_elemrev_v16qi.
3652 (vsx_st_elemrev_v8hi): Convert define_insn
3653 to define_expand, and add alternate RTL generation for P8.
3654 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3655 vsx_st_elemrev_v8hi.
3656 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3657 add alternate RTL generation for P8.
3658 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3659 vsx_st_elemrev_v16qi.
3661 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3664 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3665 TARGET_AVX512ER check.
3666 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3667 neither -mzeroupper nor -mno-zeroupper is used and
3668 TARGET_EMIT_VZEROUPPER is set.
3669 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3670 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3672 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3674 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3675 folding of vector compares.
3676 (fold_build_vec_cmp): New helper function.
3677 (fold_compare_helper): New helper function.
3678 (builtin_function_type): Add compare builtins to the list of functions
3679 having unsigned arguments. Cosmetic updates to comment indentation.
3680 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3681 the not+eq combination.
3683 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3685 PR tree-optimization/82726
3686 PR tree-optimization/70754
3687 * tree-predcom.c (order_drefs_by_pos): New function.
3688 (combine_chains): Move code setting has_max_use_after to...
3689 (try_combine_chains): ...here. New parameter. Sort combined chains
3690 according to position information.
3691 (tree_predictive_commoning_loop): Update call to above function.
3692 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3694 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3696 PR tree-optimization/82726
3698 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3700 PR tree-optimization/70754
3701 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3702 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3703 combined stmt before it if not NULL.
3704 (combine_chains): Process refs reversely and compute dominance point
3708 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3710 PR tree-optimization/79663
3711 * tree-predcom.c (combine_chains): Process refs in reverse order
3712 only for ZERO length chains, and add explaining comment.
3714 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3716 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3717 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3718 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3720 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3722 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3724 2017-11-15 Martin Liska <mliska@suse.cz>
3726 * tree-cfg.c (pass_warn_function_return::execute):
3727 Compare warn_return_type for greater than zero.
3729 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3733 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3734 to return true on Xeon and not on Xeon Phi.
3735 (ix86_check_avx256_register): Changed to ...
3736 (ix86_check_avx_upper_register): ... this. Add extra check for
3737 VALID_AVX512F_REG_OR_XI_MODE.
3738 (ix86_avx_u128_mode_needed): Changed
3739 ix86_check_avx256_register to ix86_check_avx_upper_register.
3740 (ix86_check_avx256_stores): Changed to ...
3741 (ix86_check_avx_upper_stores): ... this. Changed
3742 ix86_check_avx256_register to ix86_check_avx_upper_register.
3743 (ix86_avx_u128_mode_after): Changed
3744 avx_reg256_found to avx_upper_reg_found. Changed
3745 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3746 (ix86_avx_u128_mode_entry): Changed
3747 ix86_check_avx256_register to ix86_check_avx_upper_register.
3748 (ix86_avx_u128_mode_exit): Ditto.
3749 * config/i386/i386.h: (host_detect_local_cpu): New define.
3751 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3753 * config/arm/xgene1.md (xgene1): Split into automatons
3754 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3755 (xgene1_f_load): Adjust reservations and/or types.
3756 (xgene1_f_store): Likewise.
3757 (xgene1_load_pair): Likewise.
3758 (xgene1_store_pair): Likewise.
3759 (xgene1_fp_load1): Likewise.
3760 (xgene1_load1): Likewise.
3761 (xgene1_store1): Likewise.
3762 (xgene1_move): Likewise.
3763 (xgene1_alu): Likewise.
3764 (xgene1_simd): Likewise.
3765 (xgene1_bfm): Likewise.
3766 (xgene1_neon_load1): Likewise.
3767 (xgene1_neon_store1): Likewise.
3768 (xgene1_neon_logic): Likewise.
3769 (xgene1_neon_st1): Likewise.
3770 (xgene1_neon_ld1r): Likewise.
3771 (xgene1_alu_cond): Added.
3772 (xgene1_shift_reg): Likwise.
3773 (xgene1_bfx): Likewise.
3774 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3776 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3779 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3781 (can_widen_mult_without_libcall): New function.
3782 (expand_mul_overflow): If only checking unsigned mul overflow,
3783 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3784 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3785 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3786 (expand_DIVMOD): Formatting fix.
3787 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3788 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3789 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3791 PR tree-optimization/82977
3792 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3793 constructed temporary to strlen_to_stridx.put.
3795 2017-11-15 Martin Liska <mliska@suse.cz>
3797 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3798 * configure: Regenerate.
3800 2017-11-15 Martin Liska <mliska@suse.cz>
3803 * config/sh/sh-mem.cc: Use proper probability for
3806 2017-11-14 Jeff Law <law@redhat.com>
3808 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3809 the red zone for stack_clash_protection_final_dynamic_probe targets
3810 when the total dynamic stack size is zero bytes.
3812 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3813 blocks is post order.
3815 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3817 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3818 * final.c (rest_of_clean_state): Set it for the
3819 -fcompare-debug dump.
3820 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3821 class when TDF_COMPARE_DEBUG is set.
3823 * dwarf2out.c (gen_producer_string): Discard
3826 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3829 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3830 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3831 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3832 (__TGMATH_REAL_2_3): Remove macros.
3833 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3834 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3836 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3838 (remquo): Define using __TGMATH_REAL_3.
3840 2017-11-14 Jeff Law <law@redhat.com>
3842 * vr-values.c: New file with contents extracted from tree-vrp.c.
3843 * Makefile.in (OBJS): Add vr-values.o
3844 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3845 (set_value_range, set_and_canonicalize_value_range): Likewise.
3846 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3847 (value_range_constant_singleton, symbolic_range_p): Likewise.
3848 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3849 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3850 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3851 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3852 (find_case_label_range, find_case_label_index): Likewise.
3853 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3854 (range_int_cst_singleton_p, value_inside_range): Likewise.
3855 (get_single_symbol): Likewise.
3856 (switch_update): Move structure definition here.
3857 (to_remove_edges, to_update_switch_stmts): Provide externs.
3858 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3859 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3860 (vrp_val_is_min, set_value_range): Likewise.
3861 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3862 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3863 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3864 (range_is_nonnull, range_int_cst_p): Likewwise.
3865 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3866 (get_single_symbol, operand_less_p): Likewise
3867 (compare_values_warnv, compare_values): Likewise.
3868 (value_inside_range, value_range_constant_singleton): Likewise.
3869 (zero_nonzero_bitgs_from_vr): Likewise.
3870 (extract_range_from_binary_expr_1): Likewise.
3871 (overflow_comparison_p): Likewise.
3872 (to_remove_edges, to_update_switch_stmts): Likewise.
3873 (find_case_label-index, find_case_label_range): Likewise.
3874 (switch_update, set_value_range_to_nonnegative): Remove.
3875 (set_value_range_to_truthvalue): Likewise.
3876 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3877 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3878 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3879 (find_case_label_ranges, test_for_singularity): Likewise.
3880 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3881 (x_vr_values): Move to its remaining use site.
3883 2017-11-10 Jeff Law <law@redhat.com>
3885 * vr-values.h (VR_INITIALIZER): Move #define here.
3886 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3887 * Makefile.in (OBJS): Add tree-evrp.o
3888 * tree-vrp.h (assert_info): Move structure definition here.
3889 (set_value_range_to_varying): Prototype.
3890 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3891 (infer_value_range, register_edge_assert_for): Likewise.
3892 (stmt_interesting_for_vrp): Likewise.
3893 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3894 (set_value_range_to_varying): No longer static.
3895 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3896 (infer_value_range, register_edge_assert_for): Likewise.
3898 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3900 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3901 generate the XXBRD instruction.
3903 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3904 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3905 as being compatible if -mabi=ieeelongdouble.
3906 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3907 to setup float128 built-ins with hardware support.
3908 (BU_FLOAT128_HW_2): Likewise.
3909 (BU_FLOAT128_HW_3): Likewise.
3910 (BU_FLOAT128_HW_VSX_1): Likewise.
3911 (BU_FLOAT128_HW_VSX_2): Likewise.
3912 (scalar_extract_expq): Change float128 built-in functions to
3913 accommodate having both KFmode and TFmode functions. Use the
3914 KFmode variant as the default.
3915 (scalar_extract_sigq): Likewise.
3916 (scalar_test_neg_qp): Likewise.
3917 (scalar_insert_exp_q): Likewise.
3918 (scalar_insert_exp_qp): Likewise.
3919 (scalar_test_data_class_qp): Likewise.
3920 (sqrtf128_round_to_odd): Delete processing the round to odd
3921 built-in functions as special built-in functions, and define them
3922 as float128 built-ins. Use the KFmode variant as the default.
3923 (truncf128_round_to_odd): Likewise.
3924 (addf128_round_to_odd): Likewise.
3925 (subf128_round_to_odd): Likewise.
3926 (mulf128_round_to_odd): Likewise.
3927 (divf128_round_to_odd): Likewise.
3928 (fmaf128_round_to_odd): Likewise.
3929 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3930 support for KFmode and TFmode xststdcqp calls.
3931 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3932 point, switch the built-in handlers for the get/set float128
3933 exponent, get float128 mantissa, float128 test built-ins, and the
3934 float128 round to odd built-in functions. Eliminate creating the
3935 float128 round to odd built-in functions as special built-ins.
3936 (rs6000_init_builtins): Eliminate special creation of the float128
3937 round to odd built-in functions.
3938 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3939 function insns to support both TFmode and KFmode variants.
3940 (xsxsigqp_<mode>): Likewise.
3941 (xsiexpqpf_<mode>): Likewise.
3942 (xsiexpqp_<mode>): Likewise.
3943 (xststdcqp_<mode>): Likewise.
3944 (xststdcnegqp_<mode>): Likewise.
3945 (xststdcqp_<mode>): Likewise.
3947 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3949 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3950 unused path_in_freq_ptr parameter.
3951 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3953 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3955 * ipa-inline.c (edge_badness): Dump sreal frequency.
3956 (compute_inlined_call_time): Match natural implementaiton ...
3957 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3958 forgotten division by CGRAPH_FREQ_BASE.
3960 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3962 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3963 Solaris 11. Update comment.
3964 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3966 * config/sol2.h (STARTFILE_SPEC): Likewise.
3967 * configure: Regenerate.
3969 2017-11-14 Carl Love <cel@us.ibm.com>
3971 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3972 le_ and be_ prefixes to swap* variables. Remove
3973 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3976 2017-11-14 Jason Merrill <jason@redhat.com>
3978 Support GTY((cache)) on hash_map.
3979 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3980 (ggc_cache_remove): Override it instead of ggc_mx.
3981 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3982 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3983 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3984 (simple_cache_map_traits): Override maybe_mx.
3985 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3986 (hash_map): Friend gt_cleare_cache.
3987 (gt_cleare_cache): New.
3988 * tree.h (tree_cache_traits): New hash_map traits class.
3989 (tree_cache_map): New typedef.
3991 2017-11-14 Richard Biener <rguenther@suse.de>
3993 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3994 paramter and handling.
3995 (cleanup_control_flow_bb): Likewise.
3996 (cleanup_control_flow_pre): New helper performing a DFS walk
3997 to call cleanup_control_flow_bb in PRE order.
3998 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3999 via cleanup_control_flow_pre.
4001 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4003 * config/aarch64/aarch64-simd.md
4004 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4005 (*aarch64_simd_bsl<mode>_alt): Likewise.
4006 (aarch64_simd_bsldi_internal): New.
4007 (aarch64_simd_bsldi_alt): Likewise.
4009 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4011 * tracer.c (better_p): Do not compare frequencies.
4012 * reg-stack.c (better_edge): Likewise.
4013 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4016 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4018 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4019 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4020 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4021 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4022 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4023 * predict.c (maybe_hot_frequency_p): Inline to ...
4024 (maybe_hot_count_p): ... here; rewrite to counts.
4025 (counts_to_freqs): Rename to ...
4026 (update_max_bb_count): ... this one.
4027 (expensive_function_p): Use counts.
4028 (estimate_bb_frequencies): Update.
4029 (rebuild_frequencies): Update.
4030 * predict.h (counts_to_freqs): Rename to ...
4031 (update_max_bb_count): ... this one.
4032 * profile.c (compute_branch_probabilities): Add debug info
4033 * tree-inline.c (expand_call_inline): Update debug info.
4034 (optimize_inline_calls): Use update_max_bb_count..
4035 (tree_function_versioning): Use update_max_bb_count..
4036 * value-prof.c (gimple_value_profile_transformations):
4037 Do not use update_max_bb_count.
4039 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4041 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4042 always use frequencies.
4044 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4046 * bb-reorder.c: Remove frequencies from comments.
4047 (better_edge_p): Use profile counts.
4048 (find_traces): Dump profile counts.
4049 (rotate_loop): Use profile counts.
4050 (find_traces_1_round): Likewise.
4051 (connect_better_edge_p): Use counts instead of probabilities for
4053 (copy_bb_p): Drop early check for non-0 frequency.
4054 (sanitize_hot_paths): Update comments.
4056 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4058 * ipa-split.c (struct split_point): Add count.
4059 (consider_split): Do not compute incoming frequency; compute incoming
4060 count and store it to split_point.
4061 (split_function): Set count of the call to split part correctly.
4063 2017-11-13 Carl Love <cel@us.ibm.com>
4065 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4067 2017-11-13 Tom Tromey <tom@tromey.com>
4069 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4071 2017-11-13 Carl Love <cel@us.ibm.com>
4073 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4074 Add support for builtins:
4075 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4076 vector {un,}signed {char,int,short},
4077 vector {un,}signed {char,int,short}) arguments.
4078 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4079 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4080 Add BU_P9V_AV_2 expansions for the builtins.
4081 * config/rs6000/altivec.h (vec_first_match_index,
4082 vec_first_mismatch_index, vec_first_match_or_eos_index,
4083 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4084 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4085 new extern declaration.
4086 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4087 * config/rs6000/vsx.md (first_match_index_<mode>,
4088 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4089 first_mismatch_or_eos_index_<mode>): Add define expand.
4090 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4091 * doc/extend.texi: Update the built-in documenation file for the new
4094 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4096 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4097 into the min/max operations for _Float<N> and _Float<N>X types.
4099 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4102 * dwarf2out.c (do_eh_frame): New static variable.
4103 (dwarf2out_begin_prologue): Set it.
4104 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4106 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4108 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4110 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4112 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4115 * debug.h (dwarf2out_do_eh_frame): Declare.
4116 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4117 (dwarf2out_do_frame): Use it.
4118 (dwarf2out_do_cfi_asm): Likewise.
4119 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4120 (dwarf2out_assembly_start): Likewise.
4121 (dwarf2out_begin_prologue): Fix comment.
4122 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4123 if the target needs either debug or unwind DWARF2 info.
4124 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4125 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4126 -ftrapping-math, -ftrapv and -fwrapv.
4128 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4130 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4131 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4132 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4133 (estimate_edge_size_and_time): Likewise.
4134 (ipa_merge_fn_summary_after_inlining): Likewise.
4135 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4136 (compute_uninlined_call_time): Use sreal_frequency.
4137 (compute_inlined_call_time): Likewise.
4138 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4139 * profile-count.c: Include sreal.h.
4140 (profile_count::to_sreal_scale): New.
4141 * profile-count.h: Forward declare sreal.
4142 (profile_count::to_sreal_scale): Declare.
4144 2017-11-13 Nathan Sidwell <nathan@acm.org>
4146 * diagnostic.c (maybe_line_and_column): New.
4147 (diagnostic_get_location_text): Use it.
4148 (diagnostic_report_current_module): Likewise.
4149 (test_diagnostic_get_location_text): Add tests.
4151 2017-11-13 Luis Machado <luis.machado@linaro.org>
4153 * doc/md.texi (Specifying processor pipeline description): Fix
4154 incorrect latency for the div instruction example.
4156 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4158 PR tree-optimization/78821
4159 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4160 that bit_not_p is the same.
4161 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4162 (split_group): Count precisely bit_not_p bits in each statement.
4163 (invert_op): New function.
4164 (imm_store_chain_info::output_merged_store): Use invert_op to
4165 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4166 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4168 2017-11-13 Martin Liska <mliska@suse.cz>
4170 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4171 (struct source_info): Likewise.
4172 (add_branch_counts): Likewise.
4173 (add_line_counts): Likewise.
4174 (function_summary): Likewise.
4175 (output_intermediate_line): Likewise.
4176 (generate_results): Likewise.
4178 2017-11-13 Martin Liska <mliska@suse.cz>
4180 * gcov.c (struct block_info): Remove typedef for block_t.
4181 (struct line_info): Likewise.
4182 (line_info::has_block): Likewise.
4183 (EXIT_BLOCK): Likewise.
4184 (unblock): Likewise.
4185 (circuit): Likewise.
4186 (get_cycles_count): Likewise.
4187 (process_file): Likewise.
4188 (read_graph_file): Likewise.
4189 (solve_flow_graph): Likewise.
4190 (find_exception_blocks): Likewise.
4191 (add_line_counts): Likewise.
4192 (accumulate_line_info): Likewise.
4193 (output_line_details): Likewise.
4195 2017-11-13 Martin Liska <mliska@suse.cz>
4197 * gcov.c (struct arc_info): Remove typedef for arc_t.
4198 (struct line_info): Likewise.
4199 (add_branch_counts): Likewise.
4200 (output_branch_count): Likewise.
4201 (function_info::~function_info): Likewise.
4202 (circuit): Likewise.
4203 (output_intermediate_line): Likewise.
4204 (read_graph_file): Likewise.
4205 (solve_flow_graph): Likewise.
4206 (find_exception_blocks): Likewise.
4207 (add_line_counts): Likewise.
4208 (accumulate_line_info): Likewise.
4209 (output_line_details): Likewise.
4210 (output_function_details): Likewise.
4212 2017-11-13 Martin Liska <mliska@suse.cz>
4214 * gcov.c (struct function_info): Remove typedef for function_t.
4215 (struct source_info): Likewise.
4216 (source_info::get_functions_at_location): Likewise.
4217 (solve_flow_graph): Likewise.
4218 (find_exception_blocks): Likewise.
4219 (add_line_counts): Likewise.
4220 (output_intermediate_file): Likewise.
4221 (process_file): Likewise.
4222 (generate_results): Likewise.
4223 (release_structures): Likewise.
4224 (read_graph_file): Likewise.
4225 (read_count_file): Likewise.
4226 (accumulate_line_counts): Likewise.
4227 (output_lines): Likewise.
4229 2017-11-13 Martin Liska <mliska@suse.cz>
4231 * gcov.c (function_info::function_info): Remove num_counts
4232 and add vector<gcov_type>.
4233 (function_info::~function_info): Use the vector.
4234 (process_file): Likewise.
4235 (read_graph_file): Likewise.
4236 (read_count_file): Likewise.
4237 (solve_flow_graph): Likewise.
4239 2017-11-13 Martin Liska <mliska@suse.cz>
4241 * gcov.c (function_info::is_artificial): New function.
4242 (process_file): Erase all artificial early.
4243 (generate_results): Skip as all artificial are already
4246 2017-11-13 Martin Liska <mliska@suse.cz>
4248 * gcov.c (read_graph_file): Store to global vector of functions.
4249 (read_count_file): Iterate the vector.
4250 (process_file): Likewise.
4251 (generate_results): Likewise.
4252 (release_structures): Likewise.
4254 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4256 PR tree-optimization/82954
4257 * gimple-ssa-store-merging.c
4258 (imm_store_chain_info::coalesce_immediate_stores): If
4259 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4261 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4263 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4264 Upddate call to ENDIAN_LANE_N.
4265 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4266 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4267 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4268 and use aarch64_endian_lane_rtx.
4269 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4271 2017-11-12 Tom de Vries <tom@codesourcery.com>
4273 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4275 2017-11-12 Tom de Vries <tom@codesourcery.com>
4277 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4279 2017-11-12 Tom de Vries <tom@codesourcery.com>
4281 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4283 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4284 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4285 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4287 2017-11-11 Martin Sebor <msebor@redhat.com>
4290 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4293 * prefic.c (translate_name): Replace strncpy with memcpy to
4294 avoid -Wstringop-truncation.
4296 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4298 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4300 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4302 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4304 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4306 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4309 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4311 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4312 enable generating XXBRH if the value is in a vector register.
4313 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4314 value is in a vector register.
4315 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4316 register to register bswap64's instead of doing the GPR sequence
4317 used on previous machines.
4318 (bswapdi2_xxbrd): New insn.
4319 (bswapdi2_reg): Disallow on ISA 3.0.
4320 (register to register bswap64 splitter): Do not split the insn on
4321 ISA 3.0 systems that use XXBRD.
4323 2017-11-10 Martin Sebor <msebor@redhat.com>
4326 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4328 (find_subframework_file): Same.
4330 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4332 * auto-profile.c (afdo_indirect_call): Drop frequency.
4333 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4334 (cgraph_node::create_edge): Drop frequency argument.
4335 (cgraph_node::create_indirect_edge): Drop frequency argument.
4336 (cgraph_edge::make_speculative): Drop frequency arguments.
4337 (cgraph_edge::resolve_speculation): Do not update frequencies
4338 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4339 (cgraph_node::dump): Check consistency in IPA mode.
4340 (cgraph_edge::maybe_hot_p): Use IPA counter.
4341 (cgraph_edge::verify_count_and_frequency): Rename to ...
4342 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4343 (cgraph_node::verify_node): Update.
4344 * cgraph.h (struct cgraph_edge): Drop frequency.
4345 (cgraph_edge::frequency): New function.
4346 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4348 (cgraph_edge::rebuild_edges): Likewise.
4349 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4350 (duplicate_thunk_for_node): Do not pass frequency.
4351 (cgraph_node::create_clone): Scale only counts.
4352 (cgraph_node::create_virtual_clone): Do not pass frequency.
4353 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4354 (cgraph_node::create_version_clone): Do not pass frequency.
4355 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4356 (cgraph_node::expand_thunk): Do not pass frequency.
4357 (cgraph_node::create_wrapper): Do not pass frequency.
4358 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4360 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4361 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4362 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4363 (gather_caller_stats): Use frequency function.
4364 (ipcp_cloning_candidate_p): Use frequency function.
4365 (ipcp_propagate_stage): Use frequency function.
4366 (get_info_about_necessary_edges): Use frequency function.
4367 (update_profiling_info): Update only IPA profile.
4368 (update_specialized_profile): Use frequency functoin.
4369 (perhaps_add_new_callers): Update only IPA profile.
4370 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4371 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4372 (dump_ipa_call_summary): Use frequency function.
4373 (estimate_edge_size_and_time): Use frequency function.
4374 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4375 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4376 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4377 (update_noncloned_counts): ... ths one; scale counts only.
4378 (clone_inlined_nodes): Do not scale frequency.
4379 (inline_call): Do not pass frequency.
4380 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4381 (compute_inlined_call_time): Use IPA profile.
4382 (want_inline_small_function_p): Use IPA profile.
4383 (want_inline_self_recursive_call_p): Use IPA profile.
4384 (edge_badness): Use IPA profile.
4385 (lookup_recursive_calls): Use IPA profile.
4386 (recursive_inlining): Do not pass frequency.
4387 (resolve_noninline_speculation): Do not update frequency.
4388 (inline_small_functions): Collect max of IPA profile.
4389 (dump_overall_stats): Dump IPA porfile.
4390 (dump_inline_stats): Dump IPA porfile.
4391 (ipa_inline): Collect IPA stats.
4392 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4393 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4394 (ipa_propagate_frequency): Use frequency function.
4395 (ipa_profile): Cleanup.
4396 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4397 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4398 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4399 (input_node): Do not stream frequency.
4400 (input_edge): Do not stream frequency.
4401 (merge_profile_summaries): Scale only IPA profiles.
4402 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4403 * predict.c (drop_profile): Do not recompute frequency.
4404 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4405 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4406 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4407 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4408 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4409 * tree-inline.c (copy_bb): Do not scale frequency.
4410 (expand_call_inline): Do not scale frequency.
4411 (tree_function_versioning): Do not scale frequency.
4412 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4414 2017-11-10 Julia Koval <julia.koval@intel.com>
4416 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4417 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4418 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4419 (_mm256_maskz_gf2p8affine_epi64_epi8)
4420 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4421 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4422 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4423 (__builtin_ia32_vgf2p8affineqb_v32qi)
4424 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4425 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4427 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4431 (arm_option_override): Refactor.
4432 (arm_option_reconfigure_globals): New.
4433 (arm_options_perform_arch_sanity_checks): New.
4434 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4436 (arm_options_perform_arch_sanity_checks): Likewise
4438 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4440 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4441 (power9-qpmul): New.
4442 * rs6000/rs6000.md ("type" attr): Add qmul.
4443 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4444 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4445 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4447 2017-11-10 Martin Sebor <msebor@redhat.com>
4450 * builtins.c (compute_objsize): Handle arrays that
4451 compute_builtin_object_size likes to fail for. Make extern.
4452 * builtins.h (compute_objsize): Declare.
4453 (check_strncpy_sizes): New function.
4454 (expand_builtin_strncpy): Call check_strncpy_sizes.
4455 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4456 -Wstringop-truncation.
4457 (gimple_fold_builtin_strncat): Same.
4458 * gimple.c (gimple_build_call_from_tree): Set call location.
4459 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4460 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4461 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4462 (handle_builtin_strlen): Use strlen_to_stridx.
4463 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4465 Use strlen_to_stridx.
4466 (pass_strlen::execute): Release strlen_to_stridx.
4467 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4468 (-Wstringop-truncation): Document new option.
4470 2017-11-10 Martin Liska <mliska@suse.cz>
4472 PR gcov-profile/82702
4473 * gcov.c (main): Handle intermediate files in a different
4475 (get_gcov_intermediate_filename): New function.
4476 (output_gcov_file): Remove support of intermediate files.
4477 (generate_results): Allocate intermediate file.
4478 (release_structures): Clean-up properly fn_end.
4479 (output_intermediate_file): Start iterating with line 1.
4481 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4483 PR tree-optimization/82929
4484 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4485 ops_swapped_p non-static data member.
4486 (store_immediate_info::store_immediate_info): Clear it.
4487 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4488 ops set ops_swapped_p.
4489 (count_multiple_uses): Handle ops_swapped_p.
4491 2017-11-10 Martin Liska <mliska@suse.cz>
4493 * coverage.c (coverage_init): Stream information about
4494 support of has_unexecuted_blocks.
4495 * doc/gcov.texi: Document that.
4496 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4497 * gcov.c (read_graph_file): Likewise.
4498 (output_line_beginning): Fix a small issue with
4501 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4503 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4504 reference of trivial component.
4506 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4509 * gimple-ssa-store-merging.c
4510 (pass_store_merging::terminate_all_aliasing_chains): For
4511 gimple_store_p stmts also call refs_output_dependent_p.
4513 PR rtl-optimization/82913
4514 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4517 2017-11-09 Jeff Law <law@redhat.com>
4519 * vr-values.h: New file with vr_values class.
4520 * tree-vrp.c: Include vr-values.h
4521 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4522 data objects into the vr_values class.
4523 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4524 (get_value_range): Make it a member function within vr_values class.
4525 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4526 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4527 (op_with_constant_singleton_value_range): Likewise.
4528 (extract_range_for_var_from_comparison_expr): Likewise.
4529 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4530 (extract_range_from_binary_expr): Likewise.
4531 (extract_range_from_unary_expr): Likewise.
4532 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4533 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4534 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4535 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4536 (compare_name_with_value, compare_names): Likewise.
4537 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4538 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4539 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4540 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4541 (extract_range_from_phi_node): Likewise.
4542 (simplify_truth_ops_using_ranges): Likewise.
4543 (simplify_div_or_mod_using_ranges): Likewise.
4544 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4545 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4546 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4547 (simplify_float_conversion_using_ranges): Likewise.
4548 (simplify_internal_call_using_ranges): Likewise.
4549 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4550 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4551 poitner via x_vr_values for calls into gimple folder.
4552 (vrp_initialize_lattice): Make this the vr_values ctor.
4553 (vrp_free_lattice): Make this the vr_values dtor.
4554 (set_vr_value): New function.
4555 (class vrp_prop): Add vr_values data member. Add various member
4556 functions as well as member functions that delegate to vr_values.
4557 (check_array_ref): Make a member function within vrp_prop class.
4558 (search_for_addr_array, vrp_initialize): Likewise.
4559 (vrp_finalize): Likewise. Revamp to avoid direct access to
4560 vr_value, values_propagated, etc.
4561 (check_array_bounds): Extract vrp_prop class instance pointer from
4562 walk info structure. Use it to call member functions.
4563 (check_all_array_refs): Make a member function within vrp_prop class.
4564 Smuggle class instance pointer via walk info structure.
4565 (x_vr_values): New local static.
4566 (vrp_valueize): Use x_vr_values to get class instance.
4567 (vr_valueize_1): Likewise.
4568 (class vrp_folder): Add vr_values data member. Add various member
4569 functions as well as member functions that delegate to vr_values.
4570 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4571 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4572 class instance from vr_values. Use it to call member functions.
4573 (vrp_dom_walker): Add vr_values data member.
4574 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4575 instance via x_vr_values.
4576 (identify_jump_threads): Accept vr_values as argument. Store
4577 it into the walker structure.
4578 (evrp_dom_walker): Add vr_values class data member. Add various
4580 (evrp_dom_walker::try_find_new_range): Use vr_values data
4581 member to access the memory allocator.
4582 (evrp_dom_walker::before_dom_children): Store vr_values class
4583 instance into the vrp_folder class.
4584 (evrp_dom_walker::push_value_range): Rework to avoid direct
4585 access to num_vr_values and vr_value.
4586 (evrp_dom_walker::pop_value_range): Likewise.
4587 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4588 Use vr_values to get to dump_all_value_ranges member function.
4589 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4590 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4591 Pass vr_values class instance down to identify_jump_threads.
4592 Remove call to vrp_free_lattice.
4593 (debug_all_value_ranges): Remove.
4595 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4596 (vrp_folder): Likewise.
4598 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4599 Get it from the existing bitmap instead.
4600 (vrp_intersect_ranges_1): Likewise.
4602 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4604 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4606 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4607 set bit_not_p to it.
4608 (imm_store_chain_info::coalesce_immediate_stores): Break group
4609 if bit_not_p is different.
4610 (count_multiple_uses, split_group,
4611 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4612 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4613 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4614 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4615 to store_immediate_info ctor.
4617 2017-11-09 Jim Wilson <jimw@sifive.com>
4619 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4620 (scan_prog_file): Likewise.
4622 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4624 * bb-reorder.c (max_entry_frequency): Remove.
4625 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4626 connect_traces, push_to_next_round_p): Remove prototypes.
4627 (find_traces_1_round): Use counts only.
4628 (push_to_next_round_p): Likewise.
4629 (find_traces): Likewise.
4630 (rotate_loop): Likewise.
4631 (find_traces_1_round): Likewise.
4632 (connect_traces): Likewise.
4633 (edge_order): Likewise.
4635 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4637 * config/arm/arm.c (output_return_instruction): Add comments to
4638 indicate requirement for cmse_nonsecure_entry return to account
4639 for the size of clearing instruction output here.
4640 (thumb_exit): Likewise.
4641 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4642 return in hardfloat mode.
4644 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4646 * config/rs6000/rs6000.c (machine_function): Add a bool,
4647 "toc_is_wrapped_separately".
4648 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4649 if it wasn't explicitly set or unset, we are optimizing for speed, and
4650 doing separate shrink-wrapping.
4651 (rs6000_get_separate_components): Enable the TOC component if
4652 saving the TOC register in the prologue.
4653 (rs6000_components_for_bb): Handle the TOC component.
4654 (rs6000_emit_prologue_components): Store the TOC register where needed.
4655 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4656 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4658 2017-11-09 Martin Jambor <mjambor@suse.cz>
4660 * ipa-param-manipulation.c: New file.
4661 * ipa-param-manipulation.h: Likewise.
4662 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4663 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4664 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4665 (ipa_parm_adjustment): Likewise.
4666 (ipa_parm_adjustment_vec): Likewise.
4667 (ipa_get_vector_of_formal_parms): Moved declaration to
4668 ipa-param-manipulation.h.
4669 (ipa_get_vector_of_formal_parm_types): Likewise.
4670 (ipa_modify_formal_parameters): Likewise.
4671 (ipa_modify_call_arguments): Likewise.
4672 (ipa_combine_adjustments): Likewise.
4673 (ipa_dump_param_adjustments): Likewise.
4674 (ipa_modify_expr): Likewise.
4675 (ipa_get_adjustment_candidate): Likewise.
4676 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4677 ipa-param-manipulation.c.
4678 (ipa_get_vector_of_formal_parm_types): Likewise.
4679 (ipa_modify_formal_parameters): Likewise.
4680 (ipa_modify_call_arguments): Likewise.
4681 (ipa_modify_expr): Likewise.
4682 (get_ssa_base_param): Likewise.
4683 (ipa_get_adjustment_candidate): Likewise.
4684 (index_in_adjustments_multiple_times_p): Likewise.
4685 (ipa_combine_adjustments): Likewise.
4686 (ipa_dump_param_adjustments): Likewise.
4687 * tree-sra.c: Also include ipa-param-manipulation.h
4688 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4691 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4692 Alan Hayward <alan.hayward@arm.com>
4693 David Sherwood <david.sherwood@arm.com>
4695 * doc/sourcebuild.texi (vect_masked_store): Document.
4697 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4698 Alan Hayward <alan.hayward@arm.com>
4699 David Sherwood <david.sherwood@arm.com>
4701 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4703 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4704 Alan Hayward <alan.hayward@arm.com>
4705 David Sherwood <david.sherwood@arm.com>
4707 * doc/sourcebuild.texi (vect_variable_length): Document.
4709 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4710 Alan Hayward <alan.hayward@arm.com>
4711 David Sherwood <david.sherwood@arm.com>
4713 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4715 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4716 Alan Hayward <alan.hayward@arm.com>
4717 David Sherwood <david.sherwood@arm.com>
4719 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4721 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4722 Alan Hayward <alan.hayward@arm.com>
4723 David Sherwood <david.sherwood@arm.com>
4725 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4726 previously undocumented selectors.
4727 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4729 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4731 * doc/rtl.texi (const_vector): Say that elements can be
4732 const_wide_ints too.
4733 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4734 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4735 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4736 * optabs.c (expand_vector_broadcast): Likewise.
4738 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4739 Alan Hayward <alan.hayward@arm.com>
4740 David Sherwood <david.sherwood@arm.com>
4742 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4743 scaled index even if the unscaled address was invalid.
4744 Don't increase the complexity of using a scale in that case.
4746 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4747 Alan Hayward <alan.hayward@arm.com>
4748 David Sherwood <david.sherwood@arm.com>
4750 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4751 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4752 UNITS_PER_WORD bytes.
4753 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4754 based on REGMODE_NATURAL_SIZE of the inner mode.
4755 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4756 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4757 * expmed.c (lowpart_bit_field_p): Divide the value up into
4758 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4759 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4760 whether something is likely to occupy more than one register.
4762 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4765 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4766 profile_count::adjust_for_ipa_scaling.
4767 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4768 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4770 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4772 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4774 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4775 (split_group): Add total_orig and total_new arguments, estimate the
4776 number of statements related to the store group without store merging
4777 and with store merging.
4778 (imm_store_chain_info::output_merged_store): Adjust split_group
4779 callers, punt if estimated number of statements with store merging
4780 is not smaller than estimated number of statements without it.
4782 (handled_load): Remove has_single_use checks.
4783 (pass_store_merging::process_store): Likewise.
4785 2017-11-09 Richard Biener <rguenther@suse.de>
4787 PR tree-optimization/82902
4788 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4790 2017-11-09 Martin Liska <mliska@suse.cz>
4793 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4796 2017-11-09 Martin Liska <mliska@suse.cz>
4798 PR tree-optimization/82669
4799 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4801 2017-11-09 Martin Liska <mliska@suse.cz>
4803 PR gcov-profile/48463
4804 * coverage.c (coverage_begin_function): Output also end locus
4805 of a function and information whether the function is
4807 * gcov-dump.c (tag_function): Parse and print the information.
4808 * gcov.c (INCLUDE_MAP): Add include.
4809 (INCLUDE_SET): Likewise.
4810 (struct line_info): Move earlier in the source file because
4811 of vector<line_info> in function_info structure.
4812 (line_info::line_info): Likewise.
4813 (line_info::has_block): Likewise.
4814 (struct source_info): Add new member index.
4815 (source_info::get_functions_at_location): New function.
4816 (function_info::group_line_p): New function.
4817 (output_intermediate_line): New function.
4818 (output_intermediate_file): Use the mentioned function.
4819 (struct function_start): New.
4820 (struct function_start_pair_hash): Likewise.
4821 (process_file): Add code that identifies group functions.
4822 Assign lines either to global or function scope.
4823 (generate_results): Skip artificial functions.
4824 (find_source): Assign index for each source file.
4825 (read_graph_file): Read new flag artificial and end_line.
4826 (add_line_counts): Assign it either to global of function scope.
4827 (accumulate_line_counts): Isolate core of the function to
4828 accumulate_line_info and call it for both function and global
4830 (accumulate_line_info): New function.
4831 (output_line_beginning): Fix GNU coding style.
4832 (print_source_line): New function.
4833 (output_line_details): Likewise.
4834 (output_function_details): Likewise.
4835 (output_lines): Iterate both source (global) scope and function
4837 (struct function_line_start_cmp): New class.
4838 * doc/gcov.texi: Reflect changes in documentation.
4840 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4843 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4844 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4845 and similarly for not instead of neg.
4847 2017-11-08 Andi Kleen <ak@linux.intel.com>
4849 * config/i386/i386.opt: Add -mforce-indirect-call.
4850 * config/i386/predicates.md: Check for flag_force_indirect_call.
4851 * doc/invoke.texi: Document -mforce-indirect-call
4853 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4855 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4857 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4858 config/riscv/riscv.c (predict.h): New include.
4859 (riscv_slow_unaligned_access_p): No longer static.
4860 (riscv_block_move_straight): Add require.
4861 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4863 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4866 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4867 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4868 nonimmediate_operand predicate for operand 1 instead of
4871 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4873 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4875 * config/aarch64/constraints.md (Uml): New constraint.
4876 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4879 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4881 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4882 of two vec_duplicates into a vec_concat.
4884 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4886 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4887 Simplify vec_merge of vec_duplicate and vec_concat.
4888 * config/aarch64/constraints.md (Utq): New constraint.
4889 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4892 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4894 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4895 Simplify vec_merge of vec_duplicate and const_vector.
4896 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4898 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4899 mode iterator. Update predicate on operand 1 to
4900 handle non-const_vec constants. Delete constraints.
4901 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4903 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4905 PR tree-optimization/78821
4906 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4908 (store_operand_info::store_operand_info): Initialize it to false.
4909 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4910 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4911 store in the group, and if chain_info is non-NULL, to ignore altogether
4913 (compatible_load_p): Fail if bit_not_p does not match.
4914 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4915 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4916 (pass_store_merging::process_store): Adjust
4917 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4918 call terminate_all_aliasing_chains newly when adding a store into
4919 a chain with non-NULL chain_info.
4921 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4923 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4925 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4927 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4929 (aarch64_layout_frame): Initialise emit_frame_chain.
4930 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4931 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4933 2017-11-08 Martin Liska <mliska@suse.cz>
4935 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4936 of gimple_call_internal_p.
4938 2017-11-07 Tom de Vries <tom@codesourcery.com>
4940 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4942 2017-11-07 Tom de Vries <tom@codesourcery.com>
4944 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4947 2017-11-08 Martin Liska <mliska@suse.cz>
4950 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4952 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4954 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4956 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4959 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4960 (*cmp<mode>_ccz_1): New insn with $k alternative.
4963 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4964 RTX_COMM_COMPARE as commutative as well.
4965 (ix86_binary_operator_ok): Formatting fix.
4966 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4967 *<code><mode>3<mask_name><round_saeonly_name>,
4968 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4969 *<s>mul<mode>3_highpart<mask_name>,
4970 *vec_widen_umult_even_v16si<mask_name>,
4971 *vec_widen_umult_even_v8si<mask_name>,
4972 *vec_widen_umult_even_v4si<mask_name>,
4973 *vec_widen_smult_even_v16si<mask_name>,
4974 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4975 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4976 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4977 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4978 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4979 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4980 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4981 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4982 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4983 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4984 ix86_binary_operator_ok. Formatting fixes.
4985 (*<plusminus_insn><mode>3<mask_name><round_name>,
4986 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4989 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4991 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4992 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4993 eq and ne if TARGET_ISEL.
4994 (cmp): New code_iterator.
4995 (UNS, UNSU_, UNSIK): New code_attrs.
4996 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4997 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4999 ("eq<mode>3"): ... this.
5000 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5002 ("ne<mode>3"): ... this.
5004 2017-11-07 Julia Koval <julia.koval@intel.com>
5007 * common/config/i386/i386-common.c
5008 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5009 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5010 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5011 * config/i386/i386.opt: Ditto.
5012 * config/i386/i386.c (ix86_target_string): Ditto.
5013 (ix86_option_override_internal): Ditto.
5014 (ix86_init_mpx_builtins): Move MPX to args2.
5015 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5016 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5017 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5018 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5019 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5020 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5021 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5023 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5026 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5027 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5028 (zero-extendsidi peephole2): Remove peephole.
5030 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5033 * stor-layout.c: Include attribs.h.
5034 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5035 explicit lookup of "aligned" attribute.
5037 2017-11-07 Andrew Waterman <andrew@sifive.com>
5039 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5040 (riscv_expand_block_move): Likewise.
5041 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5043 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5044 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5045 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5046 (riscv_adjust_block_mem): Likewise.
5047 (riscv_block_move_loop): Likewise.
5048 (riscv_expand_block_move): Likewise.
5049 * config/riscv/riscv.md (movmemsi): New pattern.
5051 2017-11-07 Michael Clark <michaeljclark@mac.com>
5053 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5054 (MUSL_DYNAMIC_LINKER): Likewise.
5056 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5058 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5059 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5062 2017-11-07 Richard Biener <rguenther@suse.de>
5064 * match.pd: Fix build.
5066 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5067 Jackson Woodruff <jackson.woodruff@arm.com>
5069 PR tree-optimization/71026
5070 * match.pd: Canonicalize negate in division.
5072 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5075 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5077 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5079 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5080 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5082 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5084 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5085 non-scalar integral types.
5086 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5087 (-(A-B), -(~A)): New transformations.
5089 2017-11-07 Tom de Vries <tom@codesourcery.com>
5091 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5092 semicolon after "do {} while (0)".
5093 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5094 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5095 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5096 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5097 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5099 2017-11-07 Tom de Vries <tom@codesourcery.com>
5101 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5102 after "do {} while (0)".
5103 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5104 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5105 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5106 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5107 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5109 2017-11-07 Tom de Vries <tom@codesourcery.com>
5112 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5114 (arm_rtx_costs_internal): Add missing semicolon after
5115 HANDLE_NARROW_SHIFT_ARITH call.
5117 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5119 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5120 disable isel if it was not set explicitly.
5122 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5124 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5125 (add_type_attribute) likewise.
5127 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5129 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5130 of struct ix86_frame.
5131 (ix86_initial_elimination_offset): Likewise.
5132 (ix86_expand_split_stack_prologue): Likewise.
5134 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5136 * tree-vrp.h (enum value_range_type): Update stale comment.
5138 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5139 Alan Hayward <alan.hayward@arm.com>
5140 David Sherwood <david.sherwood@arm.com>
5142 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5143 (aarch64_expand_vec_perm_const): Take the number of units too.
5144 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5145 (aarch64_expand_vec_perm_const): Likewise.
5146 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5147 (vec_perm<mode>): Update accordingly.
5149 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5150 Alan Hayward <alan.hayward@arm.com>
5151 David Sherwood <david.sherwood@arm.com>
5153 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5154 Take the number of units too.
5155 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5156 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5157 but check for a vector mode before rather than after the call.
5158 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5159 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5160 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5161 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5162 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5163 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5164 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5165 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5166 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5167 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5168 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5169 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5170 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5171 (aarch64_sqdmull2_n<mode>): Update accordingly.
5173 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5174 Alan Hayward <alan.hayward@arm.com>
5175 David Sherwood <david.sherwood@arm.com>
5177 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5178 the number of units too.
5179 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5180 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5181 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5182 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5183 (vec_store_lanesxi<mode>): Update accordingly.
5185 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5186 Alan Hayward <alan.hayward@arm.com>
5187 David Sherwood <david.sherwood@arm.com>
5189 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5190 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5191 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5192 of units rather than the mode.
5193 * config/aarch64/iterators.md (nunits): New mode attribute.
5194 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5195 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5196 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5197 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5198 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5199 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5200 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5201 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5202 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5203 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5204 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5205 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5206 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5207 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5208 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5209 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5210 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5211 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5212 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5213 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5214 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5215 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5216 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5217 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5218 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5219 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5220 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5221 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5222 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5223 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5224 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5225 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5226 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5227 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5228 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5229 (aarch64_simd_vec_setv2di): Likewise.
5231 2017-11-06 Carl Love <cel@us.ibm.com>
5233 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5235 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5236 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5238 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5240 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5241 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5242 Add power 8 macro expansions.
5243 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5244 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5245 power 8 instructions. (VSX_XXBR): Add iterator.
5247 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5249 * config/arm/arm.md (predicable_short_it): Change default to "no",
5250 improve documentation, remove uses that are identical to the default.
5251 (enabled_for_depr_it): Rename to enabled_for_short_it.
5252 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5253 Remove default uses.
5254 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5255 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5256 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5257 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5259 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5262 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5263 float128 helper macros, which are no longer used after deleting
5264 the old 'q' built-in functions, and moving the round to odd
5265 built-in functions to being special built-in functions.
5266 (BU_FLOAT128_2): Likewise.
5267 (BU_FLOAT128_1_HW): Likewise.
5268 (BU_FLOAT128_2_HW): Likewise.
5269 (BU_FLOAT128_3_HW): Likewise.
5270 (FABSQ): Delete old 'q' built-in functions.
5271 (COPYSIGNQ): Likewise.
5272 (SQRTF128_ODD): Move round to odd built-in functions to be
5273 special built-in functions, so that we can handle
5274 -mabi=ieeelongdouble.
5275 (TRUNCF128_ODD): Likewise.
5276 (ADDF128_ODD): Likewise.
5277 (SUBF128_ODD): Likewise.
5278 (MULF128_ODD): Likewise.
5279 (DIVF128_ODD): Likewise.
5280 (FMAF128_ODD): Likewise.
5281 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5282 built-in names to 'f128'.
5283 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5284 old 'q' built-in functions, as the machine independent code for
5285 'f128' built-in functions handles this.
5286 (rs6000_expand_builtin): Add expansion for float128 round to odd
5287 functions, keying off on -mabi=ieeelongdouble of whether to use
5288 the KFmode or TFmode variant.
5289 (rs6000_init_builtins): Initialize the _Float128 round to odd
5291 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5292 _Float128 'q' built-in functions are now mapped into the new
5293 'f128' built-in functions.
5295 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5297 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5298 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5300 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5303 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5304 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5305 * toplev.c: Include "ipa-fnsummary.h".
5306 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5308 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5310 PR tree-optimization/82838
5311 * gimple-ssa-store-merging.c
5312 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5313 on a separate gimple_seq which is then appended to seq.
5315 2017-11-06 Jeff Law <law@redhat.com>
5318 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5319 (get_probe_interval): New functions.
5320 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5321 (ix86_adjust_stack_and_probe): Likewise.
5322 (output_adjust_stack_and_probe): Likewise.
5323 (ix86_emit_probe_stack_range): Likewise.
5324 (ix86_expand_prologue): Likewise.
5326 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5328 PR tree-optimization/82816
5329 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5330 if the modes of the two types are the same.
5331 (convert_plusminus_to_widen): Likewise.
5333 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5335 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5337 (usadv16qi): New define_expand.
5338 (usadv8hi): New define_expand.
5340 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5343 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5345 (inline_transform): Likewise.
5346 * predict.c (counts_to_freqs): Remove useless conditional.
5347 * profile-count.h (profile_count::apply_scale): Move sanity check.
5348 * tree-inline.c (copy_bb): Always scale.
5349 (copy_cfg_body): Likewise.
5351 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5354 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5356 (*cmp_ite0): Add enabled_for_depr_it attribute.
5357 (*cmp_ite1): Likewise.
5359 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5361 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5364 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5366 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5367 return the value by pointer.
5368 (extract_range_from_multiplicative_op_1): Update accordingly.
5369 Return as soon as an operation fails.
5371 2017-11-05 Tom de Vries <tom@codesourcery.com>
5374 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5375 (DEF_SANITIZER_BUILTIN): ... here.
5376 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5377 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5379 2017-11-05 Tom de Vries <tom@codesourcery.com>
5382 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5384 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5385 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5386 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5388 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5389 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5390 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5392 2017-11-05 Tom de Vries <tom@codesourcery.com>
5395 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5398 2017-11-04 Michael Clark <michaeljclark@mac.com>
5400 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5401 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5403 (*addsi3_extended): Likewise.
5404 (*addsi3_extended2): Likewise.
5405 (<optab>si3): Likewise.
5406 (<optab>di3): Likewise.
5407 (<optab><mode>3): Likewise.
5408 (<*optabe>si3_internal): Likewise.
5409 (zero_extendqi<SUPERQI:mode>2): Likewise.
5410 (*add<mode>hi3): Likewise.
5411 (*xor<mode>hi3): Likewise.
5412 (<optab>di3): Likewise.
5413 (*<optab>si3_extend): Likewise.
5414 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5415 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5416 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5418 2017-11-04 Andrew Waterman <andrew@sifive.com>
5420 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5421 TARGET_STRICT_ALIGN based upon -mtune argument.
5423 2017-11-04 Andrew Waterman <andrew@sifive.com>
5425 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5427 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5429 * config/i386/i386.c (choose_basereg): Use optional scratch
5430 register and add assertion.
5431 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5432 needed, and don't allocate stack.
5433 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5434 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5435 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5437 2017-11-03 Jeff Law <law@redhat.com>
5439 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5440 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5441 to probe at the start of a noreturn function.
5443 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5445 PR tree-optimization/78821
5446 * gimple-ssa-store-merging.c: Update the file comment.
5447 (MAX_STORE_ALIAS_CHECKS): Define.
5448 (struct store_operand_info): New type.
5449 (store_operand_info::store_operand_info): New constructor.
5450 (struct store_immediate_info): Add rhs_code and ops data members.
5451 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5452 and op1r arguments to the ctor, initialize corresponding data members.
5453 (struct merged_store_group): Add load_align_base and load_align
5455 (merged_store_group::merged_store_group): Initialize them.
5456 (merged_store_group::do_merge): Update them.
5457 (merged_store_group::apply_stores): Pick the constant for
5458 encode_tree_to_bitpos from one of the two operands, or skip
5459 encode_tree_to_bitpos if neither operand is a constant.
5460 (class pass_store_merging): Add process_store method decl. Remove
5461 bool argument from terminate_all_aliasing_chains method decl.
5462 (pass_store_merging::terminate_all_aliasing_chains): Remove
5463 var_offset_p argument and corresponding handling.
5464 (stmts_may_clobber_ref_p): New function.
5465 (compatible_load_p): New function.
5466 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5467 if there is overlap and rhs_code is not INTEGER_CST. For
5468 non-overlapping stores terminate group if rhs is not mergeable.
5469 (get_alias_type_for_stmts): Change first argument from
5470 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5471 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5472 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5474 (get_location_for_stmts): Change first argument from
5475 auto_vec<gimple *> & to vec<gimple *> &.
5476 (struct split_store): Remove orig_stmts data member, add orig_stores.
5477 (split_store::split_store): Create orig_stores rather than orig_stmts.
5478 (find_constituent_stmts): Renamed to ...
5479 (find_constituent_stores): ... this. Change second argument from
5480 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5481 to info structures rather than the statements.
5482 (split_group): Rename ALLOW_UNALIGNED argument to
5483 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5484 it. Adjust find_constituent_stores caller.
5485 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5486 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5487 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5488 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5489 (mem_valid_for_store_merging): New function.
5490 (handled_load): New function.
5491 (pass_store_merging::process_store): New method.
5492 (pass_store_merging::execute): Use process_store method. Adjust
5493 terminate_all_aliasing_chains caller.
5495 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5497 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5498 Return true for more constants, symbols and label references.
5499 (aarch64_valid_floating_const): Remove unused function.
5501 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5504 * config/aarch64/aarch64.c (aarch64_layout_frame):
5505 Undo forcing of LR at bottom of frame.
5507 2017-11-03 Jeff Law <law@redhat.com>
5510 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5511 for int_registers_saved.
5513 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5514 extracted from tree-ssa-dom.c.
5515 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5516 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5517 (record_equivalences_from_incoming_edge): Add additional argument
5518 to single_pred_edge_ignoring_loop_edges call.
5519 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5520 (uncprop_dom_walker::before_dom_children): Add additional argument
5521 to single_pred_edge_ignoring_loop_edges call.
5522 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5523 single_pred_edge_ignoring_loop_edges rather than open coding.
5524 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5526 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5528 * match.pd (-(-A)): Rewrite.
5530 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5532 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5533 (rs6000_emit_int_cmove): New declaration.
5534 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5535 (rs6000_emit_sISEL): Delete.
5536 (rs6000_emit_int_cmove): Make non-static.
5537 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5538 instead of rs6000_emit_sISEL.
5540 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5542 * asan.c (create_cond_insert_point): Maintain profile.
5543 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5545 * basic-block.h (struct basic_block_def): Remove frequency.
5546 (EDGE_FREQUENCY): Use to_frequency
5547 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5549 (find_traces): Update to use to_frequency.
5550 (find_traces_1_round): Likewise; use only IPA counts.
5551 (bb_to_key): Likewise.
5552 (connect_traces): Use IPA counts only.
5553 (copy_bb_p): Update to use to_frequency.
5554 (fix_up_crossing_landing_pad): Likewise.
5555 (sanitize_hot_paths): Likewise.
5556 * bt-load.c (basic_block_freq): Likewise.
5557 * cfg.c (init_flow): Set count_max to uninitialized.
5558 (check_bb_profile): Remove frequencies; check counts.
5559 (dump_bb_info): Do not dump frequencies.
5560 (update_bb_profile_for_threading): Update counts only.
5561 (scale_bbs_frequencies_int): Likewise.
5562 (MAX_SAFE_MULTIPLIER): Remove.
5563 (scale_bbs_frequencies_gcov_type): Update counts only.
5564 (scale_bbs_frequencies_profile_count): Update counts only.
5565 (scale_bbs_frequencies): Update counts only.
5566 * cfg.h (struct control_flow_graph): Add count-max.
5567 (update_bb_profile_for_threading): Update prototype.
5568 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5569 (find_many_sub_basic_blocks): Likewise.
5570 * cfgcleanup.c (try_forward_edges): Likewise.
5571 (try_crossjump_to_edge): Likewise.
5572 * cfgexpand.c (expand_gimple_cond): Likewise.
5573 (expand_gimple_tailcall): Likewise.
5574 (construct_init_block): Likewise.
5575 (construct_exit_block): Likewise.
5576 * cfghooks.c (verify_flow_info): Check consistency of counts.
5577 (dump_bb_for_graph): Do not dump frequencies.
5578 (split_block_1): Do not update frequencies.
5579 (split_edge): Do not update frequencies.
5580 (make_forwarder_block): Do not update frequencies.
5581 (duplicate_block): Do not update frequencies.
5582 (account_profile_record): Do not update frequencies.
5583 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5584 for global heuristics.
5585 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5586 (expected_loop_iterations_unbounded): Use counts only.
5587 * cfgloopmanip.c (scale_loop_profile): Simplify.
5588 (create_empty_loop_on_edge): Simplify
5590 (duplicate_loop_to_header_edge): Simplify
5591 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5592 (update_br_prob_note): Take care of removing note when profile
5594 (relink_block_chain): Do not dump frequency.
5595 (rtl_account_profile_record): Use to_frequency.
5596 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5597 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5598 (cgraph_update_edges_for_call_stmt_node): Likewise.
5599 (cgraph_edge::verify_count_and_frequency): Update.
5600 (cgraph_node::verify_node): Temporarily disable frequency verification.
5601 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5602 to_cgraph_frequency.
5603 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5604 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5606 (cgraph_node::expand_thunk): Update profile.
5607 * except.c (dw2_build_landing_pads): Do not update frequency.
5608 * final.c (compute_alignments): Use to_frequency.
5609 (dump_basic_block_info): Do not dump frequency.
5610 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5611 (dump_gimple_bb_header): Do not dump frequency.
5612 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5614 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5615 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5616 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5617 (init_before_recovery): Do not update frequency.
5618 (sched_create_recovery_edges): Do not update frequency.
5619 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5620 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5621 (ipa_cp_c_finalize): Set max_count to uninitialized.
5622 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5623 (param_change_prob): Use counts.
5624 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5626 * ipa-split.c (consider_split): Use to_frequency.
5627 (split_function): Use to_frequency.
5628 * ira-build.c (loop_compare_func): Likewise.
5629 (mark_loops_for_removal): Likewise.
5630 (mark_all_loops_for_removal): Likewise.
5631 * loop-doloop.c (doloop_modify): Do not update frequency.
5632 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5634 * lto-streamer-in.c (input_function): Update count_max.
5635 * omp-expand.c (expand_omp_taskreg): Update count_max.
5636 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5637 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5638 (maybe_hot_count_p): Use ipa counts only.
5639 (maybe_hot_bb_p): Simplify.
5640 (maybe_hot_edge_p): Simplify.
5641 (probably_never_executed): Do not take frequency argument.
5642 (probably_never_executed_bb_p): Do not pass frequency.
5643 (probably_never_executed_edge_p): Likewise.
5644 (combine_predictions_for_bb): Check that profile is nonzero.
5645 (propagate_freq): Do not set frequency.
5646 (drop_profile): Simplify.
5647 (counts_to_freqs): Simplify.
5648 (expensive_function_p): Use to_frequency.
5649 (propagate_unlikely_bbs_forward): Simplify.
5650 (determine_unlikely_bbs): Simplify.
5651 (estimate_bb_frequencies): Add hack to silence graphite issues.
5652 (compute_function_frequency): Use ipa counts.
5653 (pass_profile::execute): Update.
5654 (rebuild_frequencies): Use counts only.
5655 (force_edge_cold): Use counts only.
5656 * profile-count.c (profile_count::dump): Dump new count types.
5657 (profile_count::differs_from_p): Check compatiblity.
5658 (profile_count::to_frequency): New function.
5659 (profile_count::to_cgraph_frequency): New function.
5660 * profile-count.h (struct function): Declare.
5661 (enum profile_quality): Add profile_guessed_local and
5662 profile_guessed_global0.
5663 (class profile_proability): Decrease number of bits to 29;
5664 update from_reg_br_prob_note and to_reg_br_prob_note.
5665 (class profile_count: Update comment; decrease number of bits
5666 to 61. Check compatibility.
5667 (profile_count::compatible_p): New private member function.
5668 (profile_count::ipa_p): New member function.
5669 (profile_count::operator<): Handle global zero correctly.
5670 (profile_count::operator>): Handle global zero correctly.
5671 (profile_count::operator<=): Handle global zero correctly.
5672 (profile_count::operator>=): Handle global zero correctly.
5673 (profile_count::nonzero_p): New member function.
5674 (profile_count::force_nonzero): New member function.
5675 (profile_count::max): New member function.
5676 (profile_count::apply_scale): Handle IPA scalling.
5677 (profile_count::guessed_local): New member function.
5678 (profile_count::global0): New member function.
5679 (profile_count::ipa): New member function.
5680 (profile_count::to_frequency): Declare.
5681 (profile_count::to_cgraph_frequency): Declare.
5682 * profile.c (OVERLAP_BASE): Delete.
5683 (compute_frequency_overlap): Delete.
5684 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5685 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5686 * sched-ebb.c (rank): Use counts only.
5687 * shrink-wrap.c (handle_simple_exit): Use counts only.
5688 (try_shrink_wrapping): Use counts only.
5689 (place_prologue_for_one_component): Use counts only.
5690 * tracer.c (find_best_predecessor): Use to_frequency.
5691 (find_trace): Use to_frequency.
5692 (tail_duplicate): Use to_frequency.
5693 * trans-mem.c (expand_transaction): Do not update frequency.
5694 * tree-call-cdce.c: Do not update frequency.
5695 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5696 (gimple_merge_blocks): Likewise.
5697 (gimple_split_edge): Likewise.
5698 (gimple_duplicate_sese_region): Likewise.
5699 (gimple_duplicate_sese_tail): Likewise.
5700 (move_sese_region_to_fn): Likewise.
5701 (gimple_account_profile_record): Likewise.
5702 (insert_cond_bb): Likewise.
5703 * tree-complex.c (expand_complex_div_wide): Likewise.
5704 * tree-eh.c (lower_resx): Update profile.
5705 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5707 (initialize_cfun): Do not initialize frequencies
5708 (freqs_to_counts): Delete.
5709 (copy_cfg_body): Ignore count parameter.
5710 (copy_body): Update.
5711 (expand_call_inline): Update count_max.
5712 (optimize_inline_calls): Update count_max.
5713 (tree_function_versioning): Update count_max.
5714 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5715 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5717 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5718 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5719 (try_peel_loop): Likewise.
5720 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5722 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5723 (tree_transform_and_unroll_loop): Do not use frequencies
5724 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5725 Use reliable prediction only.
5726 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5727 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5728 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5729 probability scaling.
5730 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5731 not update frequency
5732 (any_remaining_duplicated_blocks): Likewise.
5733 (update_profile): Likewise.
5734 (estimated_freqs_path): Delete.
5735 (freqs_to_counts_path): Delete.
5736 (clear_counts_path): Delete.
5737 (ssa_fix_duplicate_block_edges): Likewise.
5738 (duplicate_thread_path): Likewise.
5739 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5740 * tree-tailcall.c (decrease_profile): Do not update frequency.
5741 (eliminate_tail_call): Likewise.
5742 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5743 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5744 (optimize_mask_stores): Likewise.
5745 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5746 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5747 (ubsan_expand_ptr_ifn): Update profile.
5748 * value-prof.c (gimple_ic): Simplify.
5749 * value-prof.h (gimple_ic): Update prototype.
5750 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5751 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5753 (want_inline_self_recursive_call_p): Likewise.
5754 (resolve_noninline_speculation): Only cummulate defined counts.
5755 (inline_small_functions): Use nonzero_p.
5756 (ipa_inline): Do not access freed node.
5758 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5760 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5761 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5763 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5765 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5766 non-legitimate address.
5768 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5770 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5771 (*lt0_<mode>di, *lt0_<mode>si): New.
5773 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5775 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5776 TARGET_PAIRED_FLOAT.
5778 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5779 Jim Wilson <jim.wilson@linaro.org>
5781 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5782 * config/aarch64/aarch64-tune.md: Regenerated.
5783 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5784 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5786 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5788 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5789 (arc_expand_prologue): Restore blink for millicode.
5790 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5792 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5795 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5796 gen_vec_duplicate after forcing the scalar into a register.
5798 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5800 * combine (try_combine): Print the insns input to try_combine to the
5803 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5806 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5807 Remove second argument from aarch64_process_target_attr call.
5808 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5810 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5812 (aarch64_handle_attr_arch): Remove second argument.
5813 (aarch64_handle_attr_cpu): Ditto.
5814 (aarch64_handle_attr_tune): Ditto.
5815 (aarch64_handle_attr_isa_flags): Ditto.
5816 (aarch64_process_one_target_attr): Ditto.
5817 (aarch64_process_target_attr): Ditto.
5818 (aarch64_option_valid_attribute_p): Remove second argument.
5819 on aarch64_process_target_attr call.
5821 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5823 * diagnostic.c: Include "selftest-diagnostic.h".
5824 (selftest::assert_location_text): New function.
5825 (selftest::test_diagnostic_get_location_text): New function.
5826 (selftest::diagnostic_c_tests): Call it.
5828 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5830 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5831 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5832 (class selftest::test_diagnostic_context): Move to...
5833 * selftest-diagnostic.c: New file.
5834 * selftest-diagnostic.h: New file.
5836 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5838 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5839 offset range for FT32B.
5840 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5841 * config/ft32/ft32.md: Add TARGET_NOPM.
5842 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5843 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5845 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5847 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5849 2017-11-02 Jeff Law <law@redhat.com>
5851 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5852 virtual keyword on FINAL OVERRIDE members.
5854 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5855 virtuals together. Add virtual destructor.
5856 (substitute_and_fold_engine): Similarly.
5858 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5860 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5862 2017-11-02 Richard Biener <rguenther@suse.de>
5864 PR tree-optimization/82795
5865 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5867 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5869 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5871 (gcc_GAS_CHECK_FEATURE): Remove.
5872 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5873 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5874 * configure: Regenerate.
5876 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5878 * config/arc/arc.c (hwloop_optimize): Account for empty
5881 2017-11-02 Richard Biener <rguenther@suse.de>
5884 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5885 Truncate ARRAY_REF index and element size.
5887 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5889 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5891 2017-11-01 Jeff Law <law@redhat.com>
5893 * tree-ssa-ccp.c (ccp_folder): New class derived from
5894 substitute_and_fold_engine.
5895 (ccp_folder::get_value): New member function.
5896 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5897 (ccp_fold_stmt): Remove prototype.
5898 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5899 * tree-ssa-copy.c (copy_folder): New class derived from
5900 substitute_and_fold_engine.
5901 (copy_folder::get_value): Renamed from get_value.
5902 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5903 * tree-vrp.c (vrp_folder): New class derived from
5904 substitute_and_fold_engine.
5905 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5906 (vrp_folder::get_value): New member function.
5907 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5908 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5909 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5910 provide a class interface to folder/substitute routines.
5911 (ssa_prop_fold_stmt_fn): Remove typedef.
5912 (ssa_prop_get_value_fn): Likewise.
5913 (subsitute_and_fold): Remove prototype.
5914 (replace_uses_in): Likewise.
5915 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5916 Renamed from replace_uses_in. Call the virtual member function
5917 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5918 (substitute_and_fold_dom_walker): Remove initialization of
5919 data member entries for calbacks. Add substitute_and_fold_engine
5920 member and initialize it.
5921 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5922 member functions for get_value, replace_phi_args_in c
5923 replace_uses_in, and fold_stmt calls.
5924 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5925 substitute_and_fold. Remove assert. Update ctor call.
5927 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5928 (ssa_prop_visit_phi_fn): Likewise.
5929 (class ssa_propagation_engine): New class to provide an interface
5931 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5933 (ssa_prop_visit_phi): Likewise.
5934 (ssa_propagation_engine::simulate_stmt): Moved into class.
5935 Call visit_phi/visit_stmt from the class rather than via
5936 file scoped static variables.
5937 (ssa_propagation_engine::simulate_block): Moved into class.
5938 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5939 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5940 set file scoped statics for the visit_stmt/visit_phi callbacks.
5941 * tree-complex.c (complex_propagate): New class derived from
5942 ssa_propagation_engine.
5943 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5944 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5945 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5947 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5948 ssa_propagation_engine.
5949 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5950 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5951 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5952 * tree-ssa-copy.c (copy_prop): New class derived from
5953 ssa_propagation_engine.
5954 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5955 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5956 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5957 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5958 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5959 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5960 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5962 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5964 PR rtl-optimization/82778
5965 PR rtl-optimization/82597
5966 * compare-elim.c (struct comparison): Add in_a_setter field.
5967 (find_comparison_dom_walker::before_dom_children): Remove killed
5968 bitmap and df_simulate_find_defs call, instead walk the defs.
5969 Compute last_setter and initialize in_a_setter. Merge definitions
5970 with first initialization for a few variables.
5971 (try_validate_parallel): Use insn_invalid_p instead of
5972 recog_memoized. Return insn rather than just the pattern.
5973 (try_merge_compare): Fix up comment. Don't uselessly test if
5974 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5976 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5979 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5980 Alan Hayward <alan.hayward@arm.com>
5981 David Sherwood <david.sherwood@arm.com>
5983 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5984 aarch64_hard_regno_nregs to get the number of registers
5987 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5988 Alan Hayward <alan.hayward@arm.com>
5989 David Sherwood <david.sherwood@arm.com>
5991 * config/aarch64/constraints.md (Upl): Rename to...
5993 * config/aarch64/aarch64.md
5994 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5997 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5998 Alan Hayward <alan.hayward@arm.com>
5999 David Sherwood <david.sherwood@arm.com>
6001 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6002 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6005 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6006 Alan Hayward <alan.hayward@arm.com>
6007 David Sherwood <david.sherwood@arm.com>
6009 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6010 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6011 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6013 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6014 of a single element.
6015 * config/aarch64/iterators.md: Add a comment above the permute
6016 unspecs to say that they are generated directly by
6017 aarch64_expand_vec_perm_const.
6018 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6020 2017-11-01 Nathan Sidwell <nathan@acm.org>
6022 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6024 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6026 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6027 and medany code models, and describe what they do.
6029 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6031 Revert accidental duplicate:
6033 * combine.c (can_change_dest_mode): Reject changes in
6034 REGMODE_NATURAL_SIZE.
6036 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6038 PR rtl-optimization/64682
6039 PR rtl-optimization/69567
6040 PR rtl-optimization/69737
6041 PR rtl-optimization/82683
6042 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6043 register mentioned in the note, drop the note, unless it came from I3,
6044 in which case it should go to I3 again.
6046 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6048 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6049 and return false if not.
6050 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6052 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6054 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6055 range is known to be empty.
6057 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6058 Alan Hayward <alan.hayward@arm.com>
6059 David Sherwood <david.sherwood@arm.com>
6061 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6062 and CONST_VECTOR_NUNITS instead of computing the number of units from
6063 the byte sizes of the vector and element.
6064 (simplify_binary_operation_1): Likewise.
6065 (simplify_const_binary_operation): Likewise.
6066 (simplify_ternary_operation): Likewise.
6068 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6069 Alan Hayward <alan.hayward@arm.com>
6070 David Sherwood <david.sherwood@arm.com>
6072 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6073 (int_mem_offset): ...this new function.
6074 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6075 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6076 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6079 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6080 Alan Hayward <alan.hayward@arm.com>
6081 David Sherwood <david.sherwood@arm.com>
6083 * lower-subreg.c (interesting_mode_p): New function.
6084 (compute_costs, find_decomposable_subregs, decompose_register)
6085 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6086 (resolve_clobber, dump_choices): Use it.
6088 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6089 Alan Hayward <alan.hayward@arm.com>
6090 David Sherwood <david.sherwood@arm.com>
6092 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6094 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6095 Alan Hayward <alan.hayward@arm.com>
6096 David Sherwood <david.sherwood@arm.com>
6098 * alias.c (find_base_value, find_base_term): Only process integer
6099 truncations. Check the precision rather than the size.
6101 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6102 Alan Hayward <alan.hayward@arm.com>
6103 David Sherwood <david.sherwood@arm.com>
6105 * machmode.h (is_narrower_int_mode): New function
6106 * optabs.c (expand_float, expand_fix): Use it.
6107 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6109 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6110 Alan Hayward <alan.hayward@arm.com>
6111 David Sherwood <david.sherwood@arm.com>
6113 * rtl.h (narrower_subreg_mode): New function.
6114 * ira-color.c (update_costs_from_allocno): Use it.
6116 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6117 Alan Hayward <alan.hayward@arm.com>
6118 David Sherwood <david.sherwood@arm.com>
6120 * optabs-query.h (convert_optab_p): New function, split out from...
6121 (convert_optab_handler): ...here.
6122 (widening_optab_handler): Delete.
6123 (find_widening_optab_handler): Remove permit_non_widening parameter.
6124 (find_widening_optab_handler_and_mode): Likewise. Provide an
6125 override that operates on mode class wrappers.
6126 * optabs-query.c (widening_optab_handler): Delete.
6127 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6128 parameter. Assert that the two modes are the same class and that
6129 the "from" mode is narrower than the "to" mode. Use
6130 convert_optab_handler instead of widening_optab_handler.
6131 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6132 instead of widening_optab_handler.
6133 * expr.c (expand_expr_real_2): Update calls to
6134 find_widening_optab_handler.
6135 * optabs.c (expand_widen_pattern_expr): Likewise.
6136 (expand_binop_directly): Take the insn_code as a parameter.
6137 (expand_binop): Only call find_widening_optab_handler for
6138 conversion optabs; use optab_handler otherwise. Update calls
6139 to find_widening_optab_handler and expand_binop_directly.
6140 Use convert_optab_handler instead of widening_optab_handler.
6141 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6142 find_widening_optab_handler and use scalar_mode rather than
6144 (convert_plusminus_to_widen): Likewise.
6146 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6147 Alan Hayward <alan.hayward@arm.com>
6148 David Sherwood <david.sherwood@arm.com>
6150 * machmode.h (fixed_size_mode): New class.
6151 * rtl.h (get_pool_mode): Return fixed_size_mode.
6152 * gengtype.c (main): Add fixed_size_mode.
6153 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6154 (get_raw_arg_mode): Likewise.
6155 * doc/tm.texi: Regenerate.
6156 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6157 * targhooks.c (default_get_reg_raw_mode): Likewise.
6158 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6159 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6160 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6161 (msp430_get_raw_result_mode): Likewise.
6162 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6163 * dbxout.c (dbxout_parms): Require fixed-size modes.
6164 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6165 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6166 * omp-low.c (lower_oacc_reductions): Likewise.
6167 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6168 (simplify_subreg): Update accordingly.
6169 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6170 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6171 that aren't fixed-size.
6172 (get_pool_mode): Return a fixed_size_mode.
6173 (output_constant_pool_2): Take a fixed_size_mode.
6175 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6176 Alan Hayward <alan.hayward@arm.com>
6177 David Sherwood <david.sherwood@arm.com>
6179 * doc/rtl.texi (vec_series): Document.
6180 (const): Say that the operand can be a vec_series.
6181 * rtl.def (VEC_SERIES): New rtx code.
6182 * rtl.h (const_vec_series_p_1): Declare.
6183 (const_vec_series_p): New function.
6184 * emit-rtl.h (gen_const_vec_series): Declare.
6185 (gen_vec_series): Likewise.
6186 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6187 (gen_vec_series): Likewise.
6188 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6189 * simplify-rtx.c (simplify_unary_operation): Handle negations
6191 (simplify_binary_operation_series): New function.
6192 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6193 (test_vector_ops_series): New function.
6194 (test_vector_ops): Call it.
6195 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6196 gen_const_vec_series.
6197 (altivec_lvsr): Likewise.
6198 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6200 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6201 Alan Hayward <alan.hayward@arm.com>
6202 David Sherwood <david.sherwood@arm.com>
6204 * doc/rtl.texi (const): Update description of address constants.
6205 Say that vector constants are allowed too.
6206 * common.md (E, F): Use CONSTANT_P instead of checking for
6208 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6209 checking for CONST_VECTOR.
6210 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6212 * expr.c (expand_expr_real_2): Check for vector modes instead
6213 of checking for CONST_VECTOR.
6214 * rtl.h (const_vec_p): New function.
6215 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6216 (unwrap_const_vec_duplicate): Handle them here too.
6218 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6219 David Malcolm <dmalcolm@redhat.com>
6220 Alan Hayward <alan.hayward@arm.com>
6221 David Sherwood <david.sherwood@arm.com>
6223 * rtl.h (vec_duplicate_p): New function.
6224 * selftest-rtl.c (assert_rtx_eq_at): New function.
6225 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6226 (assert_rtx_eq_at): Declare.
6227 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6228 * selftest-run-tests.c (selftest::run_tests): Call it.
6229 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6230 (simplify_unary_operation_1): Recursively handle vector duplicates.
6231 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6233 (simplify_subreg): Handle subregs of vector duplicates.
6234 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6235 (selftest::simplify_rtx_c_tests): New functions.
6237 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6238 Alan Hayward <alan.hayward@arm.com>
6239 David Sherwood <david.sherwood@arm.com>
6241 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6242 (gen_vec_duplicate): Likewise.
6243 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6245 (gen_const_vector): ...here.
6246 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6247 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6248 whose elements are all equal.
6249 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6250 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6251 (simplify_relational_operation): Likewise.
6252 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6254 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6255 (aarch64_expand_vector_init): Likewise.
6256 * config/arm/arm.c (neon_vdup_constant): Likewise.
6257 (neon_expand_vector_init): Likewise.
6258 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6259 (arm_block_set_unaligned_vect): Likewise.
6260 (arm_block_set_aligned_vect): Likewise.
6261 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6262 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6263 (expand_vec_perm_even_odd_pack): Likewise.
6264 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6265 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6266 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6267 gen_const_vec_duplicate.
6268 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6269 * config/mips/mips.c (mips_gen_const_int_vector): Use
6270 gen_const_vec_duplicate.
6271 (mips_expand_vector_init): Use CONST0_RTX.
6272 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6273 (define_split): Use gen_const_vec_duplicate.
6274 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6275 (define_split): Use gen_const_vec_duplicate.
6276 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6277 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6278 * config/spu/spu.c (spu_const): Likewise.
6280 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6281 Alan Hayward <alan.hayward@arm.com>
6282 David Sherwood <david.sherwood@arm.com>
6284 * combine.c (can_change_dest_mode): Reject changes in
6285 REGMODE_NATURAL_SIZE.
6287 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6289 * configure.ac (--enable-libssp): New.
6290 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6291 trying to determine target-specific default. Adjust indentation.
6292 * configure: Regenerated.
6293 * doc/install.texi (Configuration): Expand --disable-libssp
6296 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6298 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6301 2017-10-31 Martin Jambor <mjambor@suse.cz>
6304 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6306 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6308 * auto-profile.c (autofdo_source_profile::read): Use
6309 UNKNOWN_LOCATION rather than 0.
6310 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6311 (warning_at): ...this overload.
6312 (warning_at_rich_loc_n): Rename to...
6313 (warning_n): ...this overload.
6314 (error_at_rich_loc): Rename to...
6315 (error_at): ...this overload.
6316 (pedwarn_at_rich_loc): Rename to...
6317 (pedwarn): ...this overload.
6318 (permerror_at_rich_loc): Rename to...
6319 (permerror): ...this overload.
6320 (inform_at_rich_loc): Rename to...
6321 (inform): ...this overload.
6322 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6323 (diagnostic_n_impl_richloc): Rename to...
6324 (diagnostic_n_impl): ...this rich_location *-based decl.
6325 (inform_at_rich_loc): Rename to...
6326 (inform): ...this, and add an assertion.
6327 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6328 (warning_at_rich_loc): Rename to...
6329 (warning_at): ...this, and add an assertion.
6330 (warning_at_rich_loc_n): Rename to...
6331 (warning_n): ...this, and add an assertion.
6332 (warning_n): Update location_t-based implementation for removal of
6333 location_t-based diagnostic_n_impl.
6334 (pedwarn_at_rich_loc): Rename to...
6335 (pedwarn): ...this, and add an assertion.
6336 (permerror_at_rich_loc): Rename to...
6337 (permerror): ...this, and add an assertion.
6338 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6339 (error_at_rich_loc): Rename to...
6340 (error_at): ...this, and add an assertion.
6341 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6342 (driver::do_spec_on_infiles): Likewise.
6343 * substring-locations.c (format_warning_va): Update for renaming
6344 of inform_at_rich_loc.
6346 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6348 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6349 _Float<N>X built-in functions so that the variant without the
6350 "__builtin_" prefix is only enabled for the GNU C and Objective C
6351 languages when they are in non-strict ANSI/ISO mode.
6352 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6353 * target.def (floatn_builtin_p): Add a target hook to control
6354 whether _Float<N> and _Float<N>X built-in functions without the
6355 "__builtin_" prefix are enabled, and return true for C and
6356 Objective C in the default hook. Include langhooks.h in
6358 * targhooks.h (default_floatn_builtin_p): Likewise.
6359 * targhooks.c (default_floatn_builtin_p): Likewise.
6360 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6361 floatn_builtin_p target hook.
6362 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6364 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6365 Eric Botcazou <ebotcazou@adacore.com>
6367 PR rtl-optimization/81803
6368 * lra-constraints.c (curr_insn_transform): Also reload the whole
6369 register for a strict subreg no wider than a word if this is for
6370 a WORD_REGISTER_OPERATIONS target.
6372 2017-10-31 Jason Merrill <jason@redhat.com>
6374 * gdbinit.in: Skip over inlines from timevar.h.
6376 2017-10-31 Martin Liska <mliska@suse.cz>
6378 * doc/gcov.texi: Document new option.
6379 * gcov.c (print_usage): Likewise print it.
6380 (process_args): Support the argument.
6381 (format_count): New function.
6382 (format_gcov): Use the function.
6384 2017-10-31 Martin Liska <mliska@suse.cz>
6386 * gcov.c (struct name_map): do not use typedef.
6387 Define operator== and operator<.
6388 (name_search): Remove.
6389 (name_sort): Remove.
6390 (main): Do not allocate names.
6391 (process_file): Add vertical space.
6392 (generate_results): Use std::find.
6393 (release_structures): Do not release memory.
6394 (find_source): Use std::find.
6396 2017-10-31 Martin Liska <mliska@suse.cz>
6398 * gcov.c (struct line_info): Remove it's typedef.
6399 (line_info::line_info): Add proper ctor.
6400 (line_info::has_block): Do not use a typedef.
6401 (struct source_info): Do not use typedef.
6402 (circuit): Likewise.
6403 (get_cycles_count): Likewise.
6404 (output_intermediate_file): Iterate via vector iterator.
6405 (add_line_counts): Use std::vector methods.
6406 (accumulate_line_counts): Likewise.
6407 (output_lines): Likewise.
6409 2017-10-31 Martin Liska <mliska@suse.cz>
6411 * gcov.c (struct source_info): Remove typedef.
6412 (source_info::source_info): Add proper ctor.
6413 (accumulate_line_counts): Use struct, not it's typedef.
6414 (output_gcov_file): Likewise.
6415 (output_lines): Likewise.
6416 (main): Do not allocate an array.
6417 (output_intermediate_file): Use size of vector container.
6418 (process_file): Resize the vector.
6419 (generate_results): Do not preallocate, use newly added vector
6421 (release_structures): Do not release sources.
6422 (find_source): Use vector methods.
6423 (add_line_counts): Do not use typedef.
6425 2017-10-31 Martin Liska <mliska@suse.cz>
6427 * doc/gcov.texi: Document that.
6428 * gcov.c (add_line_counts): Mark lines with a non-executed
6430 (output_line_beginning): Handle such lines.
6431 (output_lines): Pass new argument.
6432 (output_intermediate_file): Print it in intermediate format.
6434 2017-10-31 Martin Liska <mliska@suse.cz>
6436 * color-macros.h: New file.
6437 * diagnostic-color.c: Factor out color related to macros to
6439 * doc/gcov.texi: Document -k option.
6440 * gcov.c (INCLUDE_STRING): Include string.h.
6441 (print_usage): Add -k option.
6442 (process_args): Parse it.
6443 (pad_count_string): New function.
6444 (output_line_beginning): Likewise.
6445 (DEFAULT_LINE_START): New macro.
6446 (output_lines): Support color output.
6448 2017-10-31 Martin Liska <mliska@suse.cz>
6450 PR gcov-profile/82633
6451 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6452 their interaction with GCOV infrastructure.
6453 * configure.ac: Add -fkeep-{inline,static}-functions to
6455 * configure: Regenerate.
6457 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6460 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6462 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6465 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6466 into a register if it does not fit into an immediate offset field.
6468 2017-10-31 Olivier Hainque <hainque@adacore.com>
6470 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6472 2017-10-31 Julia Koval <julia.koval@intel.com>
6474 * config.gcc: Add gfniintrin.h.
6475 * config/i386/gfniintrin.h: New.
6476 * config/i386/i386-builtin-types.def
6477 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6478 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6479 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6480 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6481 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6482 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6483 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6484 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6485 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6486 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6487 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6488 * config/i386/immintrin.h: Include gfniintrin.h.
6489 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6491 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6493 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6495 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6497 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6498 (arm_ashldi3_1bit): Remove pattern.
6499 (ashrdi3): Remove shift by 1 expansion.
6500 (arm_ashrdi3_1bit): Remove pattern.
6501 (lshrdi3): Remove shift by 1 expansion.
6502 (arm_lshrdi3_1bit): Remove pattern.
6503 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6504 cost of ashldi3 by 1.
6505 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6506 (<shift>di3_neon): Likewise.
6508 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6510 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6511 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6512 and (*aarch64_simd_mov<VQ:mode>).
6513 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6514 pattern alternative.
6515 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6516 attributes to match pattern alternative.
6518 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6520 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6521 * config/rs6000/emmintrin.h: New file.
6522 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6524 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6526 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6527 * (movdi_vfp_cortexa8): Remove pattern.
6529 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6531 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6533 (Specific, alpha*-dec-osf5.1): Remove.
6534 (Specific, mips-sgi-irix5): Remove.
6535 (Specific, mips-sgi-irix6): Remove.
6537 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6540 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6541 arguments to clear_bit_region_be.
6543 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6545 * gimplify.c: Include memmodel.h.
6547 2017-10-30 Martin Jambor <mjambor@suse.cz>
6549 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6550 condition whether loop should be executed at all.
6552 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6554 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6555 gimple folding of vec_madd() intrinsics.
6556 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6557 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6558 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6560 2017-10-30 Richard Biener <rguenther@suse.de>
6562 PR tree-optimization/82762
6564 2017-10-23 Richard Biener <rguenther@suse.de>
6566 PR tree-optimization/82129
6568 2017-08-01 Richard Biener <rguenther@suse.de>
6570 PR tree-optimization/81181
6571 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6572 (compute_antic): ... end of iteration here.
6574 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6576 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6577 -std=iso9899:2017 and -std=gnu17.
6578 * doc/standards.texi (C Language): Document C17 support.
6579 * doc/cpp.texi (Overview): Mention -std=c17.
6580 (Standard Predefined Macros): Document C11 and C17 values of
6581 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6582 * doc/extend.texi (Inline): Do not list individual options for
6583 standards newer than C99.
6584 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6586 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6589 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6591 * asan.c (asan_finish_file): Align asan globals array by shadow
6594 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6597 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6598 (struct store_immediate_info): Add bitregion_start and bitregion_end
6600 (store_immediate_info::store_immediate_info): Add brs and bre
6601 arguments and initialize bitregion_{start,end} from those.
6602 (struct merged_store_group): Add bitregion_start, bitregion_end,
6603 align_base and mask fields. Drop unnecessary struct keyword from
6604 struct store_immediate_info. Add do_merge method.
6605 (clear_bit_region_be): Use memset instead of loop storing zeros.
6606 (merged_store_group::do_merge): New method.
6607 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6608 stores as long as the surrounding bitregions have no gaps.
6609 (merged_store_group::merge_overlapping): Use do_merge.
6610 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6611 is byte aligned, rather than requiring that start and width are
6612 byte aligned. Drop unnecessary struct keyword from
6613 struct store_immediate_info. Allocate and populate also mask array.
6614 Make start of the arrays relative to bitregion_start rather than
6615 start and size them according to bitregion_{end,start} difference.
6616 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6617 struct store_immediate_info.
6618 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6619 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6620 struct keyword from struct store_immediate_info.
6621 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6622 between stores as long as the surrounding bitregions have no gaps.
6624 (struct split_store): Add orig non-static data member.
6625 (split_store::split_store): Initialize orig to false.
6626 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6627 if there is exactly a single original stmt. Change stmts argument
6628 to pointer from reference, if NULL, don't push anything to it. Add
6629 first argument, use it to optimize skipping over orig stmts that
6630 are known to be before bitpos already. Simplify.
6631 (split_group): Return unsigned int count how many stores are or
6632 would be needed rather than a bool. Add allow_unaligned argument.
6633 Change split_stores argument from reference to pointer, if NULL,
6634 only do a dry run computing how many stores would be produced.
6635 Rewritten algorithm to use both alignment and misalign if
6636 !allow_unaligned and handle bitfield stores with gaps.
6637 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6638 from bitregion_start instead of start. Compute allow_unaligned
6639 here, if true, do 2 split_group dry runs to compute which one
6640 produces fewer stores and prefer aligned if equal. Punt if
6641 new count is bigger or equal than original before emitting any
6642 statements, rather than during that. Remove no longer needed
6643 new_ssa_names tracking. Replace num_stmts with
6644 split_stores.length (). Use 32-bit stack allocated entries
6645 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6646 if possible. Handle bitfields with gaps.
6647 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6648 Compute bitregion_{start,end} for the stores and construct
6649 store_immediate_info with that. Formatting fixes.
6651 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6654 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6655 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6657 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6659 * gimplify.c: Include tm_p.h.
6661 * common.opt (gcoff): Re-add as ignored option.
6662 (gcoff1, gcoff2, gcoff3): Likewise.
6664 * Makefile.in (OBJS): Delete sdbout.o.
6665 (GTFILES): Delete $(srcdir)/sdbout.c.
6666 * debug.h: Delete sdb_debug_hooks.
6667 * final.c: Delete sdbout.h include.
6668 (final_scan_insn): Delete SDB_DEBUG check.
6669 (rest_of_clean_state): Likewise.
6670 * output.h: Delete sdb_begin_function_line.
6673 * toplev.c: Delete sdbout.h include.
6674 (process_options): Delete SDB_DEBUG check.
6675 * tree-core.h (tree_type_common): Delete pointer field of
6677 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6678 TYPE_SYMTAB_POINTER.
6679 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6680 (TYPE_SYMTAB_IS_POINTER): Delete.
6681 (TYPE_SYMTAB_IS_DIE): Renumber.
6682 * xcoffout.c: Refer to former sdbout.c file.
6683 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6685 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6686 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6687 (Debugging Options): Delete -gcoff.
6688 (-gstabs): Delete SDB reference.
6690 (-gcoff@var{level}): Delete.
6691 * doc/passes.texi (Debugging information output): Delete SDB and
6692 sdbout.c references.
6693 * doc/tm.texi: Regenerate.
6694 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6695 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6697 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6698 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6699 (SDB_DEBUGGING_INFO): Delete.
6700 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6701 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6702 * target.def (output_source_filename): Delete COFF reference.
6704 * common.opt (gcoff): Delete.
6705 (gxcoff+): Update Negative chain.
6706 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6708 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6709 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6710 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6712 (expand_function_start): Change sdb reference to past tense.
6713 (expand_function_end): Change sdb reference to past tense.
6714 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6715 * opts.c (debug_type_names): Delete coff entry.
6716 (common_handle_option): Delete OPT_gcoff case.
6717 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6719 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6720 * config/cris/cris.h: Delete SDB reference in comment.
6721 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6722 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6723 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6724 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6726 (ix86_expand_prologue): Likewise.
6727 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6728 * config/ia64/ia64.h: Likewise.
6729 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6730 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6732 * config/mmix/mmix.h: Likewise.
6733 * config/nds32/nds32.c: Likewise.
6734 * config/stormy/storym16.h: Likewise.
6735 * config/visium/visium.h: Likewise.
6736 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6738 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6740 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6741 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6742 FRAME_POINTER_REGNUM point at high end of local var area.
6744 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6746 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6747 Move comment around. Do not reset best_edge for a copiable
6748 destination if the copy would cause a partition change.
6749 (better_edge_p): Remove redundant check.
6751 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6753 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6755 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6757 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6758 for math functions that have _Float<N> and _Float<N>X variants.
6759 (mathfn_built_in_2): Add support for math functions that have
6760 _Float<N> and _Float<N>X variants.
6761 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6762 (expand_builtin_mathfn_ternary): Add support for fma with
6763 _Float<N> and _Float<N>X variants.
6764 (expand_builtin): Likewise.
6765 (fold_builtin_3): Likewise.
6766 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6767 create math function _Float<N> and _Float<N>X variants as external
6769 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6770 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6771 the __builtin_ prefix and if not strict ansi, without the prefix.
6772 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6773 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6774 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6775 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6776 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6777 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6778 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6779 function signatures for fma _Float<N> and _Float<N>X variants.
6780 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6781 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6782 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6783 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6784 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6785 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6786 * gencfn-macros.c (print_case_cfn): Add support for math functions
6787 that have _Float<N> and _Float<N>X variants.
6788 (print_define_operator_list): Likewise.
6789 (fltfn_suffixes): Likewise.
6791 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6792 for math functions that have _Float<N> and _Float<N>X variants.
6793 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6794 and _Float<N>X variants.
6795 (COPYSIGN): Likewise.
6798 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6799 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6801 (integer_valued_read_call_p): Likewise.
6802 * fold-const-call.c (fold_const_call_ss): Likewise.
6803 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6804 _Float<N> and _Float<N>X variants.
6805 (fold_const_call_ssss): Add support for fma _Float<N> and
6806 _Float<N>X variants.
6807 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6808 support for copysign and fma _Float<N> and _Float<N>X variants.
6809 (backprop::process_builtin_call_use): Likewise.
6810 * tree-call-cdce.c (can_test_argument_range); Add support for
6811 sqrt _Float<N> and _Float<N>X variants.
6812 (edom_only_function): Likewise.
6813 (get_no_error_domain): Likewise.
6814 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6815 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6816 copysign _Float<N> and _Float<N>X variants.
6817 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6818 handled by machine independent code.
6819 (FMAF128): Likewise.
6820 * doc/cpp.texi (Common Predefined Macros): Document defining
6821 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6822 fma _Float<N> and _Float<N>X variants.
6824 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6827 * config/i386/i386-modes.def (CCFPU): Remove definition.
6828 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6829 (ix86_cc_modes_compatible): Ditto.
6830 (ix86_expand_carry_flag_compare): Ditto.
6831 (ix86_expand_int_movcc): Ditto.
6832 (ix86_expand_int_addcc): Ditto.
6833 (ix86_reverse_condition): Ditto.
6834 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6835 Return true/false for unordered/ordered fp comparisons.
6836 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6837 (ix86_prepare_fp_compare_args): Update for rename.
6838 (ix86_expand_fp_compare): Update for rename. Generate unordered
6839 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6840 (ix86_expand_sse_compare_and_jump): Ditto.
6841 * config/i386/predicates.md (fcmov_comparison_operator):
6842 Remove CCFPU mode handling.
6843 (ix86_comparison_operator): Ditto.
6844 (ix86_carry_flag_operator): Ditto.
6845 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6846 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6847 (*cmpu<mode>_cc_i387): Ditto.
6848 (FPCMP): Remove mode iterator.
6849 (unord): Remove mode attribute.
6850 (unord_subst): New define_subst transformation
6851 (unord): New define_subst attribute.
6853 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6854 (*cmpi<unord>xf_i387): Ditto.
6855 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6856 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6857 using unord_subst transformation.
6858 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6859 (round_saeonly): Also handle CCFP mode.
6860 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6861 Remove UNSPEC_SAHF unspec handling.
6863 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6865 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6867 2017-10-27 Jeff Law <law@redhat.com>
6869 * gimple-ssa-sprintf.c: Include domwalk.h.
6870 (class sprintf_dom_walker): New class, derived from dom_walker.
6871 (sprintf_dom_walker::before_dom_children): New function.
6872 (struct call_info): Moved into sprintf_dom_walker class
6873 (compute_formath_length, handle_gimple_call): Likewise.
6874 (sprintf_length::execute): Call the dominator walker rather
6875 than walking the statements.
6877 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6878 gimple statement locations.
6879 (check_array_bounds): Corresponding changes. Get the statement's
6880 location directly from wi->stmt.
6882 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6885 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6887 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6889 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6890 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6892 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6895 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6896 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6897 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6898 maybe_get_pool_constant.
6899 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6902 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6904 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6905 2.26 caveat. Update gas and gld versions.
6906 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6909 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6911 * cgraph.h (set_malloc_flag): Declare.
6912 * cgraph.c (set_malloc_flag_1): New function.
6913 (set_malloc_flag): Likewise.
6914 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6915 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6917 (read_ipa_call_summary): Add support for reading is_return_callee.
6918 (write_ipa_call_summary): Stream is_return_callee.
6919 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6920 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6921 ipa-prop.h, ipa-fnsummary.h.
6922 (pure_const_names): Change to static.
6923 (malloc_state_e): Define.
6924 (malloc_state_names): Define.
6925 (funct_state_d): Add field malloc_state.
6926 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6927 (check_retval_uses): New function.
6928 (malloc_candidate_p): Likewise.
6929 (analyze_function): Add support for malloc attribute.
6930 (pure_const_write_summary): Stream malloc_state.
6931 (pure_const_read_summary): Add support for reading malloc_state.
6932 (dump_malloc_lattice): New function.
6933 (propagate_malloc): New function.
6934 (warn_function_malloc): New function.
6935 (ipa_pure_const::execute): Call propagate_malloc and
6936 ipa_free_fn_summary.
6937 (pass_local_pure_const::execute): Add support for malloc attribute.
6938 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6939 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6941 2017-10-27 Martin Liska <mliska@suse.cz>
6943 PR gcov-profile/82457
6944 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6945 for fork-like functions to be properly instrumented.
6947 2017-10-27 Richard Biener <rguenther@suse.de>
6950 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6951 info when we redirected EH.
6953 2017-10-26 Michael Collison <michael.collison@arm.com>
6955 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6957 (<optab>_trunchf<GPI:mode>2: New pattern.
6958 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6959 * config/aarch64/iterators.md (wv): New mode attribute.
6960 (vf, VF): New mode attributes.
6961 (vgp, VGP): New mode attributes.
6962 (s): Update attribute with SImode and DImode prefixes.
6964 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6966 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6967 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6968 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6969 (nios2_option_overide): Initialize it. Don't allow R0-relative
6970 addressing with PIC.
6971 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6972 (nios2_symbolic_constant_p): Likewise.
6973 (nios2_legitimate_address_p): Likewise.
6974 (nios2_r0rel_section_name_p): New.
6975 (nios2_symbol_ref_in_r0rel_data_p): New.
6976 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6977 (r0rel_constant_p): New.
6978 (nios2_print_operand_address): Handle r0rel_constant_p.
6979 (nios2_cdx_narrow_form_p): Likewise.
6980 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6981 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6982 (Nios II Options): Document -mr0rel-sec.
6984 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6986 * config/nios2/nios2.c: Include xregex.h.
6987 (nios2_gprel_sec_regex): New.
6988 (nios2_option_overide): Initialize it. Don't allow GP-relative
6989 addressing with PIC.
6990 (nios2_small_section_name_p): Check for regex match.
6991 * config/nios2/nios2.opt (mgprel-sec=): New option.
6992 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6993 (Nios II Options): Document -mgprel-sec.
6995 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6997 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6999 2017-10-26 Tom de Vries <tom@codesourcery.com>
7001 PR tree-optimization/82707
7002 * gimple.c (gimple_copy): Fix unsharing of
7003 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7005 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7007 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7008 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7009 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7010 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7011 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7012 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7013 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7014 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7015 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7016 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7017 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7018 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7019 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7020 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7021 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7022 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7023 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7025 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7027 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7029 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7030 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7031 warning to rs6000.c. Remove the Undocumented flag, since it has
7033 (-mabi=ibmlongdouble): Likewise.
7034 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7035 already set, set the default format for long double.
7036 (rs6000_debug_reg_global): Print whether long double is IBM or
7038 (rs6000_option_override_internal): Rework setting long double
7039 format. Only warn if the user is changing the long double default
7040 and they did not use -Wno-psabi.
7041 * doc/invoke.texi (PowerPC options): Update the documentation for
7042 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7044 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7045 Alan Hayward <alan.hayward@arm.com>
7046 David Sherwood <david.sherwood@arm.com>
7048 * rtl.h (wider_subreg_mode): New function.
7049 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7050 rather than an unsigned int *.
7051 * ira-color.c (regno_max_ref_width): Replace with...
7052 (regno_max_ref_mode): ...this new variable.
7053 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7054 Use wider_subreg_mode.
7055 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7056 rather than an unsigned int *.
7057 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7058 (process_alt_operands): Likewise.
7059 (invariant_p): Likewise.
7060 * lra-spills.c (assign_mem_slot): Likewise.
7061 (add_pseudo_to_slot): Likewise.
7062 * lra.c (collect_non_operand_hard_regs): Likewise.
7063 (add_regs_to_insn_regno_info): Likewise.
7064 * reload1.c (regno_max_ref_width): Replace with...
7065 (regno_max_ref_mode): ...this new variable.
7066 (reload): Update accordingly. Update call to
7067 ira_sort_regnos_for_alter_reg.
7068 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7069 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7070 (scan_paradoxical_subregs): Likewise.
7072 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7074 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7075 (aarch64_frame): Add emit_frame_chain boolean.
7076 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7077 Move eh_return case to aarch64_layout_frame.
7078 (aarch64_layout_frame): Initialize emit_frame_chain.
7079 (aarch64_expand_prologue): Use emit_frame_chain.
7081 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7083 * config/aarch64/aarch64.c (aarch64_layout_frame):
7084 Ensure LR is always stored at the bottom of the callee-saves.
7085 Remove rarely used frame layout which saves callee-saves at top of
7086 frame, so the store of LR can be used as a valid probe in all cases.
7088 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7090 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7091 Improve unaligned TImode/TFmode base/offset split.
7093 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7094 Alan Hayward <alan.hayward@arm.com>
7095 David Sherwood <david.sherwood@arm.com>
7097 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7098 * combine.c (find_single_use_1): Likewise.
7099 (expand_field_assignment): Likewise.
7100 (move_deaths): Likewise.
7101 * lra-constraints.c (simplify_operand_subreg): Likewise.
7102 (curr_insn_transform): Likewise.
7103 * lra.c (collect_non_operand_hard_regs): Likewise.
7104 (add_regs_to_insn_regno_info): Likewise.
7105 * rtlanal.c (reg_referenced_p): Likewise.
7106 (covers_regno_no_parallel_p): Likewise.
7108 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7110 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7111 Don't print any bits outside the precision of the value.
7112 * wide-int.cc (test_printing): Add some new tests.
7114 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7116 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7117 supports -xbrace_comment option.
7118 * configure: Regenerate.
7119 * config.in: Regenerate.
7120 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7121 (ASM_CPU_SPEC): Use it.
7123 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7125 * target.def (static_rtx_alignment): New hook.
7126 * targhooks.h (default_static_rtx_alignment): Declare.
7127 * targhooks.c (default_static_rtx_alignment): New function.
7128 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7129 * doc/tm.texi: Regenerate.
7130 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7131 instead of targetm.constant_alignment. Remove call to
7133 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7134 (cris_preferred_mininum_alignment): New function, split out from...
7135 (cris_constant_alignment): ...here.
7136 (cris_static_rtx_alignment): New function.
7137 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7139 (ix86_constant_alignment): ...here.
7140 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7141 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7142 (mmix_static_rtx_alignment): New function.
7143 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7144 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7146 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7149 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7150 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7152 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7154 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7155 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7156 mark as a sign-extending load.
7157 (local_pic_load_u): Define.
7159 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7162 * fold-const.c (operand_equal_for_comparison_p): Also return true
7163 if ARG0 is a simple variant of ARG1 with narrower precision.
7164 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7166 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7168 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7170 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7171 scatter_static, scatter_per_elt.
7172 * x86-tune-costs.h: Add new cost entries.
7174 2017-10-25 Richard Biener <rguenther@suse.de>
7176 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7177 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7178 class pass_fre): Move to ...
7179 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7180 class pass_fre): ... here and adjust for statistics.
7182 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7185 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7186 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7187 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7188 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7191 2017-10-25 Richard Biener <rguenther@suse.de>
7193 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7194 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7195 eliminate_push_avail, eliminate_insert): Move inside...
7196 (class eliminate_dom_walker): ... this class in preparation
7198 (fini_eliminate): Remove by merging with ...
7199 (eliminate): ... this function. Adjust for class changes.
7200 (pass_pre::execute): Remove fini_eliminate call.
7201 (pass_fre::execute): Likewise.
7203 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7206 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7207 (VPERMI2, VPERMI2I): New mode iterators.
7208 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7209 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7211 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7212 mode iterator. Remove 3 old define_insn patterns.
7213 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7214 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7215 VPERMI2 mode iterator, remove the other two expanders.
7216 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7217 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7218 instructions, remove the other two patterns.
7219 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7220 mode iterator, remove the other two patterns.
7221 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7222 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7223 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7224 and adjust argument order accordingly.
7225 (ix86_expand_vec_perm): Adjust caller.
7226 (expand_vec_perm_1): Likewise.
7227 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7228 (expand_vec_perm_vpermt2_vpshub2): ... this.
7229 (ix86_expand_vec_perm_const_1): Adjust caller.
7230 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7233 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7234 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7235 (vec_shl_<mode>): Remove unused expander.
7236 (avx512bw_<shift_insn><mode>3): New define_insn.
7237 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7238 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7240 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7243 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7246 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7248 PR rtl-optimization/82396
7249 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7250 (autopref_multipass_init): Simplify initialization.
7251 (autopref_rank_data): Simplify sort order.
7252 * gcc/sched-int.h (autopref_multipass_data_): Remove
7253 multi_mem_insn_p, min_offset and max_offset.
7255 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7258 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7259 Check special value of flag_omit_frame_pointer.
7260 (aarch64_can_eliminate): Likewise.
7261 (aarch64_override_options_after_change_1): Simplify handling of
7262 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7264 2017-10-24 Richard Biener <rguenther@suse.de>
7266 PR tree-optimization/82697
7267 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7268 zero for conditional load and unconditional store.
7270 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7272 * doc/install.texi: Document bootstrap-cet.
7274 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7277 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7278 ENDBR instruction at function entrance if function is only
7281 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7284 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7285 patterns to better describe from which operation the CF is computed.
7286 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7287 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7288 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7289 is 0, use _0 suffixed expanders instead of emitting a comparison
7292 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7294 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7295 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7297 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7300 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7301 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7302 * loop-iv.c (iv_get_reaching_def): Likewise.
7303 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7304 variable is promoted and the partition contains undefined values.
7306 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7308 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7310 (nios2_address_cost): Define.
7311 (nios2_legitimize_address): Recognize (exp + constant) directly.
7312 (TARGET_ADDRESS_COST): Define.
7314 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7316 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7317 (nios2_symbolic_memory_operand_p): Declare.
7318 (nios2_split_large_constant): Declare.
7319 (nios2_split_symbolic_memory_operand): Declare.
7320 * config/nios2/nios2.c: Adjust includes.
7321 (nios2_symbolic_constant_allowed): New.
7322 (nios2_symbolic_constant_p): New.
7323 (nios2_plus_symbolic_constant_p): New.
7324 (nios2_valid_addr_expr_p): Recognize addresses involving
7326 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7327 (nios2_symbolic_memory_operand_p): New.
7328 (nios2_large_constant_p): New.
7329 (nios2_split_large_constant): New.
7330 (nios2_split_plus_large_constant): New.
7331 (nios2_split_symbolic_memory_operand): New.
7332 (nios2_legitimize_address): Code refactoring. Handle addresses
7333 involving symbolic constants.
7334 (nios2_emit_move_sequence): Likewise.
7335 (nios2_print_operand): Improve error output.
7336 (nios2_print_operand_address): Handle LO_SUM.
7337 (nios2_cdx_narrow_form_p): Likewise.
7338 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7339 operands involving symbolic constants.
7340 (movhi_internal, movsi_internal): Likewise.
7341 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7342 (extendhisi2, extendqi<mode>2): Likewise.
7344 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7346 * tree-pass.h (PROP_rtl_split_insns): Define.
7347 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7349 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7351 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7353 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7356 * target.def (const_not_ok_for_debug_p): Default to
7357 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7358 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7359 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7360 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7361 which targetm.const_not_ok_for_debug_p returned true.
7362 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7364 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7366 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7367 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7368 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7370 (ix86_const_not_ok_for_debug_p): New function.
7371 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7372 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7374 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7377 * system.h: Conditionally include "unique-ptr.h" if
7378 INCLUDE_UNIQUE_PTR is defined.
7379 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7380 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7382 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7384 * config/rl78/rl78.md: New define_expand "subdi3".
7386 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7389 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7390 DF_REF_INSN if DF_REF_INSN_INFO is false.
7392 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7394 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7395 xmm_move instead of sse_move.
7396 (sse_store_index): New function.
7397 (ix86_register_move_cost): Be more sensible about mismatch stall;
7398 model AVX moves correctly; make difference between sse->integer and
7400 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7401 moves; make difference between SSE and AVX.
7402 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7403 and zmm_move. Increase size of sse load and store tables;
7404 add unaligned load and store tables; add ssemmx_to_integer.
7405 * x86-tune-costs.h: Update all entries according to real
7406 move latencies from Agner Fog's manual and chip documentation.
7408 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7411 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7412 * config/i386/constraints.md (Wf): New constraint.
7413 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7414 (cmp<dwi>_doubleword): Removed.
7415 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7416 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7417 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7418 expand with cmp<dwi>_doubleword. For LTU and GEU use
7419 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7421 * common.opt (gcolumn-info): Enable by default.
7422 * doc/invoke.texi (gcolumn-info): Document new default.
7424 2017-10-23 Richard Biener <rguenther@suse.de>
7426 PR tree-optimization/82672
7427 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7428 Fold the stmt if we propagated into it.
7430 2017-10-23 Richard Biener <rguenther@suse.de>
7432 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7433 (bitmap_remove_expr_from_set): ... this. All callers call this
7434 for non-constant values.
7435 (bitmap_set_subtract): Rename to...
7436 (bitmap_set_subtract_expressions): ... this. Adjust and
7438 (bitmap_set_contains_value): Remove superfluous check.
7439 (bitmap_set_replace_value): Inline into single caller ...
7440 (bitmap_value_replace_in_set): ... here and simplify.
7441 (dependent_clean): Merge into ...
7442 (clean): ... this using an overload. Adjust.
7443 (prune_clobbered_mems): Adjust.
7444 (compute_antic_aux): Likewise.
7445 (compute_partial_antic_aux): Likewise.
7447 2017-10-23 Richard Biener <rguenther@suse.de>
7449 PR tree-optimization/82129
7451 2017-08-01 Richard Biener <rguenther@suse.de>
7453 PR tree-optimization/81181
7454 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7455 (compute_antic): ... end of iteration here.
7457 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7459 * target.def (starting_frame_offset): New hook.
7460 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7461 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7462 * doc/tm.texi.in: Regenerate.
7463 * hooks.h (hook_hwi_void_0): Declare.
7464 * hooks.c (hook_hwi_void_0): New function.
7465 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7466 STARTING_FRAME_OFFSET.
7467 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7468 * reload1.c (reload): Likewise.
7469 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7470 instead of STARTING_FRAME_OFFSET.
7471 * function.c (try_fit_stack_local): Likewise.
7472 (assign_stack_local_1): Likewise
7473 (instantiate_virtual_regs): Likewise.
7474 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7475 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7476 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7477 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7478 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7479 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7480 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7481 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7482 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7483 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7484 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7485 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7486 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7487 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7488 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7489 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7490 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7491 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7492 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7493 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7494 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7495 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7496 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7497 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7498 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7499 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7500 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7501 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7502 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7503 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7504 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7505 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7506 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7507 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7508 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7509 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7510 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7511 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7512 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7513 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7514 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7515 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7516 return a HOST_WIDE_INT.
7517 (avr_builtin_setjmp_frame_value): Use it instead of
7518 STARTING_FRAME_OFFSET.
7519 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7520 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7521 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7523 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7524 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7525 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7526 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7527 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7528 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7529 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7530 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7531 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7532 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7533 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7534 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7536 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7537 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7538 * config/mips/mips.c (mips_compute_frame_info): Refer to
7539 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7540 (mips_starting_frame_offset): New function.
7541 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7542 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7543 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7544 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7545 and return a HOST_WIDE_INT.
7546 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7547 (mmix_initial_elimination_offset): Refer to
7548 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7549 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7550 * config/pa/pa.c (pa_starting_frame_offset): New function.
7551 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7552 (pa_expand_prologue): Likewise.
7553 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7554 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7555 !FRAME_GROWS_DOWNWARD handling to...
7556 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7557 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7558 !FRAME_GROWS_DOWNWARD handling to...
7559 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7560 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7561 !FRAME_GROWS_DOWNWARD handling to...
7562 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7563 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7565 (rs6000_starting_frame_offset): New function.
7566 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7567 !FRAME_GROWS_DOWNWARD handling to...
7568 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7569 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7570 !FRAME_GROWS_DOWNWARD handling to...
7571 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7572 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7573 !FRAME_GROWS_DOWNWARD handling to...
7574 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7575 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7576 (rs6000_starting_frame_offset): New function.
7577 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7578 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7579 * config/vax/vax.c (vax_starting_frame_offset): New function.
7580 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7581 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7582 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7583 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7584 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7585 * system.h (STARTING_FRAME_OFFSET): Poison.
7587 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7589 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7590 SCALAR_TYPE_MODE instead of TYPE_MODE.
7592 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7593 Alan Hayward <alan.hayward@arm.com>
7594 David Sherwood <david.sherwood@arm.com>
7596 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7598 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7599 Alan Hayward <alan.hayward@arm.com>
7600 David Sherwood <david.sherwood@arm.com>
7602 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7604 2017-10-23 Richard Biener <rguenther@suse.de>
7606 PR tree-optimization/82129
7607 * tree-ssa-pre.c (bitmap_set_and): Remove.
7608 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7609 canonicalizing expressions in the set to those with lowest
7610 ID rather than taking that from the first edge.
7612 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7614 * combine.c (rtx_equal_for_field_assignment_p): Use
7615 byte_lowpart_offset.
7617 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7618 Alan Hayward <alan.hayward@arm.com>
7619 David Sherwood <david.sherwood@arm.com>
7621 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7622 to a SUBREG_PROMOTED_VAR.
7624 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7625 Alan Hayward <alan.hayward@arm.com>
7626 David Sherwood <david.sherwood@arm.com>
7628 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7629 (expand_debug_source_expr): Likewise.
7630 * combine.c (combine_simplify_rtx): Likewise.
7631 * cse.c (fold_rtx): Likewise.
7632 * optabs.c (expand_float): Likewise.
7633 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7634 (simplify_binary_operation_1): Likewise.
7636 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7637 Alan Hayward <alan.hayward@arm.com>
7638 David Sherwood <david.sherwood@arm.com>
7640 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7641 (record_promoted_value): Likewise.
7642 * expr.c (expand_expr_real_2): Likewise.
7643 * ree.c (update_reg_equal_equiv_notes): Likewise.
7644 (combine_set_extension): Likewise.
7645 * rtlanal.c (low_bitmask_len): Likewise.
7646 * simplify-rtx.c (neg_const_int): Likewise.
7647 (simplify_binary_operation_1): Likewise.
7649 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7650 Alan Hayward <alan.hayward@arm.com>
7651 David Sherwood <david.sherwood@arm.com>
7653 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7654 * regcprop.c (maybe_mode_change): Likewise.
7655 * reload1.c (alter_reg): Likewise.
7657 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7659 * inchash.h (inchash::hash::add_wide_int): New function.
7660 * lto-streamer-out.c (hash_tree): Use it.
7662 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7664 * inchash.h (inchash::hash::add_wide_int): Rename to...
7665 (inchash::hash::add_hwi): ...this.
7666 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7667 (polymorphic_call_target_hasher::hash): Likewise.
7668 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7669 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7670 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7671 * lto-streamer-out.c (hash_tree): Likewise.
7672 * optc-save-gen.awk: Likewise.
7673 * tree.c (add_expr): Likewise.
7675 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7678 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7679 for ordered inequality comparisons even with TARGET_IEEE_FP.
7681 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7684 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7685 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7686 Expand with cmp<dwi>_doubleword.
7688 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7690 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7691 List CET intrinsics.
7692 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7693 specific to -fcf-protection option.
7695 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7697 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7698 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7699 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7700 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7701 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7702 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7703 (extra_objs): Add cet.o for Linux/x86 targets.
7704 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7705 * config/i386/cet.c: New file.
7706 * config/i386/cetintrin.h: Likewise.
7707 * config/i386/t-cet: Likewise.
7708 * config/i386/cpuid.h (bit_SHSTK): New.
7709 (bit_IBT): Likewise.
7710 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7711 pass IBT and SHSTK bits.
7712 * config/i386/i386-builtin-types.def
7713 (VOID_FTYPE_UNSIGNED_PVOID): New.
7714 (VOID_FTYPE_UINT64_PVOID): Likewise.
7715 * config/i386/i386-builtin.def: Add CET intrinsics.
7716 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7717 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7718 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7719 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7721 * config/i386/i386.c (rest_of_insert_endbranch): New.
7722 (pass_data_insert_endbranch): Likewise.
7723 (pass_insert_endbranch): Likewise.
7724 (make_pass_insert_endbranch): Likewise.
7725 (ix86_notrack_prefixed_insn_p): Likewise.
7726 (ix86_target_string): Add -mibt, -mshstk flags.
7727 (ix86_option_override_internal): Add flag_cf_protection
7729 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7730 (ix86_print_operand): Add 'notrack' prefix output.
7731 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7732 (ix86_expand_builtin): Expand CET intrinsics.
7733 (x86_output_mi_thunk): Add 'endbranch' instruction.
7734 * config/i386/i386.h (TARGET_IBT): New.
7735 (TARGET_IBT_P): Likewise.
7736 (TARGET_SHSTK): Likewise.
7737 (TARGET_SHSTK_P): Likewise.
7738 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7739 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7740 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7741 (builtin_setjmp_setup): New pattern.
7742 (builtin_longjmp): Likewise.
7743 (rdssp<mode>): Likewise.
7744 (incssp<mode>): Likewise.
7745 (saveprevssp): Likewise.
7746 (rstorssp): Likewise.
7747 (wrss<mode>): Likewise.
7748 (wruss<mode>): Likewise.
7749 (setssbsy): Likewise.
7750 (clrssbsy): Likewise.
7751 (nop_endbr): Likewise.
7752 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7754 * config/i386/immintrin.h: Include <cetintrin.h>.
7755 * config/i386/linux-common.h
7756 (file_end_indicate_exec_stack_and_cet): New prototype.
7757 (TARGET_ASM_FILE_END): New.
7759 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7761 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7762 latencies instead of having separate table; make difference between
7763 integer and float costs.
7764 * i386.h (processor_costs): Remove scalar_stmt_cost,
7765 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7766 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7768 * x86-tune-costs.h: Remove entries which has been removed in
7769 procesor_costs from all tables; make cond_taken_branch_cost
7770 and cond_not_taken_branch_cost COST_N_INSNS based.
7772 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7774 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7776 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7778 * config/i386/i386.md (isa): Remove fma_avx512f.
7779 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7780 <avx512>_fmadd_<mode>_mask3<round_name>,
7781 <avx512>_fmsub_<mode>_mask<round_name>,
7782 <avx512>_fmsub_<mode>_mask3<round_name>,
7783 <avx512>_fnmadd_<mode>_mask<round_name>,
7784 <avx512>_fnmadd_<mode>_mask3<round_name>,
7785 <avx512>_fnmsub_<mode>_mask<round_name>,
7786 <avx512>_fnmsub_<mode>_mask3<round_name>,
7787 <avx512>_fmaddsub_<mode>_mask<round_name>,
7788 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7789 <avx512>_fmsubadd_<mode>_mask<round_name>,
7790 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7791 (*vec_widen_umult_even_v16si<mask_name>,
7792 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7793 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7795 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7797 * extend.texi: Add 'nocf_check' documentation.
7798 * gimple.texi: Add second parameter to
7799 gimple_build_call_from_tree.
7800 * invoke.texi: Add -fcf-protection documentation.
7801 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7803 2017-10-20 Richard Biener <rguenther@suse.de>
7805 PR tree-optimization/82473
7806 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7807 the largest input type.
7809 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7811 * c-attribs.c (handle_nocf_check_attribute): New function.
7812 (c_common_attribute_table): Add 'nocf_check' handling.
7813 * gimple-parser.c: Add second argument NULL to
7814 gimple_build_call_from_tree.
7815 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7816 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7818 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7819 * common.opt: Add fcf-protection flag.
7820 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7821 * flag-types.h: Add enum cf_protection_level.
7822 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7823 Add 'nocf_check' attribute propagation to gimple call.
7824 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7825 (gimple_build_call_from_tree): Update prototype.
7826 (gimple_call_nocf_check_p): New function.
7827 (gimple_call_set_nocf_check): Likewise.
7828 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7829 * ipa-icf.c: Add nocf_check attribute in statement hash.
7830 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7831 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7832 * toplev.c (process_options): Add flag_cf_protection handling.
7834 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7836 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7838 2017-10-20 Richard Biener <rguenther@suse.de>
7840 PR tree-optimization/82603
7841 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7842 remove false predicated stores.
7844 2017-10-20 Richard Biener <rguenther@suse.de>
7846 * graphite-isl-ast-to-gimple.c
7847 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7848 Remove return value and simplify, dump copied stmt after lhs
7850 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7851 Reduce dump verbosity.
7852 (gsi_insert_earliest): Likewise.
7853 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7854 * graphite.c (print_global_statistics): Adjust dumping.
7855 (print_graphite_scop_statistics): Likewise.
7856 (print_graphite_statistics): Do not dump loops here.
7857 (graphite_transform_loops): But here.
7859 2017-10-20 Nicolas Roche <roche@adacore.com>
7861 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7862 * configure: Regenerate.
7864 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7867 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7868 functions when optimizing replace GIMPLE_RETURN stmts with
7869 calls to __builtin_unreachable ().
7872 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7873 for -fsanitize=thread link of executables.
7874 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7875 link of executables.
7878 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7880 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7881 define_insns for logical vector shifts to use VI248_AVX512BW
7882 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7883 condition, useless isa and prefix attributes. Change the first
7884 2 of these define_insns to ...
7885 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7886 define_insn for avx512vl.
7887 (<shift_insn><mode>3): ... and this, new define_insn without
7888 masking for non-avx512vl.
7891 * config/i386/sse.md (*andnot<mode>3,
7892 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7893 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7894 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7895 not applied use empty suffix even for TARGET_AVX512VL.
7896 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7897 is applied, supply evex,evex or evex,evex,evex instead of just
7900 2017-10-20 Julia Koval <julia.koval@intel.com>
7902 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7903 (OPTION_MASK_ISA_GFNI_UNSET): New.
7904 (ix86_handle_option): Handle OPT_mgfni.
7905 * config/i386/cpuid.h (bit_GFNI): New.
7906 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7907 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7908 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7909 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7910 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7911 * config/i386/i386.opt: Add mgfni.
7913 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7915 * config/msp430/msp430.c (msp430_option_override): Disable
7916 -fdelete-null-pointer-checks.
7917 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7919 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7921 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7922 of x87 and SSE instructions.
7924 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7926 * asan.c (create_cond_insert_point): Do not update edge count.
7927 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7928 (afdo_propagate_circuit): Likewise.
7929 (afdo_calculate_branch_prob): Likewise.
7930 (afdo_annotate_cfg): Likewise.
7931 * basic-block.h (struct edge_def): Remove count.
7932 (edge_def::count): New accessor.
7933 * bb-reorder.c (rotate_loop): Update.
7934 (find_traces_1_round): Update.
7935 (connect_traces): Update.
7936 (sanitize_hot_paths): Update.
7937 * cfg.c (unchecked_make_edge): Update.
7938 (make_single_succ_edge): Update.
7939 (check_bb_profile): Update.
7940 (dump_edge_info): Update.
7941 (update_bb_profile_for_threading): Update.
7942 (scale_bbs_frequencies_int): Update.
7943 (scale_bbs_frequencies_gcov_type): Update.
7944 (scale_bbs_frequencies_profile_count): Update.
7945 (scale_bbs_frequencies): Update.
7946 * cfganal.c (connect_infinite_loops_to_exit): Update.
7947 * cfgbuild.c (compute_outgoing_frequencies): Update.
7948 (find_many_sub_basic_blocks): Update.
7949 * cfgcleanup.c (try_forward_edges): Update.
7950 (try_crossjump_to_edge): Update
7951 * cfgexpand.c (expand_gimple_cond): Update
7952 (expand_gimple_tailcall): Update
7953 (construct_exit_block): Update
7954 * cfghooks.c (verify_flow_info): Update
7955 (redirect_edge_succ_nodup): Update
7956 (split_edge): Update
7957 (make_forwarder_block): Update
7958 (duplicate_block): Update
7959 (account_profile_record): Update
7960 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7961 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7962 * cfgloopmanip.c (scale_loop_profile): Update.
7964 (lv_adjust_loop_entry_edge): Update.
7965 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7966 (force_nonfallthru_and_redirect): Update.
7967 (purge_dead_edges): Update.
7968 (rtl_flow_call_edges_add): Update.
7969 * cgraphunit.c (init_lowered_empty_function): Update.
7970 (cgraph_node::expand_thunk): Update.
7971 * gimple-pretty-print.c (dump_probability): Update.
7972 (dump_edge_probability): Update.
7973 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7974 * haifa-sched.c (sched_create_recovery_edges): Update.
7975 * hsa-gen.c (convert_switch_statements): Update.
7976 * ifcvt.c (dead_or_predicable): Update.
7977 * ipa-inline-transform.c (inline_transform): Update.
7978 * ipa-split.c (split_function): Update.
7979 * ipa-utils.c (ipa_merge_profiles): Update.
7980 * loop-doloop.c (add_test): Update.
7981 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7982 * lto-streamer-in.c (input_cfg): Update.
7983 (input_function): Update.
7984 * lto-streamer-out.c (output_cfg): Update.
7985 * modulo-sched.c (sms_schedule): Update.
7986 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7987 * predict.c (maybe_hot_edge_p): Update.
7988 (unlikely_executed_edge_p): Update.
7989 (probably_never_executed_edge_p): Update.
7990 (dump_prediction): Update.
7991 (drop_profile): Update.
7992 (propagate_unlikely_bbs_forward): Update.
7993 (determine_unlikely_bbs): Update.
7994 (force_edge_cold): Update.
7995 * profile.c (compute_branch_probabilities): Update.
7996 * reg-stack.c (better_edge): Update.
7997 * shrink-wrap.c (handle_simple_exit): Update.
7998 * tracer.c (better_p): Update.
7999 * trans-mem.c (expand_transaction): Update.
8000 (split_bb_make_tm_edge): Update.
8001 * tree-call-cdce.c: Update.
8002 * tree-cfg.c (gimple_find_sub_bbs): Update.
8003 (gimple_split_edge): Update.
8004 (gimple_duplicate_sese_region): Update.
8005 (gimple_duplicate_sese_tail): Update.
8006 (gimple_flow_call_edges_add): Update.
8007 (insert_cond_bb): Update.
8008 (execute_fixup_cfg): Update.
8009 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8010 * tree-complex.c (expand_complex_div_wide): Update.
8011 * tree-eh.c (lower_resx): Update.
8012 (unsplit_eh): Update.
8013 (cleanup_empty_eh_move_lp): Update.
8014 * tree-inline.c (copy_edges_for_bb): Update.
8015 (freqs_to_counts): Update.
8016 (copy_cfg_body): Update.
8017 * tree-ssa-dce.c (remove_dead_stmt): Update.
8018 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8019 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8020 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8021 (unloop_loops): Update.
8022 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8023 * tree-ssa-loop-split.c (connect_loops): Update.
8024 (split_loop): Update.
8025 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8026 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8027 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8028 * tree-ssa-reassoc.c (branch_fixup): Update.
8029 * tree-ssa-tail-merge.c (replace_block_by): Update.
8030 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8031 (compute_path_counts): Update.
8032 (update_profile): Update.
8033 (recompute_probabilities): Update.
8034 (update_joiner_offpath_counts): Update.
8035 (estimated_freqs_path): Update.
8036 (freqs_to_counts_path): Update.
8037 (clear_counts_path): Update.
8038 (ssa_fix_duplicate_block_edges): Update.
8039 (duplicate_thread_path): Update.
8040 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8041 (case_bit_test_cmp): Update.
8042 (collect_switch_conv_info): Update.
8043 (gen_inbound_check): Update.
8044 (do_jump_if_equal): Update.
8045 (emit_cmp_and_jump_insns): Update.
8046 * tree-tailcall.c (decrease_profile): Update.
8047 (eliminate_tail_call): Update.
8048 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8049 (vect_do_peeling): Update.
8050 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8051 * ubsan.c (ubsan_expand_null_ifn): Update.
8052 (ubsan_expand_ptr_ifn): Update.
8053 * value-prof.c (gimple_divmod_fixed_value): Update.
8054 (gimple_mod_pow2): Update.
8055 (gimple_mod_subtract): Update.
8056 (gimple_ic): Update.
8057 (gimple_stringop_fixed_value): Update.
8059 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8062 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8064 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8066 PR rtl-optimization/82395
8067 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8068 based on non_spilled_static_chain_regno_p.
8070 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8072 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8073 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8074 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8076 2017-10-19 Martin Sebor <msebor@redhat.com>
8078 PR tree-optimization/82596
8079 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8081 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8083 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8084 (get_mem_refs_of_builtin_call): Likewise.
8085 * builtins.c (expand_builtin_apply): Adjust call to
8086 allocate_dynamic_stack_space.
8087 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8088 the third argument to allocate_dynamic_stack_space, otherwise -1.
8089 (expand_builtin): Deal with all alloca variants.
8090 (is_inexpensive_builtin): Likewise.
8091 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8092 * calls.c (special_function_p): Deal with all alloca variants.
8093 (initialize_argument_information): Adjust call to
8094 allocate_dynamic_stack_space.
8095 (expand_call): Likewise.
8096 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8097 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8098 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8099 use it for the stack usage computation.
8100 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8101 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8102 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8103 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8104 (in_loop_p): Remove first argument and useless check.
8105 (pass_walloca::execute): Remove useless test and adjust call to above.
8106 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8107 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8108 (gimplify_call_expr): Deal with all alloca variants.
8109 * hsa-gen.c (gen_hsa_alloca): Likewise.
8110 (gen_hsa_insns_for_call): Likewise.
8111 * ipa-pure-const.c (special_builtin_state): Likewise.
8112 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8113 * tree-object-size.c (alloc_object_size): Likewise.
8114 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8115 (call_may_clobber_ref_p_1): Likewise.
8116 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8117 (ccp_fold_stmt): Likewise.
8118 (optimize_stack_restore): Likewise.
8119 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8120 (mark_all_reaching_defs_necessary_1): Likewise.
8121 (propagate_necessity): Likewise.
8122 (eliminate_unnecessary_stmts): Likewise.
8123 * tree.c (build_common_builtin_nodes): Build
8124 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8125 (build_alloca_call_expr): New function.
8126 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8127 (CASE_BUILT_IN_ALLOCA): Likewise.
8128 (build_alloca_call_expr): Declare.
8129 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8131 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8134 * dwarf2out.c (new_die_raw): New static inline function.
8135 (new_die): Use it to create the DIE.
8136 (add_AT_external_die_ref): Likewise.
8137 (clone_die): Likewise.
8138 (clone_as_declaration): Likewise.
8139 (dwarf2out_vms_debug_main_pointer): Likewise.
8140 (base_type_die): Likewise. Remove early return for corner cases.
8141 Do not call add_pubtype on the DIE here.
8142 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8143 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8144 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8145 native order exists for base types, attach the DIE manually and call
8146 add_pubtype on it. Do not equate a reverse order DIE to the type.
8148 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8150 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8151 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8153 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8155 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8157 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8158 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8159 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8160 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8161 * opts.c (sanitizer_opts): Add builtin.
8162 * ubsan.c (instrument_builtin): New function.
8163 (pass_ubsan::execute): Call it.
8164 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8165 * doc/invoke.texi: Document -fsanitize=builtin.
8167 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8168 builtins, store max (log2 (align), 0) into uchar field instead of
8169 align into uptr field.
8170 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8171 store uchar 0 field instead of uptr 0 field.
8172 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8173 instead of passing one address of struct with 2 locations pass
8174 two addresses of structs with 1 location each.
8175 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8176 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8177 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8178 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8179 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8180 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8181 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8182 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8184 2017-10-19 Martin Liska <mliska@suse.cz>
8187 * file-find.c (remove_prefix): Remove.
8188 * file-find.h (remove_prefix): Likewise.
8189 * gcc-ar.c: Remove smartness of lookup.
8191 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8193 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8194 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8195 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8197 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8200 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8201 (setcc + and to xor + setcc): New peephole2.
8203 2017-10-19 Tom de Vries <tom@codesourcery.com>
8205 * doc/sourcebuild.texi (Test Directives, Variants of
8206 dg-require-support): Add dg-require-stack-size.
8208 2017-10-19 Martin Liska <mliska@suse.cz>
8211 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8212 that have a large alignment.
8213 (gimplify_target_expr): Likewise.
8215 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8217 PR rtl-optimization/82602
8218 * ira.c (rtx_moveable_p): Return false for volatile asm.
8220 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8223 * config/i386/i386-modes.def (CCGZ): New CC mode.
8224 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8225 * config/i386/predicates.md (ix86_comparison_operator):
8227 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8228 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8229 with double-word subtraction.
8230 (put_condition_code): Handle CCGZmode.
8232 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8234 * wide-int.cc (debug (const wide_int &)): New.
8235 (debug (const wide_int *)): New.
8236 (debug (const widest_int &)): New.
8237 (debug (const widest_int *)): New.
8239 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8242 * lra-constraints.c (curr_insn_transform): Use non-input operand
8243 instead of output one for matched reload.
8245 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8247 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8248 (tree-ssa-loop-ivopts.h): New header file.
8249 (struct builtin_info): New fields.
8250 (classify_builtin_1): Compute and record base and offset parts for
8251 memset builtin partition by calling strip_offset.
8252 (offset_cmp, fuse_memset_builtins): New functions.
8253 (finalize_partitions): Fuse adjacent memset partitions by calling
8255 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8256 Expose the interface.
8257 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8259 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8261 PR tree-optimization/82574
8262 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8263 that data reference must be executed exactly once per iteration
8264 against the outermost loop in nest.
8265 (classify_partition): Update call to above function.
8267 2017-10-18 Richard Biener <rguenther@suse.de>
8269 PR tree-optimization/82591
8270 * graphite.c (graphite_transform_loops): Move code gen message
8272 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8273 Here. Handle scop_to_isl_ast failing.
8274 (scop_to_isl_ast): Limit the number of ISL operations.
8276 2017-10-18 Richard Biener <rguenther@suse.de>
8278 * graphite-isl-ast-to-gimple.c
8279 (translate_isl_ast_to_gimple::set_rename): Simplify.
8280 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8281 (graphite_copy_stmts_from_block): ... here.
8282 (copy_bb_and_scalar_dependences): Simplify.
8283 (add_parameters_to_ivs_params): Canonicalize.
8284 (generate_entry_out_of_ssa_copies): Simplify.
8285 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8286 by passing in ISL dimension.
8287 (parameter_index_in_region_1): Rename to ...
8288 (parameter_index_in_region): ... this.
8289 (extract_affine): Adjust assert, pass down parameter index.
8290 (add_param_constraints): Use range-info when available.
8291 (build_scop_context): Adjust.
8292 * sese.c (new_sese_info): Adjust.
8293 (free_sese_info): Likewise.
8294 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8295 Remove unused typedefs.
8296 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8298 2017-10-18 Martin Liska <mliska@suse.cz>
8300 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8302 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8304 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8305 (s390_sched_init): Do not reset s390_sched_state if we entered the
8306 current basic block via a fallthru edge and all others are unlikely.
8308 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8310 * config/s390/s390.c (NUM_SIDES): New variable.
8311 (LONGRUNNING_THRESHOLD): New variable.
8312 (LATENCY_FACTOR): New variable.
8313 (s390_sched_score): Decrease score for long-running instructions on
8315 (s390_sched_variable_issue): Perform bookkeeping for long-running
8318 2017-10-18 Richard Biener <rguenther@suse.de>
8320 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8321 Simplify with removal of the parameter rename map.
8322 (set_rename): Likewise.
8323 (should_copy_to_new_region): Likewise.
8324 (graphite_copy_stmts_from_block): Likewise.
8325 (copy_bb_and_scalar_dependences): Remove initialization of
8326 unused copied_bb_map.
8328 (copy_internal_parameters): Likewise.
8329 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8330 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8331 Use INTEGRAL_TYPE_P.
8332 (parameter_index_in_region_1): Rename to ...
8333 (assign_parameter_index_in_region): ... this. Assert we have
8334 a parameter we handle.
8335 (scan_tree_for_params): Adjust.
8336 * sese.h (parameter_rename_map_t): Remove.
8337 (struct sese_info_t): Remove unused parameter_rename_map and
8338 copied_bb_map members.
8339 * sese.c (new_sese_info): Adjust.
8340 (free_sese_info): Likewise.
8342 2017-10-18 Martin Liska <mliska@suse.cz>
8345 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8346 on an abnormal edge.
8348 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8350 * doc/invoke.texi (ffunction-sections and fdata-sections):
8353 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8355 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8356 the use statement can throw internally.
8358 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8360 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8361 any RTX present on the RHS of a SET.
8362 * compare-elim.c (try_eliminate_compare): Restore comment.
8364 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8366 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8367 may be also called on tcc_type nodes.
8368 * langhooks.c (lhd_tree_size): Likewise.
8370 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8372 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8373 format_warning_at_substring.
8374 (maybe_warn): Convert source_range * param to a location_t. Pass
8375 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8376 (format_directive): Remove code to extract source_ranges and
8377 source_range * in favor of just a location_t.
8378 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8380 * substring-locations.c (format_warning_va): Convert
8381 source_range * param to a location_t.
8382 (format_warning_at_substring): Likewise.
8383 * substring-locations.h (format_warning_va): Likewise.
8384 (format_warning_at_substring): Likewise.
8386 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8388 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8390 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8391 and scatter/gather ops.
8393 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8394 vec_gather_load and vec_scatter_store.
8395 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8396 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8397 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8398 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8399 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8400 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8402 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8404 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8405 Detect FTST instruction and handle its register pops. Only pop
8406 second operand if can_pop_second_op is true.
8407 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8408 set can_pop_second_op to false in the compare_for_stack_reg call.
8410 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8411 output_fp_compare for stack register operands.
8412 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8413 instructions here. Do not emit stack register pops here. Assert
8414 that FCOMPP pops next to top stack register. Rewrite function.
8416 2017-10-17 Nathan Sidwell <nathan@acm.org>
8419 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8420 use DECL_ASSEMBLER_NAME_RAW.
8423 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8426 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8427 Wilco Dijkstra <wilco.dijkstra@arm.com>
8429 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8430 converstion to Pmode from the buf_addr.
8432 2017-10-17 Richard Biener <rguenther@suse.de>
8434 * graphite-dependences.c (scop_get_reads_and_writes): Change
8435 output parameters to references.
8437 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8439 PR 71026/tree-optimization
8440 * fold-const.c (distribute_real_division): Removed.
8441 (fold_binary_loc): Remove calls to distribute_real_divison.
8443 2017-10-17 Richard Biener <rguenther@suse.de>
8445 * graphite-scop-detection.c
8446 (scop_detection::stmt_has_simple_data_refs_p): Always use
8447 the full nest as region.
8448 (try_generate_gimple_bb): Likewise.
8449 * sese.c (scalar_evolution_in_region): Simplify now that
8450 SCEV can handle instantiation in regions.
8451 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8452 in the non-loop part of a function if requested.
8454 2017-10-17 Richard Biener <rguenther@suse.de>
8456 PR tree-optimization/82563
8457 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8459 (graphite_regenerate_ast_isl): Call it.
8460 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8462 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8464 PR tree-optimization/82549
8465 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8466 Formatting fixes. Instead of calling make_bit_field_ref with negative
8469 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8471 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8472 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8473 _mm_maskz_reduce_ss): New.
8474 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8475 __builtin_ia32_reducess_mask): Ditto..
8476 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8477 * config/i386/sse.md (reduces<mode>): Renamed to ...
8478 (reduces<mode><mask_scalar_name>): ... this.
8479 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8481 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8482 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8484 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8486 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8487 * selftest-run-tests.c (selftest::run_tests): Call
8488 selftest::unique_ptr_tests_cc_tests.
8489 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8490 * unique-ptr-tests.cc: New file.
8492 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8495 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8497 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8498 (make_hard_regno_born, make_hard_regno_dead): Update
8499 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8501 2017-10-16 Jeff Law <law@redhat.com>
8503 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8505 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8507 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8509 2017-10-16 Olivier Hainque <hainque@adacore.com>
8511 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8512 with_cpu if we were configured for an e500v2 target cpu name.
8514 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8516 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8517 * doc/invoke.texi: Document +nodsp as a valid extension for
8520 2017-10-16 Martin Liska <mliska@suse.cz>
8522 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8523 (test_set_range): Likewise.
8524 (test_range_functions): Rename to ...
8525 (test_bit_in_range): ... this.
8526 (sbitmap_c_tests): Add new test.
8528 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8530 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8532 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8533 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8535 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8537 * config/aarch64/aarch64-builtins.c
8538 (aarch64_types_quadopu_lane_qualifiers): New.
8539 (TYPES_QUADOPU_LANE): New.
8540 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8541 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8542 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8543 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8544 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8545 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8546 (Vdottype, DOTPROD): New.
8547 (sur): Add SDOT and UDOT.
8549 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8551 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8552 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8553 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8555 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8556 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8557 Enable TARGET_DOTPROD.
8558 (cortex-a75.cortex-a55): Likewise.
8559 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8561 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8563 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8564 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8566 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8568 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8569 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8570 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8571 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8572 * config/arm/types.md (neon_dot, neon_dot_q): New.
8573 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8575 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8577 * config/arm/arm.h (TARGET_DOTPROD): New.
8578 * config/arm/arm.c (arm_arch_dotprod): New.
8579 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8580 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8581 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8582 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8583 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8584 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8585 * doc/invoke.texi (armv8.2-a): Document dotprod
8587 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8589 * i386.c (ix86_vec_cost): New function.
8590 (ix86_rtx_costs): Handle vector operations better.
8591 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8592 * x86-tune-costs.h: Add new costs to all tables.
8594 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8596 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8598 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8599 divsd, sqrtss and sqrtsd
8600 * x86-tune-costs.h: Add new entries to all costs.
8601 (znver1_cost): Fix to match real instruction latencies.
8603 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8604 Michael Collison <michael.collison@arm.com>
8606 * compare-elim.c: Include emit-rtl.h.
8607 (can_merge_compare_into_arith): New function.
8608 (try_validate_parallel): Likewise.
8609 (try_merge_compare): Likewise.
8610 (try_eliminate_compare): Call the above when no previous clobber
8612 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8615 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8619 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8620 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8624 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8625 to be any operand_equal_p operands. For & (B - 1) require
8626 B to be power of 2. Recognize
8627 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8629 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8632 * optabs.c (expand_memory_blockage): Fix call of
8633 targetm.have_memory_blockage.
8635 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8638 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8639 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8640 objects to extra_objs instead of overwriting it.
8642 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8644 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8645 Use any_fp_register_operand as operand[3] predicate. Simplify
8646 equality test for operands[2] and operands[4] memory location.
8647 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8648 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8649 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8650 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8651 any_fp_register_operand as operand[1] predicate. Simplify
8652 equality test for operands[0] and operands[3] memory location.
8653 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8654 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8655 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8657 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8659 * target-insns.def: Add memory_blockage.
8660 * optabs.c (expand_memory_blockage): New function.
8661 (expand_asm_memory_barrier): Rename ...
8662 (expand_asm_memory_blockage): ... to this.
8663 (expand_mem_thread_fence): Call expand_memory_blockage
8664 instead of expand_asm_memory_barrier.
8665 (expand_mem_singnal_fence): Ditto.
8666 (expand_atomic_load): Ditto.
8667 (expand_atomic_store): Ditto.
8668 * doc/md.texi (Standard Pattern Names For Generation):
8669 Document memory_blockage instruction pattern.
8671 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8673 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8674 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8675 * config/rl78/rl78.md: New define_expand "adddi3".
8677 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8679 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8682 2017-10-13 Jeff Law <law@redhat.com>
8684 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8686 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8689 * internal-fn.c (expand_mul_overflow): If both operands have
8690 the same highpart of -1 or 0 and the topmost bit of lowpart
8691 is different, overflow is if res <= 0 rather than res < 0.
8693 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8695 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8696 TARGET_P9_VECTOR code for unaligned_load case.
8698 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8700 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8702 2017-10-13 Nathan Sidwell <nathan@acm.org>
8704 * tree-core.h (tree_contains_struct): Make bool.
8705 * tree.c (tree_contains_struct): Likewise.
8706 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8707 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8708 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8709 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8710 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8712 2017-10-13 Richard Biener <rguenther@suse.de>
8714 * graphite-isl-ast-to-gimple.c
8715 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8716 parameters and dominance check.
8717 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8718 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8719 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8720 Do not update SSA form here or do intermediate IL verification.
8721 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8722 (graphite_initialize): Remove check on the number of loops in
8723 the function and inline into graphite_transform_loops.
8724 (graphite_finalize): Inline into graphite_transform_loops.
8725 (graphite_transform_loops): Perform SSA update and IL verification
8727 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8729 2017-10-13 Richard Biener <rguenther@suse.de>
8731 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8732 graphite_expression_type_precision): Avoid global constructor
8734 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8735 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8736 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8737 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8738 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8739 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8741 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8744 * config/i386/i386.h (ix86_red_zone_size): New.
8745 * config/i386/i386.md (push peephole2s): Replace
8746 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8748 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8749 Alan Hayward <alan.hayward@arm.com>
8750 David Sherwood <david.sherwood@arm.com>
8752 * combine.c (can_change_dest_mode): Reject changes in
8753 REGMODE_NATURAL_SIZE.
8755 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8756 Alan Hayward <alan.hayward@arm.com>
8757 David Sherwood <david.sherwood@arm.com>
8759 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8760 (expand_debug_source_expr): Likewise.
8761 * combine.c (combine_simplify_rtx): Likewise.
8762 * cse.c (fold_rtx): Likewise.
8763 * fwprop.c (canonicalize_address): Likewise.
8764 * targhooks.c (default_shift_truncation_mask): Likewise.
8766 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8767 Alan Hayward <alan.hayward@arm.com>
8768 David Sherwood <david.sherwood@arm.com>
8770 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8771 (widened_mode): Likewise.
8772 (expand_unop): Likewise.
8773 * ree.c (transform_ifelse): Likewise.
8774 (merge_def_and_ext): Likewise.
8775 (combine_reaching_defs): Likewise.
8776 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8778 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8779 Alan Hayward <alan.hayward@arm.com>
8780 David Sherwood <david.sherwood@arm.com>
8782 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8783 * combine.c (gen_lowpart_for_combine): Likewise.
8784 * dwarf2out.c (rtl_for_decl_location): Likewise.
8785 * final.c (alter_subreg): Likewise.
8786 * rtlhooks.c (gen_lowpart_general): Likewise.
8787 (gen_lowpart_if_possible): Likewise.
8789 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8790 Alan Hayward <alan.hayward@arm.com>
8791 David Sherwood <david.sherwood@arm.com>
8793 * calls.c (expand_call): Use subreg_lowpart_offset.
8794 * cse.c (cse_insn): Likewise.
8795 * regcprop.c (copy_value): Likewise.
8796 (copyprop_hardreg_forward_1): Likewise.
8798 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8801 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8802 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8803 =Q constraints to +Q and into insn condition add check
8804 that operands[0] and operands[1] are equal.
8805 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8806 =Q constraints to +Q and into insn condition add check
8807 that operands[0] is equal to either operands[1] or operands[2].
8810 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8811 instead of handling MINUS_EXPR twice (once for each argument),
8812 canonicalize operand order and handle just once, use rtype where
8813 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8816 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8817 any values of __C while still being pattern recognizable as a simple
8820 2017-10-13 Richard Biener <rguenther@suse.de>
8822 PR tree-optimization/82451
8824 2017-10-02 Richard Biener <rguenther@suse.de>
8826 PR tree-optimization/82355
8827 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8828 a mapping for the enclosing loop but avoid generating one for
8830 (copy_bb_and_scalar_dependences): Remove premature codegen
8831 error on PHIs in blocks duplicated into multiple places.
8832 * graphite-scop-detection.c
8833 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8834 in the region use it as loop and nest to analyze the DR in.
8835 (try_generate_gimple_bb): Likewise.
8836 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8837 (add_loop_constraints): For blocks in a loop not in the region
8838 create a dimension with a single iteration.
8839 * sese.h (gbb_loop_at_index): Remove assert.
8841 * cfgloop.c (loop_preheader_edge): For the loop tree root
8842 return the single successor of the entry block.
8843 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8844 Reset the SCEV hashtable and niters.
8845 * graphite-scop-detection.c
8846 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8847 assert that we only have POLYNOMIAL_CHREC that vary in loops
8848 contained in the region.
8849 (scop_detection::graphite_can_represent_expr): Adjust.
8850 (scop_detection::stmt_has_simple_data_refs_p): For loops
8851 not in the region set loop to NULL. The nest is now the
8852 entry edge to the region.
8853 (try_generate_gimple_bb): Likewise.
8854 * sese.c (scalar_evolution_in_region): Adjust for
8855 instantiate_scev change.
8856 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8857 Make nest parameter the edge into the region.
8858 (create_data_ref): Likewise.
8859 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8860 entry edge into a region and adjust instantiate_scev calls.
8861 (create_data_ref): Likewise.
8862 (graphite_find_data_references_in_stmt): Likewise.
8863 (find_data_references_in_stmt): Pass the loop preheader edge
8864 from the nest argument.
8865 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8866 parameter the edge into the region.
8867 (instantiate_parameters): Use the loop preheader edge as entry.
8868 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8870 (get_instantiated_value_entry): Make instantiate_below parameter
8871 the edge into the region.
8872 (instantiate_scev_name): Likewise. Adjust dominance checks,
8873 when we cannot use loop-based instantiation instantiate by
8874 walking use-def chains.
8875 (instantiate_scev_poly): Adjust.
8876 (instantiate_scev_binary): Likewise.
8877 (instantiate_scev_convert): Likewise.
8878 (instantiate_scev_not): Likewise.
8879 (instantiate_array_ref): Remove.
8880 (instantiate_scev_3): Likewise.
8881 (instantiate_scev_2): Likewise.
8882 (instantiate_scev_1): Likewise.
8883 (instantiate_scev_r): Do not blindly handle N-operand trees.
8884 Do not instantiate array-refs. Handle all constants and invariants.
8885 (instantiate_scev): Make instantiate_below parameter
8886 the edge into the region.
8887 (resolve_mixers): Use the loop preheader edge for the region
8888 parameter to instantiate_scev_r.
8889 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8891 2017-10-13 Richard Biener <rguenther@suse.de>
8893 PR tree-optimization/82525
8894 * graphite-isl-ast-to-gimple.c
8895 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8897 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8898 Fail code generation when we cannot represent the isl integer.
8899 (binary_op_to_tree): Elide modulo operations that are no-ops
8900 in the type we code generate. Remove now superfluous code
8903 2017-10-13 Richard Biener <rguenther@suse.de>
8905 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8906 (scop_detection::harmful_loop_in_region): Remove premature
8907 IV type restriction.
8908 (scop_detection::graphite_can_represent_scev): We can handle
8909 pointer IVs just fine.
8911 2017-10-13 Alan Modra <amodra@gmail.com>
8913 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8914 "Clobbers and Scratch Registers". Add paragraph on
8915 alternative to clobbers for scratch registers and OpenBLAS
8918 2017-10-13 Alan Modra <amodra@gmail.com>
8920 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8921 example of a memory input for a string of known length. Move
8922 commentary out of table. Add a number of new examples
8923 covering array memory inputs.
8925 2017-10-12 Martin Liska <mliska@suse.cz>
8927 PR tree-optimization/82493
8928 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8929 (test_range_functions): New function.
8930 (sbitmap_c_tests): Likewise.
8931 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8932 * selftest.h (sbitmap_c_tests): New function.
8934 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8936 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8938 * config/rs6000/amo.h: Fix spacing issue.
8940 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8943 * config/i386/i386.md (*ashl<mode>3_mask_1,
8944 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8945 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8948 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8950 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8951 (profile_probability): Set max_probability
8952 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8954 * profile-count.c (profile_probability::differs_from_p): Do not
8955 rely on max_probaiblity == 10000
8957 2017-10-12 Jeff Law <law@redhat.com>
8959 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8962 2017-10-12 Martin Sebor <msebor@redhat.com>
8966 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8967 (handle_alias_pairs): Call it.
8968 * common.opt (-Wattribute-alias): New option.
8969 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8970 * doc/invoke.texi (-Wattribute-alias): Document.
8972 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8975 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8977 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8979 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8980 (make_hard_regno_born, make_hard_regno_dead): Update
8981 bb_killed_pseudos and bb_gen_pseudos.
8983 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8985 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8987 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8989 * config/alpha/alpha.c (alpha_split_conditional_move):
8990 Use std::swap instead of manually swapping.
8991 (alpha_stdarg_optimize_hook): Ditto.
8992 (alpha_canonicalize_comparison): Ditto.
8994 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8996 * tree-loop-distribution.c (struct builtin_info): New struct.
8997 (struct partition): Refactor fields into struct builtin_info.
8998 (partition_free): Free struct builtin_info.
8999 (build_size_arg_loc, build_addr_arg_loc): Delete.
9000 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9001 information from struct builtin_info.
9002 (find_single_drs): New function refactored from classify_partition.
9003 Also moved builtin validity checks to this function.
9004 (compute_access_range, alloc_builtin): New functions.
9005 (classify_builtin_st, classify_builtin_ldst): New functions.
9006 (classify_partition): Refactor code into functions find_single_drs,
9007 classify_builtin_st and classify_builtin_ldst.
9008 (distribute_loop): Don't do runtime alias check when distributing
9010 (find_seed_stmts_for_distribution): New function.
9011 (pass_loop_distribution::execute): Refactor code finding seed
9012 stmts into above function. Support distribution for the innermost
9013 two-level loop nest. Adjust dump information.
9015 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9017 * tree-loop-distribution.c: Adjust the general comment.
9018 (NUM_PARTITION_THRESHOLD): New macro.
9019 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9020 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9021 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9022 in call to build_partition_graph.
9023 (finalize_partitions): New parameter. Make loop distribution more
9024 conservative by fusing more partitions.
9025 (distribute_loop): Don't do runtime alias check in case of loop nest
9027 (find_seed_stmts_for_distribution): New function.
9028 (prepare_perfect_loop_nest): New function.
9029 (pass_loop_distribution::execute): Refactor code finding seed stmts
9030 and loop nest into above functions. Support loop nest distribution.
9031 Adjust dump information accordingly.
9033 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9035 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9036 and set PTYPE_SEQUENTIAL for merged partition.
9038 2017-10-12 Richard Biener <rguenther@suse.de>
9040 PR tree-optimization/69728
9042 2017-09-19 Richard Biener <rguenther@suse.de>
9044 PR tree-optimization/69728
9045 * graphite-sese-to-poly.c (schedule_error): New global.
9046 (add_loop_schedule): Handle empty domain by failing the
9048 (build_original_schedule): Handle schedule_error.
9050 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9051 domain by returning an unchanged schedule.
9053 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9055 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9056 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9058 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9060 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9063 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9066 * expr.c (store_field): Don't optimize away bitsize == 0 store
9067 from CALL_EXPR with addressable return type.
9069 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9071 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9072 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9073 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9074 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9075 TARGET_ISEL instead of TARGET_ISEL<sel>.
9077 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9079 * config/rs6000/rs6000.c
9080 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9082 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9084 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9086 (reg_or_cint_operand): Fix comment.
9087 (reg_or_zero_operand): New predicate.
9088 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9089 * config/rs6000/rs6000.c (output_isel): Delete.
9090 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9091 instead of reg_or_cint_operand. Output instruction directly (not via
9093 (isel_unsigned_<mode>): Ditto.
9094 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9095 gpc_reg_operand. Add an instruction alternative for this. Output
9096 instruction directly.
9097 (*isel_reversed_unsigned_<mode>): Ditto.
9099 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9101 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9102 (TARGET_CANONICALIZE_COMPARISON): Define.
9104 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9107 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9108 Check whether the dest is REG before adding REG_EQUIV note.
9110 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9113 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9115 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9116 (make_hard_regno_born, make_hard_regno_dead): Update
9117 bb_killed_pseudos and bb_gen_pseudos.
9119 2017-10-11 Nathan Sidwell <nathan@acm.org>
9121 * incpath.h (enum incpath_kind): Name enum, prefix values.
9122 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9123 * incpath.c (heads, tails): Use INC_MAX.
9124 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9125 (merge_include_chains, split_quote_chain,
9126 register_include_chains): Update incpath_kind names.
9127 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9128 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9130 (add_framework_path, darwin_register_objc_includes): Likewise.
9131 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9133 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9135 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9136 Do not use float_operator operator predicate.
9137 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9138 * config/i386/predicates.md (float_operator): Remove predicate.
9140 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9142 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9143 (*jccxf_i387): Ditto.
9144 (*jcc<mode>_i387): Ditto.
9145 (*jccu<mode>_i387): Ditto.
9146 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9147 (*jcc_*_i387 splitters): Remove.
9148 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9149 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9150 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9153 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9155 * profile-count.h (slow_safe_scale_64bit): New function.
9156 (safe_scale_64bit): New inline.
9157 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9158 * profile-count.c: Include wide-int.h
9159 (slow_safe_scale_64bit): New.
9161 2017-10-11 Nathan Sidwell <nathan@acm.org>
9163 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9164 HAS_DECL_ASSEMBLER_NAME_P.
9165 * gimple-expr.c (gimple_decl_printable_name: Check
9166 HAS_DECL_ASSEMBLER_NAME_P too.
9167 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9168 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9169 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9170 * passes.c (rest_of_decl_compilation): Check
9171 HAS_DECL_ASSEMBLER_NAME_P too.
9172 * recog.c (verify_changes): Likewise.
9173 * tree-pretty-print.c (dump_decl_name): Likewise.
9174 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9176 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9177 (SET_DECL_ASSEMBLER_NAME): Use it.
9178 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9179 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9180 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9182 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9184 * config.gcc (i386, x86_64): Add extra objects.
9185 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9186 (ix86_min_insn_size): Declare.
9187 (ix86_issue_rate): Declare.
9188 (ix86_adjust_cost): Declare.
9189 (ia32_multipass_dfa_lookahead): Declare.
9190 (ix86_macro_fusion_p): Declare.
9191 (ix86_macro_fusion_pair_p): Declare.
9192 (ix86_bd_has_dispatch): Declare.
9193 (ix86_bd_do_dispatch): Declare.
9194 (ix86_core2i7_init_hooks): Declare.
9195 (ix86_atom_sched_reorder): Declare.
9196 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9197 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9198 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9199 (rip_relative_addr_p): Rename to ...
9200 (ix86_rip_relative_addr_p): ... this one; export.
9201 (memory_address_length): Update.
9202 (ix86_issue_rate): Move to x86-tune-sched.c.
9203 (ix86_flags_dependent): Move to x86-tune-sched.c.
9204 (ix86_agi_dependent): Move to x86-tune-sched.c.
9205 (exact_dependency_1): Move to x86-tune-sched.c.
9206 (exact_store_load_dependency): Move to x86-tune-sched.c.
9207 (ix86_adjust_cost): Move to x86-tune-sched.c.
9208 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9209 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9210 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9211 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9212 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9213 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9214 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9215 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9216 (min_insn_size): Rename to ...
9217 (ix86_min_insn_size): ... this one; export.
9218 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9219 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9220 (core2i7_first_cycle_multipass_backtrack): Move to
9221 x86-tune-sched-core.c.
9222 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9223 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9224 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9225 (ix86_avoid_jump_mispredicts): Update.
9226 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9227 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9228 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9229 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9230 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9231 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9232 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9233 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9234 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9235 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9236 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9237 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9238 (BIG): Move to ix86-tune-sched-bd.c.
9239 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9240 (enum insn_path): Move to ix86-tune-sched-bd.c.
9241 (get_mem_group): Move to ix86-tune-sched-bd.c.
9242 (is_cmp): Move to ix86-tune-sched-bd.c.
9243 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9244 (is_branch): Move to ix86-tune-sched-bd.c.
9245 (is_prefetch): Move to ix86-tune-sched-bd.c.
9246 (init_window): Move to ix86-tune-sched-bd.c.
9247 (allocate_window): Move to ix86-tune-sched-bd.c.
9248 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9249 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9250 (process_end_window): Move to ix86-tune-sched-bd.c.
9251 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9252 (find_constant): Move to ix86-tune-sched-bd.c.
9253 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9254 (has_immediate): Move to ix86-tune-sched-bd.c.
9255 (get_insn_path): Move to ix86-tune-sched-bd.c.
9256 (get_insn_group): Move to ix86-tune-sched-bd.c.
9257 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9258 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9259 (add_insn_window): Move to ix86-tune-sched-bd.c.
9260 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9261 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9262 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9263 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9264 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9265 (do_dispatch): Move to ix86-tune-sched-bd.c.
9266 (has_dispatch): Move to ix86-tune-sched-bd.c.
9267 * config/i386/t-i386: Add new object files.
9268 * config/i386/x86-tune-costs.h: New file.
9269 * config/i386/x86-tune-sched-atom.c: New file.
9270 * config/i386/x86-tune-sched-bd.c: New file.
9271 * config/i386/x86-tune-sched-core.c: New file.
9272 * config/i386/x86-tune-sched.c: New file.
9274 2017-10-11 Liu Hao <lh_mouse@126.com>
9276 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9277 the generic version below instead.
9278 (should_colorize): Recognize Windows consoles as terminals
9280 * pretty-print.c [__MINGW32__] (write_all): New function.
9281 [__MINGW32__] (find_esc_head): Likewise.
9282 [__MINGW32__] (find_esc_terminator): Likewise.
9283 [__MINGW32__] (eat_esc_sequence): Likewise.
9284 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9286 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9289 2017-10-11 Richard Biener <rguenther@suse.de>
9291 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9292 Properly call analyze_scalar_evolution with the loop of the stmt.
9294 2017-10-11 Richard Biener <rguenther@suse.de>
9296 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9297 * tree-core.h (tree_base): Add chrec_var union member.
9298 * tree.h (CHREC_VAR): Remove.
9299 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9300 * tree-chrec.h (build_polynomial_chrec): Adjust.
9301 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9302 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9304 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9306 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9307 * match.pd: ... here.
9308 ((T) X == (T) Y): Relax condition.
9310 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9312 PR tree-optimization/82472
9313 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9315 (break_alias_scc_partitions): Update postorder number.
9317 2017-10-11 Martin Liska <mliska@suse.cz>
9320 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9322 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9324 2017-10-11 Martin Liska <mliska@suse.cz>
9326 * print-rtl.c (print_insn): Move declaration of idbuf
9327 to same scope as name.
9329 2017-10-11 Martin Liska <mliska@suse.cz>
9334 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9337 2017-10-11 Martin Liska <mliska@suse.cz>
9340 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9343 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9345 * config.gcc (default_gnu_indirect_function): Default to yes for
9346 arm*-*-linux* with glibc.
9348 2017-10-11 Richard Biener <rguenther@suse.de>
9350 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9351 default-defs and types we do not want to analyze.
9352 (interpret_loop_phi): Replace unreachable code with an assert.
9353 (compute_scalar_evolution_in_loop): Remove and inline ...
9354 (analyze_scalar_evolution_1): ... here, replacing condition with
9355 what makes the intent clearer. Remove handling of cases
9356 get_scalar_evolution now handles.
9358 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9360 PR rtl-optimization/81434
9361 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9362 comment for main loop. In sched_group_found if, also add checks for
9363 pass and min_cost_group.
9365 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9367 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9368 (rs6000_insn_cost): New function.
9369 * config/rs6000/rs6000.md (cost): New attribute.
9371 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9372 H.J. Lu <hongjiu.lu@intel.com>
9376 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9377 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9378 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9379 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9380 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9381 OPTION_MASK_ISA_MMX and some other bit set require both
9382 mmx and the ISAs without the mmx bit.
9383 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9384 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9385 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9386 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9387 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9388 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9389 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9390 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9391 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9392 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9393 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9394 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9395 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9396 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9397 Add OPTION_MASK_ISA_MMX.
9399 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9401 * config.gcc (armv7*-*-freebsd*): New target.
9402 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9404 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9406 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9407 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9410 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9412 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9413 a template parameter.
9414 (WIDE_INT_REF_FOR): Update accordingly.
9415 * tree.h (wi::int_traits <const_tree>): Delete.
9416 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9417 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9418 (wi::tree_to_wide_ref): New typedef.
9419 (wi::to_wide): New function.
9420 * calls.c (get_size_range): Use wi::to_wide when operating on
9422 * cgraph.c (cgraph_node::create_thunk): Likewise.
9423 * config/i386/i386.c (ix86_data_alignment): Likewise.
9424 (ix86_local_alignment): Likewise.
9425 * dbxout.c (stabstr_O): Likewise.
9426 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9427 * expr.c (const_vector_from_tree): Likewise.
9428 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9429 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9430 (fold_negate_expr_1, int_const_binop_1, const_binop)
9431 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9432 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9433 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9434 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9435 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9436 (fold_not_const, round_up_loc): Likewise.
9437 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9438 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9439 (alloca_call_type): Likewise.
9440 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9441 * godump.c (go_output_typedef): Likewise.
9442 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9443 * internal-fn.c (get_min_precision): Likewise.
9444 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9445 * ipa-polymorphic-call.c
9446 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9447 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9448 (ipa_modify_call_arguments): Likewise.
9449 * match.pd: Likewise.
9450 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9451 * print-tree.c (print_node_brief, print_node): Likewise.
9452 * stmt.c (expand_case): Likewise.
9453 * stor-layout.c (layout_type): Likewise.
9454 * tree-affine.c (tree_to_aff_combination): Likewise.
9455 * tree-cfg.c (group_case_labels_stmt): Likewise.
9456 * tree-data-ref.c (dr_analyze_indices): Likewise.
9457 (prune_runtime_alias_test_list): Likewise.
9458 * tree-dump.c (dequeue_and_dump): Likewise.
9459 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9460 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9461 * tree-pretty-print.c (dump_generic_node): Likewise.
9462 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9463 (simple_iv_with_niters): Likewise.
9464 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9465 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9466 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9467 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9468 (refine_value_range_using_guard, number_of_iterations_ne_max)
9469 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9470 (get_cst_init_from_scev, record_nonwrapping_iv)
9471 (scev_var_range_cant_overflow): Likewise.
9472 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9473 * tree-ssa-pre.c (compute_avail): Likewise.
9474 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9475 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9476 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9477 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9478 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9479 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9480 (dump_case_nodes, try_switch_expansion): Likewise.
9481 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9482 (vect_do_peeling): Likewise.
9483 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9484 * tree-vect-stmts.c (vectorizable_load): Likewise.
9485 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9486 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9487 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9488 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9489 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9490 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9491 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9492 (evrp_dom_walker::before_dom_children): Likewise.
9493 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9494 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9495 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9496 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9497 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9498 (get_range_pos_neg): Likewise.
9499 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9500 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9501 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9502 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9503 * config/avr/avr.c (avr_fold_builtin): Likewise.
9504 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9505 * config/msp430/msp430.c (msp430_attr): Likewise.
9506 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9507 * config/powerpcspe/powerpcspe-c.c
9508 (altivec_resolve_overloaded_builtin): Likewise.
9509 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9510 (rs6000_expand_ternop_builtin): Likewise.
9511 * config/rs6000/rs6000-c.c
9512 (altivec_resolve_overloaded_builtin): Likewise.
9513 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9514 (rs6000_expand_ternop_builtin): Likewise.
9515 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9517 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9519 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9520 when copying loop nest with only one inner loop.
9522 2017-10-10 Richard Biener <rguenther@suse.de>
9524 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9525 blocks if SCEV is active.
9526 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9528 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9529 (scev_initialize): Assert we are not yet initialized.
9531 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9533 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9534 inner loop's exit stmt by making it always exit the loop, otherwise
9535 we would generate an infinite empty loop.
9537 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9539 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9540 renaming variables in new preheader if it's deleted.
9542 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9544 * tree-loop-distribution.c (struct partition): Remove unused field
9545 loops of the structure.
9546 (partition_alloc, partition_free): Ditto.
9547 (build_rdg_partition_for_vertex): Ditto.
9549 2017-10-09 Jeff Law <law@redhat.com>
9551 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9552 return type to match prototype and documentation.
9554 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9556 * config/rs6000/rs6000.c (processor_costs): Move to ...
9557 * config/rs6000/rs6000.h: ... here.
9558 (rs6000_cost): Declare.
9560 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9562 * except.c (setjmp_fn): New global variable.
9563 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9564 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9565 if DONT_USE_BUILTIN_SETJMP is defined.
9567 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9569 * target.def (insn_cost): New hook.
9570 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9571 * doc/tm.texi: Regenerate.
9572 * rtlanal.c (insn_cost): Use the new hook.
9574 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9576 * combine.c (combine_validate_cost): Compute the new insn_cost,
9577 not just pattern_cost.
9578 (try_combine): Adjust comment.
9580 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9582 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9584 * combine.c (uid_insn_cost): Adjust comment.
9585 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9587 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9588 * dse.c (find_shift_sequence): Ditto.
9589 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9590 (bb_valid_for_noce_process_p): Use pattern_cost.
9591 * rtl.h (insn_rtx_cost): Delete.
9592 (pattern_cost): New prototype.
9593 (insn_cost): New prototype.
9594 * rtlanal.c (insn_rtx_cost): Rename to...
9595 (pattern_cost): ... this.
9598 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9600 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9601 (*jcc<mode>_0_r_i387): Ditto.
9602 (*jccxf_r_i387): Ditto.
9603 (*jcc<mode>_r_i387): Ditto.
9604 (*jccu<mode>_r_i387): Ditto.
9605 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9606 (*jcc): Rename from *jcc_1.
9608 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9610 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9611 deferred rescans after the lvx/stvx recombination pre-pass.
9613 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9615 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9616 memory operation instruction support.
9617 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9618 (rs6000-ibm-aix[789]*): Likewise.
9619 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9620 Document new functions.
9622 2017-10-09 Richard Biener <rguenther@suse.de>
9624 PR tree-optimization/82397
9625 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9626 equality only for semantically equal trees.
9628 2017-10-09 Richard Biener <rguenther@suse.de>
9630 PR tree-optimization/82449
9631 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9632 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9633 allow constant addresses.
9634 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9637 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9639 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9642 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9645 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9648 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9651 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9653 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9656 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9657 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9659 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9661 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9662 (WI_BINARY_PREDICATE_RESULT): Likewise.
9663 (wi::binary_traits::operator_result): New type.
9664 (wi::binary_traits::predicate_result): Likewise.
9665 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9666 (generic_wide_int::operator==, generic_wide_int::operator!=)
9667 (generic_wide_int::operator&, generic_wide_int::and_not)
9668 (generic_wide_int::operator|, generic_wide_int::or_not)
9669 (generic_wide_int::operator^, generic_wide_int::operator+
9670 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9672 (operator~, unary operator-, operator==, operator!=, operator&)
9673 (operator|, operator^, operator+, binary operator-, operator*): New
9675 * expr.c (get_inner_reference): Use wi::bit_and_not.
9676 * fold-const.c (fold_binary_loc): Likewise.
9677 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9678 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9679 (bit_value_binop): Likewise.
9680 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9681 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9682 (extract_range_from_binary_expr_1): Likewise.
9683 (masked_increment): Likewise.
9684 (simplify_bit_ops_using_ranges): Likewise.
9686 2017-10-09 Martin Jambor <mjambor@suse.cz>
9689 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9690 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9691 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9693 (hsa_fixup_mov_insn_type): New function.
9694 (hsa_op_with_type::get_in_type): Use it.
9695 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9696 immediates in an assert.
9697 (hsa_op_with_type::extend_int_to_32bit): New method.
9698 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9699 types. Convert to dest type if necessary.
9700 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9701 (reg_for_gimple_ssa): Pass false as min32int to
9702 hsa_type_for_scalar_tree_type.
9703 (gen_hsa_addr): Fixup type when creating addresable temporary.
9704 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9705 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9706 necessary. Call hsa_fixup_mov_insn_type.
9707 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9708 extend operands and convert to dest type if necessary.
9709 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9710 to dest type if necessary.
9711 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9712 if conversion nt necessary and size matches.
9713 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9714 to dest type if necessary.
9715 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9716 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9718 (gen_hsa_clrsb): Likewise.
9719 (gen_hsa_ffs): Likewise.
9720 (gen_hsa_divmod): Extend operands and convert to dest type if
9722 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9724 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9726 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9727 Remove empty default arguments. Use a brace block as output
9729 (conditional return): Ditto.
9731 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9732 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9734 (group_ending_nop): Ditto.
9735 (doloop_end): Ditto.
9736 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9737 (splitters for those): Ditto.
9739 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9741 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9742 a conditional jump (and the compare for it) so that pc_rtx is the
9744 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9745 for the deleted and renamed ctr<mode>_internal[234] patterns.
9746 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9747 Delete second conditional return pattern.
9748 (ctr<mode>_internal2): Delete this second bdnz pattern.
9749 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9750 (ctr<mode>_internal4): Delete this second bdz pattern.
9752 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9754 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9755 (always_initialized_rtx_for_ssa_name_p): New predicate.
9756 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9757 (finish_out_of_ssa): Free new field of SA.
9758 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9759 * tree-ssa-coalesce.c: Include tree-ssa.h.
9760 (get_parm_default_def_partitions): Remove extern keyword.
9761 (get_undefined_value_partitions): New function.
9762 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9763 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9765 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9767 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9769 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9771 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9773 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9774 for targets that preffer 128bit.
9776 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9778 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9780 2017-10-08 Olivier Hainque <hainque@adacore.com>
9782 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9783 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9785 (thumb_set_return_address): Likewise.
9787 2017-10-08 Olivier Hainque <hainque@adacore.com>
9789 * common/config/arm/arm-common.c (arm_except_unwind_info):
9790 Handle DWARF2_UNWIND_INFO.
9792 2017-10-07 Michael Collison <michael.collison@arm.com>
9794 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9797 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9799 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9800 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9801 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9802 defined, force the creation of a new block for a dispatch label.
9804 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9806 * invoke.texi (Wsuggest-attribute=cold): Document.
9807 * common.opt (Wsuggest-attribute=cold): New
9808 * ipa-pure-const.c (warn_function_cold): New function.
9809 * predict.c (compute_function_frequency): Use it.
9810 * predict.h (warn_function_cold): Declare.
9812 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9814 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9817 2017-10-06 Martin Liska <mliska@suse.cz>
9819 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9820 keyword for member functions.
9821 (struct sanopt_tree_couple): New struct.
9822 (struct sanopt_tree_couple_hash): New function.
9823 (struct sanopt_ctx): Add new hash_map.
9824 (has_dominating_ubsan_ptr_check): New function.
9825 (record_ubsan_ptr_check_stmt): Likewise.
9826 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9827 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9828 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9830 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9833 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9834 aarch64_simd_valid_immediate on CONST_VECTORs.
9835 (aarch64_reg_or_bic_imm): Likewise.
9837 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9839 PR rtl-optimization/82396
9840 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9842 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9844 * graphite-dependences.c (scop_get_reads): Move code to...
9845 (scop_get_must_writes): Move code to...
9846 (scop_get_may_writes): Move code to...
9847 (scop_get_reads_and_writes): ... here.
9848 (scop_get_dependences): Call scop_get_reads_and_writes.
9850 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9852 PR tree-optimization/82434
9853 * fold-const.h (can_native_encode_type_p,
9854 can_native_encode_string_p): Remove.
9855 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9856 don't encode anything, just return what would be otherwise returned.
9857 (native_encode_fixed, native_encode_complex, native_encode_vector):
9859 (native_encode_string): Likewise. Inline by hand
9860 can_native_encode_string_p.
9861 (can_native_encode_type_p): Remove.
9862 (can_native_encode_string_p): Remove.
9863 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9864 STRING_CSTs using can_native_encode_string_p, test all
9865 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9866 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9867 argument from native_encode_expr.
9868 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9869 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9872 2017-10-06 Richard Biener <rguenther@suse.de>
9874 PR tree-optimization/82397
9875 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9876 operand_equal_p but rely on data_ref_compare_tree for detecting
9878 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9879 to match up with dr_group_sort_cmp.
9881 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9884 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9886 * config/s390/s390-builtin-types.def: Regenerate.
9888 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9891 * config/s390/s390-builtin-types.def: Regenerate.
9892 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9893 Change flag from B_VXE to B_VX.
9894 (s390_vec_min_dbl): Remove B_VXE flag.
9896 2017-10-06 Richard Biener <rguenther@suse.de>
9898 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9899 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9900 translate_isl_ast_to_gimple::is_valid_rename,
9901 translate_isl_ast_to_gimple::get_rename,
9902 translate_isl_ast_to_gimple::get_def_bb_for_const,
9903 translate_isl_ast_to_gimple::get_new_name,
9904 translate_isl_ast_to_gimple::collect_all_ssa_names,
9905 translate_isl_ast_to_gimple::copy_loop_phi_args,
9906 translate_isl_ast_to_gimple::collect_all_ssa_names,
9907 translate_isl_ast_to_gimple::copy_loop_phi_args,
9908 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9909 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9910 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9911 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9912 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9913 translate_isl_ast_to_gimple::copy_cond_phi_args,
9914 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9915 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9916 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9917 translate_isl_ast_to_gimple::rename_uses,
9918 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9919 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9920 (set_rename_for_each_def): Likewise.
9921 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9922 here. Handle rewriting SCEV analyzable uses here.
9923 (copy_bb_and_scalar_dependences): Generate code for PHI
9925 (graphite_regenerate_ast_isl): Adjust.
9926 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9927 (add_write, add_read): New functions.
9928 (build_cross_bb_scalars_def): Use it and simplify.
9929 (build_cross_bb_scalars_use): Likewise.
9930 (graphite_find_cross_bb_scalar_vars): Inline into...
9931 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9932 simulating out-of-SSA. Compute liveout and add dependencies.
9933 (build_scops): Force an empty entry block.
9934 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9936 (sese_build_liveouts): Declare.
9937 (sese_trivially_empty_bb_p): Likewise.
9938 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9939 compute liveout and debug_liveout.
9940 (sese_bad_liveouts_use): Remove.
9941 (sese_reset_debug_liveouts_bb): Likewise.
9942 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9943 (sese_build_liveouts): Build liveout and debug_liveout and store
9945 (new_sese_info): Adjust.
9946 (free_sese_info): Likewise.
9947 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9948 do not build liveout here.
9949 (move_sese_in_condition): Adjust region entry.
9950 (scev_analyzable_p): Match up with chrec_apply requirements.
9951 (sese_trivially_empty_bb_p): New.
9952 * tree-into-ssa.c (get_reaching_def): Properly support generating
9953 default-defs for incremental rewrite of anonymous names.
9955 2017-10-06 Richard Biener <rguenther@suse.de>
9957 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9958 precision do not perform modulo reduction.
9960 2017-10-06 Richard Biener <rguenther@suse.de>
9962 PR tree-optimization/82436
9963 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9964 conservatively choose the vectorization factor when checking
9965 whether we can perform the required load permutation.
9966 (vect_transform_slp_perm_load): Assert when we may not fail.
9968 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9970 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9971 message for incompatible -msdata=* and -mcall-* options.
9973 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9975 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9976 rate for post-reload scheduling.
9978 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9980 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9982 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9984 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9985 to improve monte carlo in scimark.
9987 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9989 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9990 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9991 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9992 pentium4_cost, nocona_cost): Set reassociation width to 1.
9993 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9994 width to 2 for fp operations and 1 otherwise.
9995 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9997 (atom_cost): Set reassociation width to 2.
9998 (slm_cost, generic_cost): Set fp reassociation width
9999 to 2 and 1 otherwise.
10000 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10001 (core_cost): Set fp reassociation width to 4 and vector to 2.
10002 (ix86_reassociation_width): Rewrite using cost table; special case
10003 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10004 and TARGET_AVX128_OPTIMAL.
10005 * config/i386/i386.h (processor_costs): Add
10006 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10007 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10008 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10009 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10010 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10011 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10013 2017-10-05 Nathan Sidwell <nathan@acm.org>
10015 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10017 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10019 * config/arm/arm.c (arm_test_fpu_data): New.
10020 (arm_run_selftests): Call arm_test_fpu_data.
10022 2017-10-04 Nathan Sidwell <nathan@acm.org>
10024 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10026 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10028 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10030 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10031 check type for aarch64_simd_valid_immediate.
10032 (aarch64_output_simd_mov_immediate): Update prototype.
10033 (aarch64_simd_valid_immediate): Update prototype.
10034 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10035 support for ORR-immediate.
10036 (and<mode>3): modified pattern to add support for BIC-immediate.
10037 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10038 now checks for valid immediate for BIC and ORR based on new enum
10040 (aarch64_output_simd_mov_immediate): Function now used to output
10041 BIC/ORR imm as well based on new enum argument.
10042 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10044 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10045 (aarch64_reg_or_bic_imm): Likewise.
10047 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10049 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10050 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10052 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10056 PR rtl-optimization/82396
10057 * haifa-sched.c (autopref_multipass_init): Simplify
10059 (autopref_rank_data): Simplify sort order.
10060 * sched-int.h (autopref_multipass_data_): Remove
10061 multi_mem_insn_p, min_offset and max_offset.
10063 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10065 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10067 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10069 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10070 vect_doubleint_cvt.
10072 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10074 * doc/sourcebuild.texi: Document vect_long_mult.
10076 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10078 PR tree-optimization/82413
10079 * fold-const.c (build_range_check): Use widest_int when comparing
10080 the maximum ETYPE value with HIGH.
10082 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10084 PR rtl-optimization/82396
10085 * haifa-sched.c (autopref_multipass_init): Simplify
10087 (autopref_rank_data): Simplify sort order.
10088 * sched-int.h (autopref_multipass_data_): Remove
10089 multi_mem_insn_p, min_offset and max_offset.
10091 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10093 PR tree-optimization/82381
10094 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10095 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10098 PR tree-optimization/82374
10099 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10100 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10101 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10102 current_function_decl to the new decl.
10104 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10106 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10107 helper macro for IEEE float128 hardware built-in functions.
10108 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10110 (TRUNCF128_ODD): Likewise.
10111 (ADDF128_ODD): Likewise.
10112 (SUBF128_ODD): Likewise.
10113 (MULF128_ODD): Likewise.
10114 (DIVF128_ODD): Likewise.
10115 (FMAF128_ODD): Likewise.
10116 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10117 UNSPEC_TRUNC_ROUND_TO_ODD.
10118 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10119 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10120 floating point round to odd instructions.
10121 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10122 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10123 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10124 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10125 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10126 (trunc<mode>sf2_hw): Change the truncate with round to odd
10127 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10128 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10129 to odd hardware instructions.
10130 (sub<mode>3_odd): Likewise.
10131 (mul<mode>3_odd): Likewise.
10132 (div<mode>3_odd): Likewise.
10133 (sqrt<mode>2_odd): Likewise.
10134 (fma<mode>4_odd): Likewise.
10135 (fms<mode>4_odd): Likewise.
10136 (nfma<mode>4_odd): Likewise.
10137 (nfms<mode>4_odd): Likewise.
10138 (trunc<mode>df2_odd): Change the truncate with round to odd
10139 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10141 * doc/extend.texi (PowerPC built-in functions): Update documentation
10142 for existing IEEE float128-bit built-in functions. Add built-in
10143 functions that generate the IEEE 128-bit floating point round to
10146 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10148 PR rtl-optimization/77729
10149 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10150 to (X&(C1&~C2))|C2 transformations.
10152 2017-10-03 Martin Jambor <mjambor@suse.cz>
10154 PR tree-optimization/82363
10155 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10156 mismatch, mark lacc written regardless of racc.
10158 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10160 PR tree-optimization/82381
10161 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10162 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10163 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10164 fallthrough into reassoc_stmt_dominates_stmt_p.
10167 * combine.c (combine_instructions): Don't combine in unreachable
10170 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10173 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10174 function to not use the have_cpu variable. Do not set cpu_index,
10175 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10176 or the default cpu.
10177 (rs6000_valid_attribute_p): Remove duplicate initializations of
10178 old_optimize and func_optimize.
10179 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10180 (rs6000_activate_target_options): Make global.
10181 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10184 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10186 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10187 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10188 Set *poffset to 0 and *psize and *pmax_size to -1 if
10189 *poffset + *psize overflows in HOST_WIDE_INT.
10191 PR tree-optimization/82387
10192 PR tree-optimization/82388
10193 PR tree-optimization/82389
10194 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10195 instead of live_bytes non-NULL.
10197 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10200 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10203 2017-10-02 Richard Biener <rguenther@suse.de>
10205 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10206 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10207 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10209 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10211 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10212 requested precision matches the type's.
10213 * calls.c (alloc_max_size): Calculate the new candidate size as
10214 a widest_int and use wi::to_widest when comparing it with the
10215 current candidate size.
10216 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10217 zero rather than integer_zero_node.
10218 * match.pd: Check for a no-op conversion before using wi::add
10219 rather than after. Use tree_to_uhwi when summing small shift
10220 counts into an unsigned int.
10222 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10223 Alan Hayward <alan.hayward@arm.com>
10224 David Sherwood <david.sherwood@arm.com>
10227 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10228 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10229 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10230 POINTER_AND_FP_REGS.
10232 2017-10-02 Richard Biener <rguenther@suse.de>
10234 PR tree-optimization/82355
10235 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10236 a mapping for the enclosing loop but avoid generating one for
10237 the loop tree root.
10238 (copy_bb_and_scalar_dependences): Remove premature codegen
10239 error on PHIs in blocks duplicated into multiple places.
10240 * graphite-scop-detection.c
10241 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10242 in the region use it as loop and nest to analyze the DR in.
10243 (try_generate_gimple_bb): Likewise.
10244 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10245 (add_loop_constraints): For blocks in a loop not in the region
10246 create a dimension with a single iteration.
10247 * sese.h (gbb_loop_at_index): Remove assert.
10249 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10251 * omp-expand.c (adjust_context_scope): New function.
10252 (expand_parallel_call): Call adjust_context_scope.
10254 2017-10-01 Jeff Law <law@redhat.com>
10256 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10257 dom_opt_dom_walker class with direct access to private members.
10258 Add comments. Call test_for_singularity.
10259 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10260 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10261 m_dummy_cond anymore.
10262 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10264 (pass_dominator:execute): Build the dummy_cond here and pass it
10265 to the dom_opt_dom_walker ctor.
10266 (test_for_singularity): New function.
10268 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10269 Maya Rashish <coypu@sdf.org>
10271 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10272 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10273 (alpha*-*-netbsd*) Use nbsd_tm_file.
10274 (arm*-*-netbsdelf*) Likewise.
10275 (i[34567]86-*-netbsdelf*) Likewise.
10276 (x86_64-*-netbsd*) Likewise.
10277 (mips*-*-netbsd*) Likewise.
10278 (powerpc-*-netbsd*) Likewise.
10279 (sh*-*-netbsd*) Likewise.
10280 (sparc-*-netbsdelf*) Likewise.
10281 (sparc64-*-netbsd*) Likewise.
10282 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10284 (vax-*-netbsdelf*) Likewise.
10285 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10286 (UINT_FAST8_TYPE) Likewise.
10287 (INT_FAST16_TYPE) Check CHAR_FAST16.
10288 (UINT_FAST16_TYPE) Likewise.
10290 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10293 * config/i386/i386.md
10294 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10295 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10296 *divmodsi4_zext_2): New define_insn_and_split.
10297 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10298 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10299 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10300 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10301 New define_insn_and_split.
10302 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10303 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10304 operands[1] having DImode when mode is SImode.
10306 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10307 always SImode for DIV and MOD in REG_EQUAL notes.
10309 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10311 PR middle-end/82319
10312 * match.pd: Fix handling of NaNs in pattern.
10314 2017-09-29 Jeff Law <law@redhat.com>
10316 * sbitmap.c (bitmap_bit_in_range_p): New function.
10317 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10318 * tree-ssa-dse.c (live_bytes_read): New function.
10319 (dse_classify_store): Ignore reads of dead bytes.
10321 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10322 typos and whitespace errors.
10323 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10324 * config/s390/s390.c (s390_emit_prologue): Likewise.
10326 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10329 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10330 with a symbol for LRA.
10332 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10334 PR rtl-optimization/82338
10335 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10337 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10339 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10340 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10341 (qsort_chk): Declare.
10342 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10343 (qsort_chk): New function.
10345 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10347 PR tree-optimization/82337
10348 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10349 phi definition if the PHI result appears in an abnormal PHI.
10350 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10351 basis appears in an abnormal PHI.
10353 2017-09-29 Richard Biener <rguenther@suse.de>
10355 * graphite-isl-ast-to-gimple.c
10356 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10357 (binary_op_to_tree): Use it.
10358 (get_rename_from_scev): Likewise.
10359 (copy_loop_phi_nodes): Likewise.
10360 (copy_bb_and_scalar_dependences): Likewise.
10361 (translate_pending_phi_nodes): Likewise.
10363 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10366 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10367 for movabsq $(i32 << shift), r64.
10369 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10371 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10372 index when encoding %esp as %rsp to avoid 0x67 prefix.
10374 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10376 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10377 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10379 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10381 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10382 Extensions with more than 16 double VFP registers.
10383 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10384 to_clear_mask and all code related to it. Replace the remaining
10385 entry by a sbitmap and adapt code accordingly.
10387 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10389 * brig-builtins.def: Change pure attributes to const.
10391 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10393 * config.gcc (default_gnu_indirect_function): Default to yes for
10394 sparc*-*-linux* with glibc.
10396 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10398 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10399 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10400 when creating .init_array and .fini_array sections with priority
10403 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10406 * config/aarch64/aarch64.c
10407 (aarch64_builtin_support_vector_misalignment): Always return false
10408 when misalignment is unknown.
10410 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10412 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10413 this function to return false if the definition used by the swap
10414 instruction is artificial, or if the memory address from which the
10415 constant value is loaded is not represented by a base address held
10416 in a register or if the base address register is a frame or stack
10417 pointer. Additionally, return false if the base address of the
10418 loaded constant is a SYMBOL_REF but is not considered to be a
10420 (replace_swapped_load_constant): New function.
10421 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10422 loaded constant vector with a load of a swapped constant vector.
10424 2017-09-27 Carl Love <cel@us.ibm.com>
10426 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10427 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10428 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10431 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10433 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10434 first, always call autopref_rank_data otherwise.
10436 2017-09-27 Richard Biener <rguenther@suse.de>
10438 * graphite-scop-detection.c (find_scop_parameters): Move
10439 loop bound handling ...
10440 (gather_bbs::before_dom_children): ... here, avoiding the need
10441 to build scop_info->loop_nest.
10442 (record_loop_in_sese): Remove.
10443 * sese.h (sese_info_t::loop_nest): Remove.
10444 * sese.c (new_sese_info): Do not allocate loop_nest.
10445 (free_sese_info): Do not free loop_nest.
10447 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10450 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10451 lhs from calls if the lhs has addressable type.
10453 2017-09-27 Richard Biener <rguenther@suse.de>
10455 * graphite.h (scop::max_alias_set): New member.
10456 * graphite-scop-detection.c: Remove references to non-existing
10457 --param in comments.
10458 (build_alias_sets): Record the maximum alias set used for drs.
10459 (build_scops): Support zero as unlimited for
10460 --param graphite-max-arrays-per-scop.
10461 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10462 and inline into ...
10463 (build_poly_sr_1): ... here. Compute alias set based on the
10464 maximum alias set used for drs rather than
10465 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10467 2017-09-27 Richard Biener <rguenther@suse.de>
10469 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10470 --param loop-block-tile-size=0 to disable tiling.
10472 2017-09-27 Richard Biener <rguenther@suse.de>
10474 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10475 (graphite-max-nb-scop-params): Document special value zero.
10476 * domwalk.h (dom_walker::STOP): New symbolical constant.
10477 (dom_walker::dom_walker): Add optional parameter for bb to
10479 (dom_walker::~dom_walker): Declare.
10480 (dom_walker::before_dom_children): Document STOP return value.
10481 (dom_walker::m_user_bb_to_rpo): New member.
10482 (dom_walker::m_bb_to_rpo): Likewise.
10483 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10484 mapping here if not provided by the user.
10485 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10486 provided by the user.
10487 (dom_walker::STOP): Define.
10488 (dom_walker::walk): Do not compute bb to RPO mapping here.
10489 Support STOP return value from before_dom_children to stop
10491 * graphite-optimize-isl.c (optimize_isl): If the schedule
10492 is the same still generate code if -fgraphite-identity
10493 or -floop-parallelize-all are given.
10494 * graphite-scop-detection.c: Include cfganal.h.
10495 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10497 (gather_bbs::before_dom_children): Return STOP for BBs
10499 (build_scops): Compute bb to RPO mapping and pass it to
10500 the domwalk. Treat --param graphite-max-nb-scop-params=0
10501 as not limiting the number of params.
10502 * graphite.c (graphite_initialize): Remove limit on the
10503 number of basic-blocks in a function.
10504 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10505 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10506 default value of 10.
10508 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10510 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10511 Adjust code to eliminate needing to do the shift right 32-bits
10512 operation after XSCVDPSPN.
10514 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10516 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10517 ((X / Y) != 0 -> X >= Y): Likewise.
10519 2017-09-26 Carl Love <cel@us.ibm.com>
10521 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10522 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10523 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10524 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10525 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10526 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10527 definitions and overloading.
10528 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10529 statement for P9V_BUILTIN_XST_LEN_R.
10530 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10531 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10532 define_expand and define_insn for the instructions and builtins.
10533 * doc/extend.texi: Update the built-in documentation file for the new
10534 built-in functions.
10535 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10536 define_insn for the instructions
10538 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10541 * gcc/config/netbsd-protos.h: New file.
10542 * gcc/config/netbsd.c: New file.
10543 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10544 * gcc/config/t-netbsd: New file.
10545 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10546 (tmake_file) Add t-netbsd.
10547 (extra_objs) Add netbsd.o.
10549 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10553 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10555 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10557 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10558 sign extension from a vector register to a GPR by doing a 32-bit
10559 direct move and then an EXTSW.
10560 (extendsi<mode>2 splitter): Likewise.
10561 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10562 right or vector extract after doing XSCVDPSPN. Use
10563 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10565 (movdi_from_sf_zero_ext): Likewise.
10566 (reload_gpr_from_vsxsf): Likewise.
10567 (p8_mfvsrd_4_disf): Delete, no longer used.
10568 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10569 then needing to move the SFmode to a GPR to use the XSCVDPSP
10570 instruction instead of FRSP and XSCVDPSPN.
10571 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10572 it is adjacent to the other XSCVSPDP insns.
10573 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10574 SFmode to be in traditional Altivec registers.
10575 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10576 (vsx_xscvspdpn): Likewise.
10577 (vsx_xscvspdpn_scalar): Likewise.
10579 2017-09-26 Martin Jambor <mjambor@suse.cz>
10581 * tree-sra.c (compare_access_positions): Put integral types first,
10582 stabilize sorting of integral types, remove conditions putting
10583 non-full-precision integers last.
10584 (sort_and_splice_var_accesses): Disable scalarization if a
10585 non-integert would be represented by a non-full-precision integer.
10587 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10589 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10590 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10591 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10592 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10593 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10594 conditionals inside the function instead of around it. Call
10595 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10596 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10598 2017-09-26 Richard Biener <rguenther@suse.de>
10600 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10602 (scop_detection::build_scop_breadth): ... this. Removed.
10603 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10604 (scop_detection::harmful_stmt_in_bb): Likewise.
10605 (scop_detection::graphite_can_represent_stmt): Likewise.
10606 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10607 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10608 (scop_detection::can_represent_loop_1): ... this. Removed.
10609 (scop_detection::harmful_loop_in_region): Simplify after inlining
10610 the above and remove more quadraticness.
10611 (build_scops): Adjust.
10612 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10615 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10618 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10619 REGNO (base) == SP_REG if base is a REG.
10621 PR middle-end/35691
10622 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10623 if it is different SSA_NAME.
10624 (optimize_range_tests_cmp_bitwise): New function.
10625 (optimize_range_tests): Call it.
10627 2017-09-26 Richard Biener <rguenther@suse.de>
10629 PR tree-optimization/82321
10630 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10631 for the def being inside the loop.
10633 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10635 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10637 * config/s390/s390-builtins.def: Fix constraint on op4.
10639 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10641 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10642 independent expanders.
10643 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10644 ("vec_ordered", "vec_unordered"): New expanders.
10646 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10648 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10651 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10653 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10654 vec_unpacks_lo_v16qi.
10655 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10657 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10659 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10660 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10661 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10663 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10665 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10667 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10669 ("*vec_slb<mode>"): New insn pattern.
10670 ("vec_shr_<mode>"): New expander.
10671 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10672 and force the shift count operand to V16QImode.
10673 ("vec_srb<mode>"): Set shift count mode to V16QI.
10675 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10677 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10678 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10679 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10681 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10684 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10686 2017-09-26 Richard Biener <rguenther@suse.de>
10688 PR tree-optimization/82320
10689 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10692 2017-09-25 Jeff Law <law@redhat.com>
10694 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10695 prototype for new argument.
10696 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10697 mostly extracted from rs6000_emit_allocate_stack.
10698 (rs6000_emit_probe_stack_range_stack_clash): New function.
10699 (rs6000_emit_allocate_stack): Call
10700 rs6000_emit_probe_stack_range_stack_clash as needed.
10701 (rs6000_emit_probe_stack_range): Add additional argument
10702 to call to gen_probe_stack_range{si,di}.
10703 (output_probe_stack_range): New.
10704 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10705 (output_probe_stack_range_stack_clash): New.
10706 (rs6000_emit_prologue): Emit notes into dump file as requested.
10707 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10708 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10709 Add additional operand and pass it to output_probe_stack_range.
10711 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10713 PR tree-optimization/82163
10714 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10715 (checking_verify_loop_closed_ssa): New parameter.
10716 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10717 (check_loop_closed_ssa_stmt): Delete.
10718 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10719 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10720 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10723 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10725 * brig-builtins.def: Treat HSAIL barrier builtins as
10726 setjmp/longjump style functions.
10728 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10730 * target.def (constant_alignment): New hook.
10731 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10732 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10733 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10734 * doc/tm.texi: Regenerate.
10735 * targhooks.h (default_constant_alignment): Declare.
10736 (constant_alignment_word_strings): Likewise.
10737 * targhooks.c (default_constant_alignment): New function.
10738 (constant_alignment_word_strings): Likewise.
10739 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10740 instead of CONSTANT_ALIGNMENT.
10741 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10742 (force_const_mem): Likewise.
10743 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10744 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10745 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10746 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10747 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10749 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10750 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10751 constant_alignment_word_strings.
10752 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10753 (CONSTANT_ALIGNMENT): Likewise.
10754 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10755 (arm_constant_alignment): New function.
10756 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10757 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10758 constant_alignment_word_strings.
10759 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10760 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10761 constant_alignment_word_strings.
10762 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10763 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10764 (cris_constant_alignment): New function.
10765 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10766 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10767 (epiphany_constant_alignment): New function.
10768 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10769 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10770 constant_alignment_word_strings.
10771 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10772 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10773 constant_alignment_word_strings.
10774 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10775 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10776 constant_alignment_word_strings.
10777 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10778 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10779 * config/i386/i386.c (ix86_constant_alignment): Make static.
10780 Use the same interface as the target hook.
10781 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10782 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10783 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10784 constant_alignment_word_strings.
10785 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10786 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10787 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10788 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10789 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10790 constant_alignment_word_strings.
10791 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10792 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10793 constant_alignment_word_strings.
10794 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10795 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10796 constant_alignment_word_strings.
10797 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10798 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10800 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10801 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10802 * config/mips/mips.c (mips_constant_alignment): New function.
10803 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10804 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10805 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10806 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10807 (mmix_constant_alignment): Make static. Use the same interface
10808 as the target hook.
10809 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10810 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10811 constant_alignment_word_strings.
10812 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10813 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10814 constant_alignment_word_strings.
10815 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10816 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10817 constant_alignment_word_strings.
10818 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10819 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10820 (rs6000_constant_alignment): New function.
10821 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10822 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10823 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10824 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10825 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10826 (rs6000_constant_alignment): New function.
10827 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10828 * config/s390/s390.c (s390_constant_alignment): New function.
10829 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10830 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10831 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10832 constant_alignment_word_strings.
10833 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10834 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10835 (sparc_constant_alignment): New function.
10836 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10837 * config/spu/spu.c (spu_constant_alignment): New function.
10838 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10839 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10840 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10841 constant_alignment_word_strings.
10842 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10843 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10844 constant_alignment_word_strings.
10845 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10846 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10847 constant_alignment_word_strings.
10848 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10849 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10850 (visium_constant_alignment): New function.
10851 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10852 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10853 (xtensa_constant_alignment): New function.
10854 * system.h (CONSTANT_ALIGNMENT): Poison.
10856 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10858 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10859 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10860 (rs6000_builtin_valid_without_lhs): New helper function.
10861 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10862 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10864 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10866 * target.h (vec_perm_indices): Use unsigned short rather than
10868 (auto_vec_perm_indices): Likewise.
10869 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10870 Use unsigned int rather than unsigned char.
10871 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10873 2017-09-25 Richard Biener <rguenther@suse.de>
10875 * cfgloop.h (sort_sibling_loops): Declare.
10876 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10877 (sort_sibling_loops): New function sorting the sibling loop list
10879 * graphite.c (graphite_transform_loops): Sort sibling loops.
10881 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10883 * target.def (vec_perm_const_ok): Change sel parameter to
10885 * optabs-query.c (can_vec_perm_p): Update accordingly.
10886 * doc/tm.texi: Regenerate.
10887 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10888 auto_vec_perm_indices and remove separate nelt field.
10889 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10890 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10891 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10892 (aarch64_expand_vec_perm_const): Update accordingly.
10893 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10894 to vec_perm_indices.
10895 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10896 auto_vec_perm_indices and remove separate nelt field.
10897 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10898 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10899 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10901 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10902 to vec_perm_indices.
10903 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10904 sel to vec_perm_indices.
10905 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10906 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10907 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10909 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10912 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10915 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10916 on the FUNCTION_DECL function context if it has a DIE that is a
10919 2017-09-25 Richard Biener <rguenther@suse.de>
10921 PR tree-optimization/82285
10922 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10925 2017-09-25 Tom de Vries <tom@codesourcery.com>
10929 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10932 2017-09-25 Richard Biener <rguenther@suse.de>
10934 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10935 ISL errors other than isl_error_quota happen. Dump if the
10936 schedule is the same.
10937 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10938 errors instead of aborting inside ISL.
10940 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10943 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10944 of libgcc_eh for m64.
10945 * config/i386/darwin64.h: Likewise.
10947 2017-09-25 Richard Biener <rguenther@suse.de>
10949 PR middle-end/82144
10950 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10951 attribute for incomplete types nor twice for complete ones.
10953 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10956 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10957 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10960 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10963 * config/i386/i386.opt (mprefer-avx256): Use
10964 ix86_target_flags variable.
10965 * config/i386/i386.c (ix86_target_string): Move
10966 -mprefer-avx256 to flag2_opts.
10968 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10970 PR middle-end/35691
10971 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10972 and x != -1 | y != -1 into (x & y) != -1.
10974 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10976 * config.gcc: Add new case statement to set
10977 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10978 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10979 s390x-*-linux* case statements. Added aarch64 to the list of
10980 supported architectures.
10982 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10984 PR tree-optimization/82289
10985 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10986 STMT_VINFO_RELEVANT_P.
10988 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10989 Alan Hayward <alan.hayward@arm.com>
10990 David Sherwood <david.sherwood@arm.com>
10992 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10993 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10994 (extract_range_from_binary_expr_1): Don't call
10995 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10997 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10998 Alan Hayward <alan.hayward@arm.com>
10999 David Sherwood <david.sherwood@arm.com>
11001 * target.def (preferred_vector_alignment): New hook.
11002 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11004 * doc/tm.texi: Regenerate.
11005 * targhooks.h (default_preferred_vector_alignment): Declare.
11006 * targhooks.c (default_preferred_vector_alignment): New function.
11007 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11009 (DR_TARGET_ALIGNMENT): New macro.
11010 (aligned_access_p): Update commentary.
11011 (vect_known_alignment_in_bytes): New function.
11012 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11014 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11015 Calculate the misalignment based on the target alignment rather than
11017 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11018 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11019 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11020 the target alignment, rather than masking the element misalignment
11021 with the number of elements in a vector. Also use the target
11022 alignment when calculating the maximum number of peels.
11023 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11024 instead of TYPE_ALIGN_UNIT.
11025 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11026 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11027 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11028 (vect_create_data_ref_ptr): Likewise.
11029 (vect_setup_realignment): Realign by ANDing with
11030 -DR_TARGET_MISALIGNMENT.
11031 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11032 the number of peels based on DR_TARGET_ALIGNMENT.
11033 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11034 with the guaranteed alignment boundary when deciding whether
11036 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11037 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11038 (ensure_base_align): Remove stmt_info parameter. Get the
11039 target base alignment from DR_TARGET_ALIGNMENT.
11040 (vectorizable_store): Update call accordingly. Interpret
11041 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11043 (vectorizable_load): Likewise.
11045 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11046 Alan Hayward <alan.hayward@arm.com>
11047 David Sherwood <david.sherwood@arm.com>
11049 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11050 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11051 (vect_enhance_data_refs_alignment): Likewise.
11053 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11055 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11056 error. Only quit immediately if parsing is complete.
11057 (BEGIN): Initialize fatal_err and parse_done.
11058 (begin fpu, end fpu): Check number of arguments.
11059 (begin arch, end arch): Likewise.
11060 (begin cpu, end cpu): Likewise.
11061 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11062 (optalias): Likewise.
11064 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11066 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11067 * config/arm/arm-isa.h: Delete. Move definitions to ...
11068 * arm-cpus.in: ... here. Use new feature and fgroup values.
11069 * config/arm/arm.c (arm_option_override): Use lower case for feature
11071 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11072 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11073 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11075 (print_isa_bits_for): New function.
11076 (gen_isa): New function.
11077 (gen_comm_data): Use print_isa_bits_for.
11078 (define feature): New keyword.
11079 (define fgroup): New keyword.
11080 * config/arm/t-arm (TM_H): Remove.
11081 (GTM_H): Add arm-isa.h.
11082 (arm-isa.h): Add rule to generate file.
11083 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11084 case for feature bit names.
11086 2017-09-22 Richard Biener <rguenther@suse.de>
11088 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11090 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11091 print of no dependency loops ...
11092 * graphite.c (graphite_transform_loops): ... here.
11093 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11095 (same_close_phi_node, remove_duplicate_close_phi,
11096 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11097 (canonicalize_loop_closed_ssa): ... here and simplify.
11098 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11099 (optimize_isl): Use dump_printf_loc to tell when we stopped
11100 optimizing because of an ISL timeout.
11102 2017-09-22 Richard Biener <rguenther@suse.de>
11104 PR tree-optimization/82291
11105 * tree-if-conv.c (predicate_mem_writes): Make sure to
11106 remove writes in blocks predicated with false.
11108 2017-09-22 Richard Biener <rguenther@suse.de>
11110 * sese.c: Include cfganal.h.
11111 (if_region_set_false_region): Remove.
11112 (create_if_region_on_edge): Likewise.
11113 (move_sese_in_condition): Re-implement without destroying
11116 2017-09-22 Richard Biener <rguenther@suse.de>
11118 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11119 Verify both BBs contain loop PHI nodes before dispatching to
11120 copy_loop_phi_args.
11121 (graphite_regenerate_ast_isl): Do not recompute dominators,
11122 do not verify three times. Restructure for clarity.
11123 * graphite-scop-detection.c (same_close_phi_node,
11124 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11125 defined_in_loop_p, canonicalize_loop_closed_ssa,
11126 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11127 checking and SSA rewrite, move to ...
11128 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11129 (graphite_initialize): Do not pass in ctx, do not reset the
11130 SCEV cache, compute only dominators.
11131 (graphite_transform_loops): Allocate ISL ctx after
11132 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11133 Maintain post-dominators only around build_scops.
11134 * sese.c (if_region_set_false_region): Make static. Free
11135 and recompute dominators.
11136 (move_sese_in_condition): Assert we don't get called with
11137 post-dominators computed.
11138 * sese.h (if_region_set_false_region): Remove.
11140 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11142 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11143 mode attribute for TARGET_AVX512VL.
11145 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11147 * config/i386/i386.opt (mprefer-avx256): New option.
11148 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11150 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11151 for TARGET_PREFER_AVX256.
11152 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11154 2017-09-21 Jeff Law <law@redhat.com>
11156 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11157 Fix dump output if the only stack space is for pushed registers.
11159 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11161 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11164 2017-09-21 Martin Sebor <msebor@redhat.com>
11167 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11168 code (in C++) or code that triggers warnings.
11170 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11172 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11174 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11176 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11177 * sched-rgn.c: Ditto.
11178 * sel-sched-ir.c: Ditto.
11180 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11182 * toplev.h (set_random_seed): Adjust return type.
11183 * toplev.c (init_local_tick): Move eager initialization of random_seed
11184 to get_random_seed. Adjust comment.
11185 (init_random_seed): Inline to get_random_seed, delete.
11186 (get_random_seed): Initialize random_seed lazily.
11187 (set_random_seed): Do not return previous value.
11188 (print_switch_value): Do not call get_random_seed.
11190 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11192 * cgraph.c (delete_function_version): New, broken out from...
11193 (cgraph_node::delete_function_version): ...here. Rename to
11194 cgraph_node::delete_function_version_by_decl. Update all uses.
11195 (cgraph_node::remove): Call delete_function_version.
11197 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11200 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11201 VAR_DECLs to which addressable non-volatile parameters are mapped
11202 and for id->retvar after the return value assignment. Clear
11203 id->retval and id->retbnd after inlining.
11205 2017-09-21 Richard Biener <rguenther@suse.de>
11207 PR tree-optimization/82276
11208 PR tree-optimization/82244
11209 * tree-vrp.c (build_assert_expr_for): Set
11210 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11212 (remove_range_assertions): Revert earlier change.
11214 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11217 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11219 2017-09-21 Richard Biener <rguenther@suse.de>
11221 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11222 Restore valid IL after code generation errors.
11223 * graphite.c (graphite_transform_loops): Diagnose code
11224 generation issues as MSG_MISSED_OPTIMIZATION and continue
11225 with processing SCOPs.
11227 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11228 Alan Hayward <alan.hayward@arm.com>
11229 David Sherwood <david.sherwood@arm.com>
11231 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11232 rather than choosing between plus_constant and gen_rtx_<CODE>.
11233 * expr.c (emit_push_insn): Likewise.
11234 (expand_expr_real_2): Likewise.
11236 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11237 Alan Hayward <alan.hayward@arm.com>
11238 David Sherwood <david.sherwood@arm.com>
11240 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11242 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11243 Alan Hayward <alan.hayward@arm.com>
11244 David Sherwood <david.sherwood@arm.com>
11246 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11247 calling tree_to_uhwi.
11249 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11250 Alan Hayward <alan.hayward@arm.com>
11251 David Sherwood <david.sherwood@arm.com>
11253 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11254 INTEGER_CST rather than a negative test for ADDR_EXPR.
11256 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11257 Alan Hayward <alan.hayward@arm.com>
11258 David Sherwood <david.sherwood@arm.com>
11260 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11261 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11263 2017-09-21 Richard Biener <rguenther@suse.de>
11265 PR tree-optimization/71351
11266 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11267 graphite_create_new_loop_guard): Remove, fold remaining parts
11269 (translate_isl_ast_node_for): ... here and simplify.
11271 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11274 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11275 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11276 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11277 alternative always use QI mode, for -Os imov (=R,R) alternative
11278 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11281 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11282 Jeff Law <law@redhat.com>
11284 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11285 (allocate_stack_space): New function, partially extracted from
11286 s390_emit_prologue.
11287 (s390_emit_prologue): Track offset to most recent stack probe.
11288 Code to allocate space moved into allocate_stack_space.
11289 Dump actions when no stack is allocated.
11290 (s390_prologue_plus_offset): New function.
11291 (s390_emit_stack_probe): Likewise.
11293 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11295 * common.opt (Wa, Wl, Wp, g, gz=): Add
11297 (gno-column-info): Remove.
11298 (gcolumn-info): Drop RejectNegative.
11299 (gno-): New prefix.
11300 (gno-record-gcc-switches): Remove.
11301 (grecord-gcc-switches): Drop RejectNegative.
11302 (gno-split-dwarf): Remove.
11303 (gsplit-dwarf): Drop RejectNegative.
11304 (gno-strict-dwarf): Remove.
11305 (gstrict-dwarf): Drop RejectNegative.
11306 * config/darwin.opt (gfull, gused): Add RejectNegative.
11307 * dwarf2out.c (gen_producer_string): Drop
11308 gno-record-gcc-switches handler.
11309 * optc-gen.awk: Add g to prefixes with negative forms.
11310 * opts-common.c (remapping_prefix_p): New.
11311 (find_opt): Check it.
11312 (generate_canonical_option): Test g prefix.
11313 (option_map): Add -gno- mapping.
11314 (add_misspelling_candidates): Check remapping_prefix_p.
11316 2017-09-20 Jeff Law <law@redhat.com>
11318 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11319 thinko in stack clash protection support.
11321 * explow.c (compute_stack_clash_protection_loop_data): Use
11322 CONST_INT_P instead of explicit test. Verify object is a
11323 CONST_INT_P before looking at INTVAL.
11324 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11325 instead of explicit test.
11327 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11330 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11331 address instead of to r1 and r11.
11333 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11335 * config.gcc: Support "knm".
11336 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11337 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11339 * config/i386/i386.c (m_KNM): Define.
11340 (processor_target_table): Add "knm".
11342 (ix86_option_override_internal): Add "knm".
11343 (ix86_issue_rate): Add PROCESSOR_KNM.
11344 (ix86_adjust_cost): Ditto.
11345 (ia32_multipass_dfa_lookahead): Ditto.
11346 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11347 (fold_builtin_cpu): Add M_INTEL_KNM.
11348 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11349 (processor_type): Add PROCESSOR_KNM.
11350 * config/i386/x86-tune.def: Add m_KNM.
11351 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11353 2017-09-20 Richard Biener <rguenther@suse.de>
11355 PR tree-optimization/80213
11356 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11357 are allowed in empty BBs as well.
11358 (canonicalize_loop_closed_ssa): Also look for other complex
11360 (scop_detection::get_sese): Include the loop-closed PHI block
11362 (scop_detection::merge_sese): Remove code adding extra blocks.
11363 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11364 (build_scops): Assert the final returned scop is invalid.
11366 2017-09-20 Richard Biener <rguenther@suse.de>
11368 PR tree-optimization/82264
11369 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11371 (vn_phi_lookup): Likewise.
11372 (vn_phi_insert): Likewise.
11374 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11376 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11377 that fits into uhwi or shwi, add DW_AT_const_value regardless
11378 of early_dwarf without going through RTL, using add_AT_unsigned
11381 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11382 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11383 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11384 (DEBUG_MACRO_SECTION): Likewise.
11385 (DEBUG_LTO_MACRO_SECTION): Likewise.
11386 (DEBUG_STR_DWO_SECTION): Likewise.
11387 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11388 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11389 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11390 (DEBUG_LTO_LINE_STR_SECTION): Define.
11391 (init_sections_and_labels): Initialize debug_line_str_section
11392 variable. Initialize debug_loc_section for -gdwarf-5 to
11393 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11395 2017-09-20 Richard Biener <rguenther@suse.de>
11397 * graphite-sese-to-poly.c (extract_affine): Properly handle
11398 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11400 2017-09-20 Richard Biener <rguenther@suse.de>
11402 PR tree-optimization/81373
11403 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11404 Force SESE live-out defs to be handled even if they are
11407 2017-09-19 Jeff Law <law@redhat.com>
11409 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11410 nothing for stack adjustments with REG_STACK_CHECK.
11411 * sched-deps.c (parse_add_or_inc): Reject insns with
11412 REG_STACK_CHECK from dependency breaking.
11413 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11414 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11415 * reg-notes.def (STACK_CHECK): New note.
11417 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11418 (ix86_expand_prologue): Dump stack clash info as needed.
11419 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11421 * function.c (dump_stack_clash_frame_info): New function.
11422 * function.h (dump_stack_clash_frame_info): Prototype.
11423 (enum stack_clash_probes): New enum.
11425 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11426 flag_stack_clash_protection.
11427 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11428 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11429 (arm_frame_pointer_required): Likewise.
11430 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11431 (ia64_expand_prologue): Likewise.
11432 * config/mips/mips.c (mips_expand_prologue): Likewise.
11433 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11434 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11435 (sparc_flat_expand_prologue): Likewise.
11436 * config/spu/spu.c (spu_expand_prologue): Likewise.
11438 * explow.c: Include "params.h".
11439 (anti_adjust_stack_and_probe_stack_clash): New function.
11440 (get_stack_check_protect): Likewise.
11441 (compute_stack_clash_protection_loop_data): Likewise.
11442 (emit_stack_clash_protection_loop_start): Likewise.
11443 (emit_stack_clash_protection_loop_end): Likewise.
11444 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11445 Use anti_adjust_stack_and_probe_stack_clash.
11446 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11447 (emit_stack_clash_protection_loop_start): Likewise.
11448 (emit_stack_clash_protection_loop_end): Likewise.
11449 * rtl.h (get_stack_check_protect): Prototype.
11450 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11451 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11452 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11454 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11456 * doc/tm.texi: Rebuilt.
11457 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11458 get_stack_check_protect.
11459 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11460 * config/arm/arm.c (arm_expand_prologue): Likewise.
11461 (arm_frame_pointer_required): Likewise.
11462 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11463 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11464 * config/mips/mips.c (mips_expand_prologue): Likewise.
11465 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11466 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11467 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11468 (sparc_flat_expand_prologue): Likewise.
11470 * common.opt (-fstack-clash-protection): New option.
11471 * flag-types.h (enum stack_check_type): Note difference between
11472 -fstack-check= and -fstack-clash-protection.
11473 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11474 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11475 * toplev.c (process_options): Issue warnings/errors for cases
11476 not handled with -fstack-clash-protection.
11477 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11478 (-fstack-check): Note additional problem with -fstack-check=generic.
11479 Note that -fstack-check is primarily for Ada and refer users
11480 to -fstack-clash-protection for stack-clash-protection.
11481 Document new params for stack clash protection.
11483 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11485 * config/i386/i386.c (ix86_split_long_move): Do not handle
11486 address used for LEA in a special way.
11488 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11490 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11492 2017-09-19 Martin Sebor <msebor@redhat.com>
11495 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11496 of incompatible types.
11498 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11500 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11501 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11502 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11503 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11505 2017-09-19 Richard Biener <rguenther@suse.de>
11507 PR tree-optimization/82244
11508 * tree-vrp.c (remove_range_assertions): Do not propagate
11509 a constant to abnormals but replace the assert with a copy.
11511 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11513 PR rtl-optimization/57878
11514 PR rtl-optimization/68988
11515 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11516 avoidance test involving non_reload_pseudos. Move frequency test
11517 below the general fragmentation avoidance test.
11519 2017-09-19 Richard Biener <rguenther@suse.de>
11521 PR tree-optimization/69728
11522 * graphite-sese-to-poly.c (schedule_error): New global.
11523 (add_loop_schedule): Handle empty domain by failing the
11525 (build_original_schedule): Handle schedule_error.
11527 2017-09-19 Richard Biener <rguenther@suse.de>
11529 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11530 Do not iterate to sibling loops but only to siblings of inner
11533 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11536 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11537 that the registers are different.
11539 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11541 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11542 to processor_model and "amdfam17h" to arch_names_table.
11543 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11545 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11548 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11550 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11551 Alan Hayward <alan.hayward@arm.com>
11552 David Sherwood <david.sherwood@arm.com>
11554 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11556 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11558 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11559 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11560 vect_schedule_slp_instance.
11561 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11562 Update call to vect_slp_analyze_node_operations. Simplify return
11564 (vect_slp_analyze_bb_1): Update call accordingly.
11565 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11566 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11567 (vect_schedule_slp): Update call accordingly.
11569 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11570 Alan Hayward <alan.hayward@arm.com>
11571 David Sherwood <david.sherwood@arm.com>
11573 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11574 with types that aren't in fact scalar.
11576 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11578 * tree-vect-slp.c (vect_record_max_nunits): New function,
11580 (vect_build_slp_tree_1): ...here.
11581 (vect_build_slp_tree_2): Call it for phis too.
11583 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11585 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11586 to vect_get_vec_def_for_operand when getting the mask operand.
11588 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11589 Alan Hayward <alan.hayward@arm.com>
11590 David Sherwood <david.sherwood@arm.com>
11592 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11595 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11596 Alan Hayward <alan.hayward@arm.com>
11597 David Sherwood <david.sherwood@arm.com>
11599 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11600 calculation for vector booleans.
11602 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11603 Alan Hayward <alan.hayward@arm.com>
11604 David Sherwood <david.sherwood@arm.com>
11606 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11608 (vect_transform_stmt): ...here.
11609 (vect_analyze_stmt): Use it instead of calling
11610 vectorizable_live_operation directly.
11612 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11614 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11615 non-SIMT targets in acc vector loops.
11617 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11619 * configure.ac: Add arc and check if assembler supports gdwarf2.
11620 * configure: Regenerate.
11622 2017-09-18 Richard Biener <rguenther@suse.de>
11624 PR tree-optimization/82220
11625 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11626 epilogue niters from the min_profitable_iters compute.
11628 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11631 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11632 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11633 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11635 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11638 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11639 switching to a new text section.
11641 2017-09-18 Richard Biener <rguenther@suse.de>
11643 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11645 (build_alias_set): Reject aliases with no access function.
11647 2017-09-18 Richard Biener <rguenther@suse.de>
11649 PR tree-optimization/79622
11650 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11652 (build_cross_bb_scalars_use): Likewise.
11654 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11656 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11658 2017-09-18 Alan Modra <amodra@gmail.com>
11661 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11662 stack_pointer_rtx for count 0. Update comments. Break up
11663 large rtl expression.
11665 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11667 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11668 Increase to 20 bytes.
11669 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11670 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11671 or avx version of the stub.
11673 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11676 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11677 compute the minimum stack alignment. Also update preferred stack
11678 boundary for leaf functions.
11680 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11682 PR tree-optimization/82228
11683 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11686 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11688 * common/config/nds32/nds32-common.c
11689 (nds32_option_optimization_table): Refine formatting.
11690 (nds32_option_optimization_table): Use -fsched-pressure and
11691 -fomit-frame-pointer for specific optimization level.
11693 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11695 * config/nds32/nds32.c: Refine formatting and comments.
11696 * config/nds32/nds32.h: Likewise.
11697 * config/nds32/nds32.md: Likewise.
11698 * config/nds32/nds32-cost.c: Likewise.
11699 * config/nds32/nds32-isr.c: Likewise.
11700 * config/nds32/nds32-md-auxiliary.c: Likewise.
11701 * config/nds32/nds32-multiple.md: Likewise.
11702 * config/nds32/nds32-predicates.c: Likewise.
11704 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11705 Jakub Jelinek <jakub@redhat.com>
11707 Add support for -std=c++2a.
11708 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11710 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11712 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11715 * doc/extend.texi (Common Function Attributes): Add
11716 references to ARM, AArch64, and S/390 specific attributes.
11717 (Function Specific Option Pragmas): Add AArch64 and S/390
11718 to list of back ends that support the target pragma.
11720 2017-09-15 Nathan Sidwell <nathan@acm.org>
11722 * doc/standards.texi: Fix C++17 description. Update URLs for
11725 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11727 * common.opt (Wcast-align=strict): New warning option.
11728 * doc/invoke.texi: Document -Wcast-align=strict.
11730 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11732 * cgraph.h (cgraph_thunk_info): Add comments.
11733 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11734 assert for VIRTUAL_* arguments stricter.
11736 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11738 PR tree-optimization/71026
11739 * match.pd: Move RDIV patterns from fold-const.c
11740 * fold-const.c (distribute_real_division): Removed.
11741 (fold_binary_loc): Remove calls to distribute_real_divison.
11743 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11745 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11746 c++1z and gnu++1z as deprecated. Change other references to
11747 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11748 Change -Wc++1z-compat to -Wc++17-compat.
11749 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11750 * dwarf2out.c (highest_c_language): Handle C++17.
11751 (gen_compile_unit_die): Likewise.
11753 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11755 PR rtl-optimization/82192
11756 * combine.c (make_extraction): Don't look through non-paradoxical
11757 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11760 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11761 Alan Hayward <alan.hayward@arm.com>
11762 David Sherwood <david.sherwood@arm.com>
11764 * target.def (function_arg_offset): New hook.
11765 * targhooks.h (default_function_arg_offset): Declare.
11766 * targhooks.c (default_function_arg_offset): New function.
11767 * function.c (locate_and_pad_parm): Use
11768 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11769 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11770 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11771 * doc/tm.texi: Regenerate.
11772 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11773 * config/spu/spu.c (spu_function_arg_offset): New function.
11774 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11775 * system.h (FUNCTION_ARG_OFFSET): Poison.
11777 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11778 Alan Hayard <alan.hayward@arm.com>
11779 David Sherwood <david.sherwood@arm.com>
11781 * target.def (truly_noop_truncation): New hook.
11782 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11783 than TRULY_NOOP_TRUNCATION.
11784 * hooks.h (hook_bool_uint_uint_true): Declare.
11785 * hooks.c (hook_bool_uint_uint_true): New function.
11786 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11787 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11788 * doc/tm.texi: Regenerate.
11789 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11790 rather than TRULY_NOOP_TRUNCATION in comments.
11791 (simplify_comparison): Likewise.
11792 (record_truncated_value): Likewise.
11793 * expmed.c (extract_bit_field_1): Likewise.
11794 (extract_split_bit_field): Likewise.
11795 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11796 instead of TRULY_NOOP_TRUNCATION.
11797 * function.c (assign_parm_setup_block): Likewise.
11798 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11799 * rtlhooks.c: Include target.h.
11800 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11801 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11802 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11803 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11804 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11805 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11806 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11807 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11808 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11809 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11810 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11811 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11812 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11813 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11814 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11815 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11816 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11817 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11818 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11819 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11820 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11821 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11822 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11823 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11824 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11825 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11826 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11827 rather than TRULY_NOOP_TRUNCATION in comments.
11828 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11829 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11830 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11831 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11832 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11833 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11834 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11835 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11836 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11837 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11838 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11839 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11840 rather than TRULY_NOOP_TRUNCATION in comments.
11841 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11842 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11843 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11844 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11845 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11846 TRULY_NOOP_TRUNCATION condition.
11847 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11848 (TRULY_NOOP_TRUNCATION): Delete.
11849 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11850 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11851 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11852 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11853 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11854 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11855 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11856 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11857 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11858 rather than TRULY_NOOP_TRUNCATION in comments.
11859 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11860 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11861 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11862 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11863 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11864 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11866 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11869 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11870 (*cmp_ior): Likewise.
11871 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11872 (*ior_scc_scc_cmp): Likewise.
11873 (*and_scc_scc): Likewise.
11874 (*and_scc_scc_cmp): Likewise.
11876 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11877 Alan Hayard <alan.hayward@arm.com>
11878 David Sherwood <david.sherwood@arm.com>
11880 * target.def (can_change_mode_class): New hook.
11881 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11882 (hard_regno_nregs): Likewise.
11883 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11884 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11885 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11886 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11887 (LOAD_EXTEND_OP): Update accordingly.
11888 * doc/tm.texi: Regenerate.
11889 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11890 CANNOT_CHANGE_MODE_CLASS.
11891 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11892 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11893 * combine.c (simplify_set): Update accordingly.
11894 * emit-rtl.c (validate_subreg): Likewise.
11895 * recog.c (general_operand): Likewise.
11896 * regcprop.c (mode_change_ok): Likewise.
11897 * reload1.c (choose_reload_regs): Likewise.
11898 (inherit_piecemeal_p): Likewise.
11899 * rtlanal.c (simplify_subreg_regno): Likewise.
11900 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11901 instead of CANNOT_CHANGE_MODE_CLASS.
11902 (reload_cse_simplify_operands): Likewise.
11903 * reload.c (push_reload): Use targetm.can_change_mode_class
11904 instead of CANNOT_CHANGE_MODE_CLASS.
11905 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11906 REG_CANNOT_CHANGE_MODE_P.
11907 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11908 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11909 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11910 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11911 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11912 (arm_can_change_mode_class): New function.
11913 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11914 than CANNOT_CHANGE_MODE_CLASS in comments.
11915 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11916 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11917 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11918 (ix86_can_change_mode_class): ...this new function, inverting the
11919 sense of the return value.
11920 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11921 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11922 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11923 (ia64_can_change_mode_class): New function.
11924 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11925 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11926 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11927 (m32c_can_change_mode_class): ...this new function, inverting the
11928 sense of the return value.
11929 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11930 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11931 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11932 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11933 (mips_can_change_mode_class): ...this new function, inverting the
11934 sense of the return value.
11935 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11936 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11937 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11938 (msp430_can_change_mode_class): New function.
11939 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11940 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11941 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11942 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11943 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11944 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11945 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11946 (pa_cannot_change_mode_class): Replace with...
11947 (pa_can_change_mode_class): ...this new function, inverting the
11948 sense of the return value.
11949 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11950 than CANNOT_CHANGE_MODE_CLASS in comments.
11951 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11952 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11953 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11954 (pdp11_cannot_change_mode_class): Replace with...
11955 (pdp11_can_change_mode_class): ...this new function, inverting the
11956 sense of the return value.
11957 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11958 * config/powerpcspe/powerpcspe-protos.h
11959 (rs6000_cannot_change_mode_class_ptr): Delete.
11960 * config/powerpcspe/powerpcspe.c
11961 (rs6000_cannot_change_mode_class_ptr): Delete.
11962 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11963 (rs6000_option_override_internal): Assign to
11964 targetm.can_change_mode_class instead of
11965 rs6000_cannot_change_mode_class_ptr.
11966 (rs6000_cannot_change_mode_class): Replace with...
11967 (rs6000_can_change_mode_class): ...this new function, inverting the
11968 sense of the return value.
11969 (rs6000_debug_cannot_change_mode_class): Replace with...
11970 (rs6000_debug_can_change_mode_class): ...this new function.
11971 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11972 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11973 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11974 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11975 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11977 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11978 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11979 (rs6000_option_override_internal): Assign to
11980 targetm.can_change_mode_class instead of
11981 rs6000_cannot_change_mode_class_ptr.
11982 (rs6000_cannot_change_mode_class): Replace with...
11983 (rs6000_can_change_mode_class): ...this new function, inverting the
11984 sense of the return value.
11985 (rs6000_debug_cannot_change_mode_class): Replace with...
11986 (rs6000_debug_can_change_mode_class): ...this new function.
11987 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11988 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11989 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11990 (s390_can_change_mode_class): ...this new function, inverting the
11991 sense of the return value.
11992 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11993 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11994 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11995 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11996 (sh_cannot_change_mode_class): Replace with...
11997 (sh_can_change_mode_class): ...this new function, inverting the
11998 sense of the return value.
11999 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12000 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12001 (sparc_can_change_mode_class): New function.
12002 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12003 * config/spu/spu.c (spu_can_change_mode_class): New function.
12004 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12005 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12006 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12007 (visium_can_change_mode_class): New function.
12008 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12010 2017-09-15 Richard Biener <rguenther@suse.de>
12012 PR tree-optimization/82217
12013 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12014 but not undefined case.
12016 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12019 * postreload.c (reload_cse_simplify_operands): Skip
12020 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12022 2017-09-15 Richard Biener <rguenther@suse.de>
12024 PR tree-optimization/68823
12025 * graphite-scop-detection.c (build_alias_set): If we have a
12026 possible dependence check whether we can handle them by just
12027 looking at the DRs DR_ACCESS_FNs.
12028 (build_scops): If build_alias_set fails, fail the SCOP.
12030 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12032 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12033 to support float128 built-in functions that require the ISA 3.0
12035 (BU_FLOAT128_3_HW): Likewise.
12036 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12037 built-in functions.
12038 (FMAF128): Likewise.
12040 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12041 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12042 floating point instructions.
12043 (rs6000_invalid_builtin): Likewise.
12044 (rs6000_builtin_mask_names): Likewise.
12045 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12046 (RS6000_BTM_FLOAT128_HW): Likewise.
12047 (RS6000_BTM_COMMON): Likewise.
12048 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12050 * doc/extend.texi (RS/6000 built-in functions): Document the
12051 IEEE 128-bit floating point square root and fused multiply-add
12052 built-in functions.
12054 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12056 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12057 reg (r2) isn't in the set of registers defined in the prologue.
12059 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12060 Alan Hayward <alan.hayward@arm.com>
12061 David Sherwood <david.sherwood@arm.com>
12063 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12064 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12065 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12066 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12067 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12069 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12070 max_vectorization_factor.
12071 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12073 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12074 Alan Hayward <alan.hayward@arm.com>
12075 David Sherwood <david.sherwood@arm.com>
12077 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12078 (vect_worthwhile_without_simd_p): Declare.
12079 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12080 (vectorizable_reduction): Use it.
12081 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12082 (vectorizable_operation): Likewise.
12084 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12085 Alan Hayward <alan.hayward@arm.com>
12086 David Sherwood <david.sherwood@arm.com>
12088 * tree-vectorizer.h (vect_get_num_copies): New function.
12089 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12090 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12091 (vectorizable_induction): Likewise.
12092 (vectorizable_live_operation): Likewise.
12093 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12094 (vectorizable_bswap): Likewise.
12095 (vectorizable_call): Likewise.
12096 (vectorizable_conversion): Likewise.
12097 (vectorizable_assignment): Likewise.
12098 (vectorizable_shift): Likewise.
12099 (vectorizable_operation): Likewise.
12100 (vectorizable_store): Likewise.
12101 (vectorizable_load): Likewise.
12102 (vectorizable_condition): Likewise.
12103 (vectorizable_comparison): Likewise.
12104 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12106 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12107 Alan Hayward <alan.hayward@arm.com>
12108 David Sherwood <david.sherwood@arm.com>
12110 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12111 of vect_init_vector.
12113 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12114 Alan Hayward <alan.hayward@arm.com>
12115 David Sherwood <david.sherwood@arm.com>
12117 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12118 an inline wrapper that provides a location.
12119 (gimple_build_vector): Likewise.
12120 * gimple-fold.c (gimple_build_vector_from_val): New function.
12121 (gimple_build_vector): Likewise.
12122 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12123 functions to build the initial value. Always return a gimple value.
12124 (get_initial_defs_for_reduction): Likewise. Only compute
12126 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12127 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12128 (vectorizable_induction): Use gimple_build_vector rather than
12131 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12132 Alan Hayward <alan.hayward@arm.com>
12133 David Sherwood <david.sherwood@arm.com>
12135 * target.h (vec_perm_indices): New typedef.
12136 (auto_vec_perm_indices): Likewise.
12137 * optabs-query.h: Include target.h
12138 (can_vec_perm_p): Take a vec_perm_indices *.
12139 * optabs-query.c (can_vec_perm_p): Likewise.
12140 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12141 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12142 * tree-vect-generic.c (lower_vec_perm): Likewise.
12143 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12144 (vect_grouped_load_supported): Likewise.
12145 (vect_shift_permute_load_chain): Likewise.
12146 (vect_permute_store_chain): Use auto_vec_perm_indices.
12147 (vect_permute_load_chain): Likewise.
12148 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12149 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12150 Update uses of can_vec_perm_p.
12151 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12152 mode with a number of elements. Take a vec_perm_indices *.
12153 (vect_create_epilog_for_reduction): Update accordingly.
12154 Use auto_vec_perm_indices.
12155 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12156 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12157 (vect_transform_slp_perm_load): Likewise.
12158 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12159 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12160 (vect_gen_perm_mask_checked): Likewise.
12161 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12162 (vect_gen_perm_mask_checked): Likewise.
12163 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12164 (vectorizable_store): Likewise.
12165 (vectorizable_load): Likewise.
12166 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12167 (vectorizable_bswap): Likewise.
12169 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12170 Alan Hayward <alan.hayward@arm.com>
12171 David Sherwood <david.sherwood@arm.com>
12173 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12174 * tree.c (build_vector): Likewise.
12175 (build_vector_from_ctor): Update accordingly.
12176 (build_vector_from_val): Likewise.
12177 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12178 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12179 * tree-vect-generic.c (add_rshift): Likewise.
12180 (expand_vector_divmod): Likewise.
12181 (optimize_vector_constructor): Likewise.
12182 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12183 (vect_transform_slp_perm_load): Likewise.
12184 (vect_schedule_slp_instance): Likewise.
12185 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12186 (vectorizable_call): Likewise.
12187 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12188 * expmed.c (make_tree): Likewise.
12189 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12190 a vector passed to build_vector.
12191 (fold_convert_const): Likewise.
12192 (exact_inverse): Likewise.
12193 (fold_ternary_loc): Likewise.
12194 (fold_relational_const): Likewise.
12195 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12196 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12197 (const_unop): Likewise. Store the reduction accumulator in a
12198 variable rather than an array.
12199 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12200 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12201 the new vector, rather than constructing it after the input arrays.
12202 (native_interpret_vector): Use auto_vec<tree> when building
12203 a vector passed to build_vector. Add elements in order.
12204 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12205 auto_vec<tree> when building a vector passed to build_vector.
12206 (vect_create_epilog_for_reduction): Likewise.
12207 (vectorizable_induction): Likewise.
12208 (get_initial_def_for_reduction): Likewise. Fix indentation of
12210 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12212 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12213 passed to build_vector.
12215 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12216 Alan Hayward <alan.hayward@arm.com>
12217 David Sherwood <david.sherwood@arm.com>
12219 * tree-core.h (tree_base::u): Add an "nelts" field.
12220 (tree_vector): Use VECTOR_CST_NELTS as the length.
12221 * tree.c (tree_size): Likewise.
12222 (make_vector): Initialize VECTOR_CST_NELTS.
12223 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12224 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12225 TYPE_VECTOR_SUBPARTS.
12226 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12227 as the number of units, setting it from VECTOR_CST_NELTS.
12228 (const_vector_from_tree): Likewise.
12229 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12230 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12231 (fold_negate_expr_1): Likewise.
12232 (fold_convert_const): Likewise.
12233 (const_binop): Likewise. Differentiate the number of output and
12235 (const_unop): Likewise.
12236 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12237 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12238 in cases that did the opposite.
12240 2017-09-14 Richard Biener <rguenther@suse.de>
12242 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12245 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12247 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12249 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12252 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12253 if and where to split a bb, except for splitting before debug insn
12254 sequences followed by non-label real insn. Delete debug insns
12255 in between basic blocks.
12257 * combine.c (make_compound_operation_int): Formatting fixes.
12259 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12260 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12261 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12262 * config/sol2.h (LINK_EH_SPEC): Likewise.
12263 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12264 * config/s390/linux.h (LINK_SPEC): Likewise.
12265 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12266 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12267 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12268 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12269 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12270 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12271 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12272 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12274 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12276 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12278 (ENDFILE_LINUX_SPEC): Likewise.
12279 (LINK_EH_SPEC): Likewise.
12280 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12281 (LINK_OS_LINUX_SPEC32): Likewise.
12282 (LINK_OS_LINUX_SPEC64): Likewise.
12283 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12284 (LINK_OS_LINUX_SPEC): Likewise.
12286 2017-09-13 Martin Liska <mliska@suse.cz>
12288 PR middle-end/82154
12289 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12290 CASE_HIGH is NULL_TREE.
12292 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12293 Alan Hayward <alan.hayward@arm.com>
12294 David Sherwood <david.sherwood@arm.com>
12296 * target.def (secondary_memory_needed): New hook.
12297 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12298 instead of SECONDARY_MEMORY_NEEDED.
12299 (secondary_memory_needed_mode): Likewise.
12300 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12301 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12302 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12303 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12304 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12305 * doc/tm.texi: Regenerate.
12306 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12307 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12308 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12309 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12310 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12311 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12312 mode argument first and change the reg_class arguments to reg_class_t.
12313 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12314 Make static. Update the call to inline_secondary_memory_needed.
12315 (ix86_register_move_cost): Update the call to
12316 inline_secondary_memory_needed.
12317 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12318 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12320 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12321 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12323 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12324 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12325 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12326 and match hook interface. Add comment from mips.h.
12327 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12328 * config/mmix/mmix.md (truncdfsf2): Refer to
12329 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12331 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12332 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12333 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12334 (pa_secondary_memory_needed): New function.
12335 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12336 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12337 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12338 (pdp11_secondary_memory_needed): Make static and match hook interface.
12339 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12340 * config/powerpcspe/powerpcspe-protos.h
12341 (rs6000_secondary_memory_needed_ptr): Delete.
12342 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12344 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12345 (rs6000_option_override_internal): Assign to
12346 targetm.secondary_memory_needed rather than
12347 rs6000_secondary_memory_needed_ptr.
12348 (rs6000_secondary_memory_needed): Match hook interface.
12349 (rs6000_debug_secondary_memory_needed): Likewise.
12350 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12351 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12352 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12353 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12354 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12355 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12357 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12358 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12359 (rs6000_option_override_internal): Assign to
12360 targetm.secondary_memory_needed rather than
12361 rs6000_secondary_memory_needed_ptr.
12362 (rs6000_secondary_memory_needed): Match hook interface.
12363 (rs6000_debug_secondary_memory_needed): Likewise.
12364 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12365 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12366 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12367 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12368 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12369 (sparc_secondary_memory_needed): New function.
12370 * lra-constraints.c (check_and_process_move): Refer to
12371 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12373 (curr_insn_transform): Likewise.
12374 (process_alt_operands): Use targetm.secondary_memory_needed
12375 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12376 (check_secondary_memory_needed_p): Likewise.
12377 (choose_split_class): Likewise.
12378 * reload.c: Unconditionally include code that was previously
12379 conditional on SECONDARY_MEMORY_NEEDED.
12380 (push_secondary_reload): Use targetm.secondary_memory_needed
12381 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12382 (push_reload): Likewise.
12383 * reload1.c: Unconditionally include code that was previously
12384 conditional on SECONDARY_MEMORY_NEEDED.
12385 (choose_reload_regs): Use targetm.secondary_memory_needed
12386 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12387 (gen_reload): Likewise.
12388 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12390 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12391 Alan Hayward <alan.hayward@arm.com>
12392 David Sherwood <david.sherwood@arm.com>
12394 * target.def (secondary_memory_needed_mode): New hook:
12395 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12396 * targhooks.h (default_secondary_memory_needed_mode): New function.
12397 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12398 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12399 * doc/tm.texi: Regenerate.
12400 * lra-constraints.c (check_and_process_move): Use
12401 targetm.secondary_memory_needed_mode instead of
12402 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12403 (curr_insn_transform): Likewise.
12404 * reload.c (get_secondary_mem): Likewise.
12405 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12406 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12408 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12409 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12410 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12411 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12412 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12414 * config/powerpcspe/powerpcspe-protos.h
12415 (rs6000_secondary_memory_needed_mode): Delete.
12416 * config/powerpcspe/powerpcspe.c
12417 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12418 (rs6000_secondary_memory_needed_mode): Make static.
12419 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12420 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12422 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12424 (rs6000_secondary_memory_needed_mode): Make static.
12425 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12426 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12427 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12428 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12429 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12431 (sparc_secondary_memory_needed_mode): New function.
12432 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12434 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12436 * config/aarch64/constraints.md (Umq): New constraint.
12437 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12439 (mov<mode>): Update condition.
12441 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12443 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12444 when bitposition is the same.
12446 2017-09-13 Richard Biener <rguenther@suse.de>
12448 * dwarf2out.c (output_die_symbol): Remove.
12449 (output_die): Do not output a DIEs symbol.
12451 2017-09-13 Richard Biener <rguenther@suse.de>
12453 PR middle-end/82128
12454 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12455 default-def to avoid breaking iterator update with the weird
12456 interaction with cgraph_update_edges_for_call_stmt_node.
12458 2017-09-13 Richard Biener <rguenther@suse.de>
12460 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12461 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12462 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12463 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12465 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12467 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12468 Disable pc relative literal load irrespective of
12469 TARGET_FIX_ERR_A53_84341 for default.
12471 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12473 * config/sparc/sparc.c (output_return): Output the source location of
12474 the insn in the delay slot, if any.
12475 (output_sibcall): Likewise.
12477 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12480 * common.opt (-static-pie): New alias.
12481 (shared): Negate static-pie.
12482 (-no-pie): Update help text.
12484 (static-pie): New option.
12485 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12486 -static-pie support.
12487 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12488 (LINK_EH_SPEC): Likewise.
12489 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12490 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12491 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12492 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12493 (init_gcc_specs): Likewise.
12494 (init_spec): Likewise.
12495 (display_help): Update help message for -pie.
12496 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12499 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12501 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12502 (movdi_aarch64): Likewise.
12503 (movti_aarch64): Likewise.
12505 2017-09-12 Simon Wright <simon@pushface.org>
12508 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12509 calculation of the minor version, always output as 0.
12511 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12514 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12515 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12516 on it early, rather than manual conversion late. For
12517 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12518 instead of performing manual conversion.
12520 2017-09-12 Carl Love <cel@us.ibm.com>
12522 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12523 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12525 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12526 VMULOSW): Add definitions.
12527 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12528 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12529 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12530 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12531 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12533 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12535 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12537 (movti_aarch64): Likewise.
12538 (movdf_aarch64): Likewise.
12539 (movtf_aarch64): Likewise.
12540 (load_pairdi): Likewise.
12541 (store_pairdi): Likewise.
12542 (load_pairdf): Likewise.
12543 (store_pairdf): Likewise.
12544 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12545 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12546 (ldr_got_small_<mode>): Likewise.
12547 (ldr_got_small_28k_<mode>): Likewise.
12548 (ldr_got_tiny): Likewise.
12549 * config/aarch64/iterators.md (ldst_sz): New.
12550 (ldpstp_sz): Likewise.
12551 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12553 (thunderx_load): Split load_8 to load_16.
12554 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12556 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12557 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12558 (xgene1_store_pair): Split store_8 to store_16.
12559 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12560 (falkor_st_0_st_sd): Split store_8 to store_16.
12562 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12564 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12565 and store1/2/3/4 to store_4/8/12/16.
12566 * config/aarch64/aarch64.md: Update for rename.
12567 * config/arm/arm.md: Likewise.: Likewise.
12568 * config/arm/arm.c: Likewise.
12569 * config/arm/thumb1.md: Likewise.
12570 * config/arm/thumb2.md: Likewise.
12571 * config/arm/vfp.md: Likewise.
12572 * config/arm/arm-generic.md: Likewise.
12573 * config/arm/arm1020e.md: Likewise.
12574 * config/arm/arm1026ejs.md: Likewise.
12575 * config/arm/arm1136jfs.md: Likewise.
12576 * config/arm/arm926ejs.md: Likewise.
12577 * config/arm/cortex-a15.md: Likewise.
12578 * config/arm/cortex-a17.md: Likewise.
12579 * config/arm/cortex-a5.md: Likewise.
12580 * config/arm/cortex-a53.md: Likewise.
12581 * config/arm/cortex-a57.md: Likewise.
12582 * config/arm/cortex-a7.md: Likewise.
12583 * config/arm/cortex-a8.md: Likewise.
12584 * config/arm/cortex-a9.md: Likewise.
12585 * config/arm/cortex-m4.md: Likewise.
12586 * config/arm/cortex-m7.md: Likewise.
12587 * config/arm/cortex-r4.md: Likewise.
12588 * config/arm/exynos-m1.md: Likewise.
12589 * config/arm/fa526.md: Likewise.
12590 * config/arm/fa606te.md: Likewise.
12591 * config/arm/fa626te.md: Likewise.
12592 * config/arm/fa726te.md: Likewise.
12593 * config/arm/fmp626.md: Likewise.
12594 * config/arm/iwmmxt.md: Likewise.
12595 * config/arm/ldmstm.md: Likewise.
12596 * config/arm/marvell-pj4.md: Likewise.
12597 * config/arm/xgene1.md: Likewise.
12598 * config/aarch64/thunderx.md: Likewise.
12599 * config/aarch64/thunderx2t99.md: Likewise.
12600 * config/aarch64/falkor.md: Likewise.
12602 2017-09-12 Martin Liska <mliska@suse.cz>
12604 * attribs.c (private_lookup_attribute): New function.
12605 * attribs.h (private_lookup_attribute): Declared here.
12606 (lookup_attribute): Called from this place.
12608 2017-09-12 Richard Biener <rguenther@suse.de>
12610 PR tree-optimization/82157
12611 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12612 stmts with side-effects.
12614 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12615 Alan Hayward <alan.hayward@arm.com>
12616 David Sherwood <david.sherwood@arm.com>
12618 * target.def (hard_regno_nregs): New hook.
12619 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12620 * targhooks.h (default_hard_regno_nregs): Declare.
12621 * targhooks.c (default_hard_regno_nregs): New function.
12622 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12623 (TARGET_HARD_REGNO_NREGS): ...this hook.
12624 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12625 (CLASS_MAX_NREGS): Likewise.
12626 * doc/tm.texi: Regenerate.
12627 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12628 instead of HARD_REGNO_NREGS.
12629 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12630 HARD_REGNO_NREGS in the comment.
12631 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12632 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12633 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12634 Return an unsigned int.
12635 (TARGET_HARD_REGNO_NREGS): Redefine.
12636 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12637 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12638 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12639 (arc_hard_regno_nregs): New function.
12640 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12641 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12642 (arm_hard_regno_nregs): New function.
12643 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12644 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12645 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12646 (TARGET_HARD_REGNO_NREGS): Redefine.
12647 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12648 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12649 (HARD_REGNO_NREGS): Delete.
12650 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12651 (cr16_hard_regno_nregs): New function.
12652 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12653 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12654 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12655 (cris_hard_regno_nregs): New function.
12656 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12657 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12658 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12659 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12660 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12661 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12662 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12663 (frv_hard_regno_nregs): Make static. Take and return an
12665 (frv_class_max_nregs): Remove outdated copy of documentation.
12666 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12667 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12668 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12669 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12670 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12671 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12672 (TARGET_HARD_REGNO_NREGS): Redefine.
12673 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12674 (CLASS_MAX_NREGS): Update comment.
12675 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12676 (ia64_hard_regno_nregs): New function.
12677 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12678 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12679 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12680 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12681 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12683 (m32c_hard_regno_nregs): Likewise. Make static.
12684 (TARGET_HARD_REGNO_NREGS): Redefine.
12685 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12686 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12687 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12688 (m68k_hard_regno_nregs): New function.
12689 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12690 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12691 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12692 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12693 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12694 Take and return an unsigned int.
12695 (TARGET_HARD_REGNO_NREGS): Redefine.
12696 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12697 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12698 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12699 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12700 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12701 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12702 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12703 (msp430_hard_regno_nregs): Make static. Take and return an
12705 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12706 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12707 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12708 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12709 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12710 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12711 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12712 (TARGET_HARD_REGNO_NREGS): Redefine.
12713 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12714 (PA_HARD_REGNO_NREGS): ...this.
12715 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12716 (PA_HARD_REGNO_NREGS): ...this.
12717 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12718 (pa_hard_regno_nregs): New function.
12719 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12720 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12721 (pdp11_hard_regno_nregs): New function.
12722 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12723 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12724 (rs6000_hard_regno_nregs_hook): New function.
12725 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12726 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12727 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12728 Take and return an unsigned int. Move earlier in file.
12729 (TARGET_HARD_REGNO_NREGS): Redefine.
12730 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12731 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12732 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12733 (rl78_hard_regno_nregs): Make static. Take and return an
12735 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12736 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12737 (rs6000_hard_regno_nregs_hook): New function.
12738 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12739 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12740 (TARGET_HARD_REGNO_NREGS): Redefine.
12741 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12742 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12743 instead of HARD_REGNO_NREGS.
12744 (s390_hard_regno_nregs): New function.
12745 (s390_hard_regno_mode_ok): Add comment from s390.h.
12746 (TARGET_HARD_REGNO_NREGS): Redefine.
12747 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12748 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12749 (sh_hard_regno_nregs): New function.
12750 (sh_pass_in_reg_p): Use it.
12751 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12752 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12753 (sparc_hard_regno_nregs): New function.
12754 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12755 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12756 (spu_function_arg_advance): Use it, supplying a valid register number.
12757 (TARGET_HARD_REGNO_NREGS): Redefine.
12758 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12759 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12760 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12761 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12762 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12763 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12764 (CLASS_MAX_NREGS): Remove copy of old documentation.
12765 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12766 (visium_hard_regno_nregs): New function.
12767 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12768 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12769 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12770 xtensa_hard_regno_nregs): New function.
12771 * system.h (HARD_REGNO_NREGS): Poison.
12773 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12775 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12776 hard_regno_nregs instead of HARD_REGNO_NREGS.
12777 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12778 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12779 (c6x_expand_epilogue): Likewise.
12780 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12781 (frv_read_iacc_argument): Likewise.
12782 * config/sh/sh.c: Include regs.h.
12783 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12784 (regs_used): Likewise.
12785 (output_stack_adjust): Likewise.
12786 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12787 * expmed.c: Include regs.h.
12788 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12789 * ree.c: Include regs.h.
12790 (combine_reaching_defs): Use hard_regno_nregs instead of
12792 (add_removable_extension): Likewise.
12794 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12796 * regs.h (hard_regno_nregs): Turn into a function.
12797 (end_hard_regno): Update accordingly.
12798 * caller-save.c (setup_save_areas): Likewise.
12799 (save_call_clobbered_regs): Likewise.
12800 (replace_reg_with_saved_mem): Likewise.
12801 (insert_restore): Likewise.
12802 (insert_save): Likewise.
12803 * combine.c (can_change_dest_mode): Likewise.
12804 (move_deaths): Likewise.
12805 (distribute_notes): Likewise.
12806 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12807 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12808 (rs6000_split_multireg_move): Likewise.
12809 (rs6000_register_move_cost): Likewise.
12810 (rs6000_memory_move_cost): Likewise.
12811 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12812 (rs6000_split_multireg_move): Likewise.
12813 (rs6000_register_move_cost): Likewise.
12814 (rs6000_memory_move_cost): Likewise.
12815 * cselib.c (cselib_reset_table): Likewise.
12816 (cselib_lookup_1): Likewise.
12817 * emit-rtl.c (set_mode_and_regno): Likewise.
12818 * function.c (aggregate_value_p): Likewise.
12819 * ira-color.c (setup_profitable_hard_regs): Likewise.
12820 (check_hard_reg_p): Likewise.
12821 (calculate_saved_nregs): Likewise.
12822 (assign_hard_reg): Likewise.
12823 (improve_allocation): Likewise.
12824 (calculate_spill_cost): Likewise.
12825 * ira-emit.c (modify_move_list): Likewise.
12826 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12827 (ira_hard_reg_in_set_p): Likewise.
12828 * ira.c (setup_reg_mode_hard_regset): Likewise.
12829 (clarify_prohibited_class_mode_regs): Likewise.
12830 (check_allocation): Likewise.
12831 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12832 (lra_setup_reg_renumber): Likewise.
12833 (setup_try_hard_regno_pseudos): Likewise.
12834 (spill_for): Likewise.
12835 (assign_hard_regno): Likewise.
12836 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12837 * lra-constraints.c (in_class_p): Likewise.
12838 (lra_constraint_offset): Likewise.
12839 (simplify_operand_subreg): Likewise.
12840 (lra_constraints): Likewise.
12841 (split_reg): Likewise.
12842 (split_if_necessary): Likewise.
12843 (invariant_p): Likewise.
12844 (inherit_in_ebb): Likewise.
12845 * lra-lives.c (process_bb_lives): Likewise.
12846 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12847 (get_hard_regs): Likewise.
12848 (do_remat): Likewise.
12849 * lra-spills.c (assign_spill_hard_regs): Likewise.
12850 * mode-switching.c (create_pre_exit): Likewise.
12851 * postreload.c (reload_combine_recognize_pattern): Likewise.
12852 * recog.c (peep2_find_free_register): Likewise.
12853 * regcprop.c (kill_value_regno): Likewise.
12854 (set_value_regno): Likewise.
12855 (copy_value): Likewise.
12856 (maybe_mode_change): Likewise.
12857 (find_oldest_value_reg): Likewise.
12858 (copyprop_hardreg_forward_1): Likewise.
12859 * regrename.c (check_new_reg_p): Likewise.
12860 (regrename_do_replace): Likewise.
12861 * reload.c (push_reload): Likewise.
12862 (combine_reloads): Likewise.
12863 (find_dummy_reload): Likewise.
12864 (operands_match_p): Likewise.
12865 (find_reloads): Likewise.
12866 (find_equiv_reg): Likewise.
12867 (reload_adjust_reg_for_mode): Likewise.
12868 * reload1.c (count_pseudo): Likewise.
12869 (count_spilled_pseudo): Likewise.
12870 (find_reg): Likewise.
12871 (clear_reload_reg_in_use): Likewise.
12872 (free_for_value_p): Likewise.
12873 (allocate_reload_reg): Likewise.
12874 (choose_reload_regs): Likewise.
12875 (reload_adjust_reg_for_temp): Likewise.
12876 (emit_reload_insns): Likewise.
12877 (delete_output_reload): Likewise.
12878 * rtlanal.c (subreg_get_info): Likewise.
12879 * sched-deps.c (sched_analyze_reg): Likewise.
12880 * sel-sched.c (init_regs_for_mode): Likewise.
12881 (mark_unavailable_hard_regs): Likewise.
12882 (choose_best_reg_1): Likewise.
12883 (verify_target_availability): Likewise.
12884 * valtrack.c (dead_debug_insert_temp): Likewise.
12885 * var-tracking.c (track_loc_p): Likewise.
12886 (emit_note_insn_var_location): Likewise.
12887 * varasm.c (make_decl_rtl): Likewise.
12888 * reginfo.c (choose_hard_reg_mode): Likewise.
12889 (init_reg_modes_target): Refer directly to
12890 this_target_regs->x_hard_regno_nregs.
12892 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12894 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12895 instead of hard_regno_nregs.
12897 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12899 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12900 end_hard_regno instead of hard_regno_nregs.
12901 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12902 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12903 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12904 * ira-color.c (improve_allocation): Likewise.
12905 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12906 * lra-lives.c (mark_regno_live): Likewise.
12907 (mark_regno_dead): Likewise.
12908 * lra-remat.c (operand_to_remat): Likewise.
12909 * lra.c (collect_non_operand_hard_regs): Likewise.
12910 * postreload.c (reload_combine_note_store): Likewise.
12911 (move2add_valid_value_p): Likewise.
12912 * reload.c (regno_clobbered_p): Likewise.
12914 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12916 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12918 * config/v850/v850.c (v850_reorg): Likewise.
12919 * reload.c (refers_to_regno_for_reload_p): Likewise.
12920 (find_equiv_reg): Likewise.
12921 * reload1.c (reload_reg_reaches_end_p): Likewise.
12923 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12925 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12927 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12928 * config/arm/arm.c (output_move_neon): Likewise.
12929 (arm_attr_length_move_neon): Likewise.
12930 (neon_split_vcombine): Likewise.
12931 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12932 (c6x_mark_reg_written): Likewise.
12933 (c6x_dwarf_register_span): Likewise.
12934 * config/i386/i386.c (ix86_save_reg): Likewise.
12935 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12936 (rws_access_reg): Likewise.
12937 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12938 * mode-switching.c (create_pre_exit): Likewise.
12939 * ree.c (combine_reaching_defs): Likewise.
12940 (add_removable_extension): Likewise.
12941 * regcprop.c (find_oldest_value_reg): Likewise.
12942 (copyprop_hardreg_forward_1): Likewise.
12943 * reload.c (reload_inner_reg_of_subreg): Likewise.
12944 (push_reload): Likewise.
12945 (combine_reloads): Likewise.
12946 (find_dummy_reload): Likewise.
12947 (reload_adjust_reg_for_mode): Likewise.
12948 * reload1.c (find_reload_regs): Likewise.
12949 (forget_old_reloads_1): Likewise.
12950 (reload_reg_free_for_value_p): Likewise.
12951 (reload_adjust_reg_for_temp): Likewise.
12952 (emit_reload_insns): Likewise.
12953 (delete_output_reload): Likewise.
12954 * sel-sched.c (choose_best_reg_1): Likewise.
12955 (choose_best_pseudo_reg): Likewise.
12957 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12958 Alan Hayward <alan.hayward@arm.com>
12959 David Sherwood <david.sherwood@arm.com>
12961 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12962 * target.def (slow_unaligned_access): New hook.
12963 * targhooks.h (default_slow_unaligned_access): Declare.
12964 * targhooks.c (default_slow_unaligned_access): New function.
12965 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12966 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12967 * doc/tm.texi: Regenerate.
12968 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12969 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12970 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12972 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12973 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12975 (rs6000_slow_unaligned_access): New function.
12976 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12977 (expand_block_compare): Likewise.
12978 (expand_strn_compare): Likewise.
12979 (rs6000_rtx_costs): Likewise.
12980 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12981 (riscv_slow_unaligned_access): Likewise.
12982 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12983 (riscv_slow_unaligned_access_p): ...this and make static.
12984 (riscv_option_override): Update accordingly.
12985 (riscv_slow_unaligned_access): New function.
12986 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12987 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12988 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12989 (rs6000_slow_unaligned_access): New function.
12990 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12991 (rs6000_rtx_costs): Likewise.
12992 * config/rs6000/rs6000-string.c (expand_block_compare)
12993 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12994 of SLOW_UNALIGNED_ACCESS.
12995 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12996 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12997 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12998 of SLOW_UNALIGNED_ACCESS.
12999 * expmed.c (simple_mem_bitfield_p): Likewise.
13000 * expr.c (alignment_for_piecewise_move): Likewise.
13001 (emit_group_load_1): Likewise.
13002 (emit_group_store): Likewise.
13003 (copy_blkmode_from_reg): Likewise.
13004 (emit_push_insn): Likewise.
13005 (expand_assignment): Likewise.
13006 (store_field): Likewise.
13007 (expand_expr_real_1): Likewise.
13008 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13009 * lra-constraints.c (simplify_operand_subreg): Likewise.
13010 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13011 * gimple-ssa-store-merging.c: Likewise in block comment at start
13013 * tree-ssa-strlen.c: Include target.h.
13014 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13015 of SLOW_UNALIGNED_ACCESS.
13016 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13018 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13020 PR rtl-optimization/82185
13021 * expmed.c (emit_store_flag_int): Only test tem if it has been
13024 2017-09-12 Richard Biener <rguenther@suse.de>
13026 PR middle-end/82149
13027 * match.pd ((FTYPE) N CMP CST): Fix typo.
13029 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13031 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13033 (mips_near_type_p): Add 'short_call' attribute as a synonym
13035 * doc/extend.texi (short_call): Document new function attribute.
13037 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13040 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13041 assertion check that in the condition.
13042 (get_atomic_generic_size): Likewise. Before testing if parameter
13043 has pointer type, if it has array type, call for C++
13044 default_conversion to perform array-to-pointer conversion.
13046 2017-09-12 Richard Biener <rguenther@suse.de>
13048 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13049 for operations we cannot scalarize.
13051 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13053 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13054 vectors heap vectors. Clean up comments.
13055 Make visited_bbs a reference.
13056 (profitable_jump_thread_path): Make GC
13057 vectors heap vectors. Clean up comments.
13059 (convert_and_register_jump_thread_path): Make GC vectors heap
13061 (check_subpath_and_update_thread_path): Same. Clean up comments.
13062 Make visited_bbs a reference.
13063 (handle_phi): Abstract common code to to
13064 register_jump_thread_path_if_profitable.
13065 Rename VAR_BB to DEF_BB.
13067 Make GC vectors heap vectors.
13068 Make visited_bbs a reference.
13069 (handle_assignment): Same.
13070 (register_jump_thread_path_if_profitable): New.
13071 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13073 Make GC vectors heap vectors. Clean up comments.
13074 Make visited_bbs a reference.
13075 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13076 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13079 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13082 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13083 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13085 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13087 Revert r251800 and r251799.
13089 2017-09-11 Martin Jambor <mjambor@suse.cz>
13092 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13093 arguments in advance.
13094 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13095 use it to find predecessor edges.
13096 (naive_outof_ssa): Collect vector of predecessors.
13098 2017-09-08 Jason Merrill <jason@redhat.com>
13100 PR c++/70029 - ICE with ref-qualifier and -flto
13101 * langhooks.h (struct lang_hooks_for_types): Add
13102 copy_lang_qualifiers.
13103 * attribs.c (build_type_attribute_qual_variant): Use it.
13104 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13106 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13107 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13109 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13112 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13113 (*mulsi3_sp64): New instruction.
13114 (mulsi3): New expander.
13116 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13118 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13120 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13122 * sancov.c: Include memmodel.h.
13124 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13127 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13130 2017-09-07 Carl Love <cel@us.ibm.com>
13132 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13133 the sldi instruction.
13135 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13137 * sancov.c: Include tm_p.h.
13139 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13142 * output.h (switch_to_other_text_partition): New declaration.
13143 * varasm.c (switch_to_other_text_partition): New function.
13144 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13145 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13146 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13147 to the other text partition before emitting LCL label and switch back
13148 after emitting the word after it.
13150 2017-09-07 Richard Biener <rguenther@suse.de>
13152 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13153 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13154 critical edges here, after loop init.
13155 (pass_data_pre): Remove PROP_no_crit_edges flags.
13156 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13157 for valueization of call args to avoid leaking VN_TOP.
13158 (visit_use): Assert we do not visit default defs.
13159 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13160 Use error_mark_node to more easily detect leaking VN_TOP.
13161 All default-defs are varying, not VN_TOP. Mark them visited.
13162 (run_scc_vn): Make code match comment.
13164 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13166 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13167 OPTION_MASK_FLOAT128_KEYWORD.
13168 (POWERPC_MASKS): Likewise.
13169 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13170 support for the -mfloat128-type option, and make -mfloat128
13171 default on PowerPC Linux systems. Define or undefine
13172 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13173 Define __float128 to be __ieee128 if IEEE 128-bit support is
13174 enabled, or undefine it.
13175 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13176 Delete defining __FLOAT128_TYPE__.
13177 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13178 -mfloat128-type option and make -mfloat128 default on PowerPC
13180 (TARGET_FLOAT128_TYPE): Likewise.
13181 (-mfloat128-type): Likewise.
13182 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13183 Delete the -mfloat128-type option and make -mfloat128 default on
13184 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13185 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13186 128-bit floating point is enabled. Change tests from using
13187 -mfloat128-type to -mfloat128.
13188 (rs6000_mangle_type): Use the correct mangling for the __float128
13189 type even if normal long double is restricted to 64-bits.
13190 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13192 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13193 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13194 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13196 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13197 documentation for -mfloat128.
13199 2017-09-06 Olivier Hainque <hainque@adacore.com>
13201 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13203 2017-09-06 Wish Wu <wishwu007@gmail.com>
13204 Jakub Jelinek <jakub@redhat.com>
13206 * asan.c (initialize_sanitizer_builtins): Add
13207 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13208 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13209 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13210 BT_FN_VOID_UINT64_PTR variables.
13211 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13212 (BT_FN_VOID_UINT16_UINT16): Likewise.
13213 (BT_FN_VOID_UINT32_UINT32): Likewise.
13214 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13215 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13216 (BT_FN_VOID_UINT64_PTR): Likewise.
13217 * common.opt (flag_sanitize_coverage): New variable.
13218 (fsanitize-coverage=trace-pc): Remove.
13219 (fsanitize-coverage=): Add.
13220 * flag-types.h (enum sanitize_coverage_code): New enum.
13221 * fold-const.c (fold_range_test): Disable non-short-circuit
13222 optimization if flag_sanitize_coverage.
13223 (fold_truth_andor): Likewise.
13224 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13225 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13226 (coverage_sanitizer_opts): New array.
13227 (get_closest_sanitizer_option): Add OPTS argument, handle also
13228 OPT_fsanitize_coverage_.
13229 (parse_sanitizer_options): Adjusted to also handle
13230 OPT_fsanitize_coverage_.
13231 (common_handle_option): Add OPT_fsanitize_coverage_.
13232 * sancov.c (instrument_comparison, instrument_switch): New function.
13233 (sancov_pass): Add trace-cmp support.
13234 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13235 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13236 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13237 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13238 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13239 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13240 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13241 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13242 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13243 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13245 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13247 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13248 error. Only quit immediately if parsing is complete.
13249 (BEGIN): Initialize fatal_err and parse_done.
13250 (begin fpu, end fpu): Check number of arguments.
13251 (begin arch, end arch): Likewise.
13252 (begin cpu, end cpu): Likewise.
13253 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13254 (optalias): Likewise.
13256 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13258 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13259 * config/arm/arm-isa.h: Delete. Move definitions to ...
13260 * arm-cpus.in: ... here. Use new feature and fgroup values.
13261 * config/arm/arm.c (arm_option_override): Use lower case for feature
13263 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13264 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13265 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13267 (print_isa_bits_for): New function.
13268 (gen_isa): New function.
13269 (gen_comm_data): Use print_isa_bits_for.
13270 (define feature): New keyword.
13271 (define fgroup): New keyword.
13272 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13273 (arm-isa.h): Add rule to generate file.
13274 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13275 case for feature bit names.
13277 2017-09-06 Richard Biener <rguenther@suse.de>
13279 * tree-ssa-pre.c (NECESSARY): Remove.
13280 (create_expression_by_pieces): Do not touch pass-local flags.
13281 (insert_into_preds_of_block): Likewise.
13282 (do_pre_regular_insertion): Likewise.
13283 (eliminate_insert): Likewise.
13284 (eliminate_dom_walker::before_dom_children): Likewise.
13285 (fini_eliminate): Do not look at inserted_exprs.
13286 (mark_operand_necessary): Remove.
13287 (remove_dead_inserted_code): Replace with simple work-list
13288 algorithm based on inserted_exprs and SSA uses.
13289 (pass_pre::execute): Re-order fini_eliminate and
13290 remove_dead_inserted_code.
13292 2017-09-06 Olivier Hainque <hainque@adacore.com>
13294 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13295 for VxWorks 7. Adjust surrounding comments.
13297 2017-09-06 Richard Biener <rguenther@suse.de>
13299 * gimple-ssa-strength-reduction.c
13300 (find_candidates_dom_walker::before_dom_children): Also allow
13303 2017-09-06 Richard Biener <rguenther@suse.de>
13305 PR tree-optimization/82108
13306 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13307 for gap in the non-permutation SLP case.
13309 2017-09-06 Martin Jambor <mjambor@suse.cz>
13311 PR tree-optimization/82078
13312 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13313 add_access_to_work_queue...
13314 (build_accesses_from_assign): ...here.
13315 (propagate_all_subaccesses): Make sure racc is the group
13316 representative, if there is one.
13318 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13320 PR middle-end/82095
13321 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13324 2017-09-06 Richard Biener <rguenther@suse.de>
13326 * gimple-ssa-strength-reduction.c
13327 (find_candidates_dom_walker::before_doom_children): Use a
13328 type and not a mode check.
13330 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13333 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13335 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13337 2017-09-05 Jeff Law <law@redhat.com>
13339 PR tree-optimization/64910
13340 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13341 cases where we have 3 or more operands.
13343 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13345 PR middle-end/81768
13346 * omp-low.c (lower_omp_for): Recompute tree invariant if
13347 gimple_omp_for_initial/final is ADDR_EXPR.
13349 PR middle-end/81768
13350 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13351 into gimple val before gimplification fo the COND_EXPR.
13353 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13355 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13356 REGION_COPY argument.
13357 (thread_through_all_blocks): Remove unused argument to
13358 duplicate_thread_path.
13360 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13361 Alan Hayward <alan.hayward@arm.com>
13362 David Sherwood <david.sherwood@arm.com>
13364 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13365 Take a scalar_mode rather than a machine_mode.
13366 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13367 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13368 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13369 (aarch64_gen_adjusted_ldpstp): Likewise.
13370 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13372 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13373 Alan Hayward <alan.hayward@arm.com>
13374 David Sherwood <david.sherwood@arm.com>
13376 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13377 Take a scalar_int_mode instead of a machine_mode.
13378 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13379 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13380 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13381 (aarch64_simd_attr_length_rglist): Delete.
13382 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13383 a scalar_int_mode instead of a machine_mode.
13384 (aarch64_add_offset): Likewise.
13385 (aarch64_internal_mov_immediate): Likewise
13386 (aarch64_add_constant_internal): Likewise.
13387 (aarch64_add_constant): Likewise.
13388 (aarch64_movw_imm): Likewise.
13389 (aarch64_rtx_arith_op_extract_p): Likewise.
13390 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13391 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13392 Remove assert that the mode isn't a vector.
13393 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13394 (aarch64_expand_mov_immediate): Update calls after above changes.
13395 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13396 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13397 (aarch64_move_imm): Likewise.
13398 (aarch64_can_const_movi_rtx_p): Likewise.
13399 (aarch64_strip_extend): Likewise.
13400 (aarch64_extr_rtx_p): Likewise.
13401 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13402 a CONST_INT when the mode parameter is VOIDmode.
13403 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13405 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13407 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13408 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13409 (bitwise_type_for_mode): Update accordingly.
13411 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13413 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13414 * stor-layout.c (mode_for_size_tree): Likewise.
13415 (mode_for_array): Update accordingly.
13416 (layout_decl): Likewise.
13417 (compute_record_mode): Likewise. Only set the mode once.
13419 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13421 * target.def (get_mask_mode): Change return type to opt_mode.
13423 * doc/tm.texi: Regenerate.
13424 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13425 * targhooks.c (default_get_mask_mode): Likewise.
13426 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13427 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13428 targetm.get_mask_mode.
13429 * tree.c (build_truth_vector_type): Likewise.
13431 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13433 * machmode.h (mode_for_vector): Return an opt_mode.
13434 * stor-layout.c (mode_for_vector): Likewise.
13435 (mode_for_int_vector): Update accordingly.
13436 (layout_type): Likewise.
13437 * config/i386/i386.c (emit_memmov): Likewise.
13438 (ix86_expand_set_or_movmem): Likewise.
13439 (ix86_expand_vector_init): Likewise.
13440 (ix86_get_mask_mode): Likewise.
13441 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13443 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13444 * expmed.c (extract_bit_field_1): Likewise.
13445 * expr.c (expand_expr_real_2): Likewise.
13446 * optabs-query.c (can_vec_perm_p): Likewise.
13447 (can_vec_mask_load_store_p): Likewise.
13448 * optabs.c (expand_vec_perm): Likewise.
13449 * targhooks.c (default_get_mask_mode): Likewise.
13450 * tree-vect-stmts.c (vectorizable_store): Likewise.
13451 (vectorizable_load): Likewise.
13452 (get_vectype_for_scalar_type_and_size): Likewise.
13454 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13456 * machmode.h (mode_for_int_vector): New function.
13457 * stor-layout.c (mode_for_int_vector): Likewise.
13458 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13459 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13460 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13461 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13462 (s390_expand_vcond): Likewise.
13464 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13466 * machmode.h (opt_machine_mode): New type.
13467 (opt_mode<T>): Allow construction from anything that can be
13469 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13470 (mode_for_size): Return an opt_machine_mode.
13471 * stor-layout.c (mode_for_size): Likewise.
13472 (mode_for_size_tree): Update call accordingly.
13473 (bitwise_mode_for_mode): Likewise.
13474 (make_fract_type): Likewise.
13475 (make_accum_type): Likewise.
13476 * caller-save.c (replace_reg_with_saved_mem): Update call
13478 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13479 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13480 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13481 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13482 * expmed.c (extract_bit_field_1): Likewise.
13483 * reload.c (get_secondary_mem): Likewise.
13484 * varasm.c (assemble_integer): Likewise.
13485 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13488 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13490 * machmode.h (decimal_float_mode_for_size): New function.
13491 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13492 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13493 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13494 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13495 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13496 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13498 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13500 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13501 (get_builtin_sync_mode): Likewise.
13502 (expand_ifn_atomic_compare_exchange): Likewise.
13503 (expand_builtin_atomic_clear): Likewise.
13504 (expand_builtin_atomic_test_and_set): Likewise.
13505 (fold_builtin_atomic_always_lock_free): Likewise.
13506 * calls.c (compute_argument_addresses): Likewise.
13507 (emit_library_call_value_1): Likewise.
13508 (store_one_arg): Likewise.
13509 * combine.c (combine_instructions): Likewise.
13510 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13511 * config/arm/arm.c (arm_function_value): Likewise.
13512 (aapcs_allocate_return_reg): Likewise.
13513 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13514 * config/i386/i386.c (construct_container): Likewise.
13515 (ix86_gimplify_va_arg): Likewise.
13516 (ix86_expand_sse_cmp): Likewise.
13517 (emit_memmov): Likewise.
13518 (emit_memset): Likewise.
13519 (expand_small_movmem_or_setmem): Likewise.
13520 (ix86_expand_pextr): Likewise.
13521 (ix86_expand_pinsr): Likewise.
13522 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13523 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13525 * config/mips/mips.c (mips_function_value_1) Likewise.
13526 (mips_block_move_straight): Likewise.
13527 (mips_expand_ins_as_unaligned_store): Likewise.
13528 * config/powerpcspe/powerpcspe.c
13529 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13530 (rs6000_darwin64_record_arg_flush): Likewise.
13531 * config/rs6000/rs6000.c
13532 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13533 (rs6000_darwin64_record_arg_flush): Likewise.
13534 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13535 (sparc_function_value_1): Likewise.
13536 * config/spu/spu.c (adjust_operand): Likewise.
13537 (spu_emit_branch_or_set): Likewise.
13538 (arith_immediate_p): Likewise.
13539 * emit-rtl.c (gen_lowpart_common): Likewise.
13540 * expr.c (expand_expr_real_1): Likewise.
13541 * function.c (assign_parm_setup_block): Likewise.
13542 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13543 * reload1.c (alter_reg): Likewise.
13544 * stor-layout.c (mode_for_vector): Likewise.
13545 (layout_type): Likewise.
13547 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13549 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13550 (spu_convert_move): Likewise.
13551 * lower-subreg.c (resolve_simple_move): Likewise.
13553 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13556 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13557 define_insn to a define_expand.
13558 (altivec_vsum2sws_direct): New define_insn.
13559 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13561 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13563 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13566 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13571 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13572 realign stack if stack alignment needed is less than incoming
13575 2017-09-05 Marek Polacek <polacek@redhat.com>
13578 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13581 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13583 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13585 2017-09-05 Richard Biener <rguenther@suse.de>
13587 PR tree-optimization/82084
13588 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13590 2017-09-05 Richard Biener <rguenther@suse.de>
13592 PR tree-optimization/82102
13593 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13595 2017-09-05 Martin Liska <mliska@suse.cz>
13597 PR tree-optimization/82032
13598 * tree-cfg.c (generate_range_test): New function.
13599 * tree-cfg.h (generate_range_test): Declared here.
13600 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13601 (cleanup_control_expr_graph): Use it.
13602 * tree-switch-conversion.c (try_switch_expansion): Remove
13604 (emit_case_nodes): Use generate_range_test.
13606 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13609 * config/i386/i386.md (*<btsc><mode>_mask): Add
13610 TARGET_USE_BT to insn constraint.
13611 (*btr<mode>_mask): Ditto.
13613 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13615 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13616 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13618 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13621 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13622 TARGET_NEON and TARGET_IWMMXT.
13623 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13624 TARGET_NEON and TARGET_IWMMXT.
13625 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13627 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13629 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13630 (ix86_rewrite_tls_address): Ditto.
13631 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13632 (ix86_rewrite_tls_address_1): Ditto.
13633 (ix86_rewrite_tls_address): Ditto.
13634 * config/i386/predicates.md (tls_address_pattern): New predicate.
13635 * config/i386/i386.md (TLS address splitter): New splitter.
13637 2017-09-04 Richard Biener <rguenther@suse.de>
13639 PR tree-optimization/82084
13640 * fold-const.h (can_native_encode_string_p): Declare.
13641 * fold-const.c (can_native_encode_string_p): Factor out from ...
13642 (native_encode_string): ... here.
13643 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13644 vectorizing stores from constants we later cannot handle.
13646 2017-09-04 Marek Polacek <polacek@redhat.com>
13649 * doc/invoke.texi: Update -Wtautological-compare documentation.
13651 2017-09-04 Jeff Law <law@redhat.com>
13653 PR tree-optimization/64910
13654 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13655 swap the first and last operand if the last is a constant.
13657 2017-09-04 Marek Polacek <polacek@redhat.com>
13660 * convert.c (do_narrow): When sanitizing signed integer overflows,
13661 bail out for signed types.
13662 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13664 2017-09-04 Richard Biener <rguenther@suse.de>
13666 PR tree-optimization/82060
13667 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13668 Move devirtualization after stmt folding and before EH/AB/noreturn
13669 cleanup to get the stmt refs canonicalized. Use a bool instead
13670 of gimple_modified_p since that doesn't work for NOPs. Schedule
13671 NOPs generated by folding for removal.
13673 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13674 Alan Hayward <alan.hayward@arm.com>
13675 David Sherwood <david.sherwood@arm.com>
13677 * coretypes.h (pad_direction): New enum.
13678 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13679 (FUNCTION_ARG_PADDING): Likewise.
13680 * target.def (function_arg_padding): New hook.
13681 * targhooks.h (default_function_arg_padding): Declare.
13682 * targhooks.c (default_function_arg_padding): New function.
13683 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13684 (TARGET_FUNCTION_ARG_PADDING): ...this.
13685 * doc/tm.texi: Regenerate.
13686 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13687 instead of direction.
13688 (compute_argument_addresses): Likewise.
13689 (load_register_parameters): Likewise.
13690 (emit_library_call_value_1): Likewise.
13691 (store_one_arg): Use targetm.calls.function_arg_padding instead
13692 of FUNCTION_ARG_PADDING.
13693 (must_pass_in_stack_var_size_or_pad): Likewise.
13694 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13695 (emit_group_store): Likewise.
13696 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13697 instead of FUNCTION_ARG_PADDING.
13698 (emit_push_insn): Likewise, and propagate enum change throughout
13700 * function.h (direction): Delete.
13701 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13703 * function.c (assign_parm_find_stack_rtl): Likewise.
13704 (assign_parm_setup_block_p): Likewise.
13705 (assign_parm_setup_block): Likewise.
13706 (gimplify_parameters): Likewise.
13707 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13708 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13710 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13711 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13712 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13713 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13714 (aarch64_function_arg_padding): ...this new function.
13715 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13716 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13717 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13718 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13719 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13720 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13721 (arm_pad_arg_upward): Replace with...
13722 (arm_function_arg_padding): ...this new function.
13723 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13725 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13726 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13727 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13728 (ia64_hpux_function_arg_padding): Replace with...
13729 (ia64_function_arg_padding): ...this new function. Use pad_direction
13730 instead of direction. Check for TARGET_HPUX.
13731 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13732 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13733 (iq2000_function_arg_padding): New function.
13734 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13735 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13736 (mips_function_arg_padding): ...this new function.
13737 (mips_pad_reg_upward): Update accordingly.
13738 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13739 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13740 targetm.calls.function_arg_padding.
13741 (FUNCTION_ARG_PADDING): Delete.
13742 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13743 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13744 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13745 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13746 (nios2_block_reg_padding): Return pad_direction instead of direction.
13747 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13748 instead of direction.
13749 (nios2_function_arg_padding): Likewise. Make static.
13750 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13751 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13752 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13753 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13754 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13755 (pa_function_arg_padding): Make static. Return pad_direction instead
13757 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13758 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13759 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13760 instead of direction. Use targetm.calls.function_arg_padding.
13761 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13762 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13763 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13764 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13765 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13767 (function_arg_padding): Rename to...
13768 (rs6000_function_arg_padding): ...this. Make static. Return
13769 pad_direction instead of direction.
13770 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13771 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13772 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13773 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13774 instead of direction. Use targetm.calls.function_arg_padding.
13775 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13776 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13777 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13778 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13779 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13780 (function_arg_padding): Rename to...
13781 (rs6000_function_arg_padding): ...this. Make static. Return
13782 pad_direction instead of direction.
13783 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13784 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13785 * config/s390/s390.c (s390_function_arg_padding): New function.
13786 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13787 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13788 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13789 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13790 (function_arg_padding): Rename to...
13791 (sparc_function_arg_padding): ...this. Make static. Return
13792 pad_direction instead of direction.
13793 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13794 * config/spu/spu.c (spu_function_arg_padding): New function.
13795 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13796 * system.h (FUNCTION_ARG_PADDING): Poison.
13798 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13799 Alan Hayward <alan.hayward@arm.com>
13800 David Sherwood <david.sherwood@arm.com>
13802 * target.def (modes_tieable_p): New hook.
13803 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13804 (TARGET_MODES_TIEABLE_P): ...this.
13805 * doc/tm.texi.in: Regenerate.
13806 * hooks.h (hook_bool_mode_mode_true): Declare.
13807 * hooks.c (hook_bool_mode_mode_true): New function.
13808 * combine.c (subst): Use targetm.modes_tieable_p instead of
13810 * dse.c (find_shift_sequence): Likewise.
13811 * expmed.c (extract_low_bits): Likewise.
13812 * lower-subreg.c: Include target.h.
13813 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13815 * rtlanal.c (rtx_cost): Likewise.
13816 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13817 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13818 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13819 (TARGET_MODES_TIEABLE_P): Redefine.
13820 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13821 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13822 (TARGET_MODES_TIEABLE_P): Redefine.
13823 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13824 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13825 (arc_modes_tieable_p): New function.
13826 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13827 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13828 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13829 (arm_modes_tieable_p): Make static.
13830 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13831 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13832 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13833 (TARGET_MODES_TIEABLE_P): Redefine.
13834 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13835 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13836 (TARGET_MODES_TIEABLE_P): Redefine.
13837 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13838 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13839 (cr16_modes_tieable_p): New function.
13840 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13841 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13842 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13843 (TRULY_NOOP_TRUNCATION): Update comment.
13844 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13845 (TRULY_NOOP_TRUNCATION): Update comment.
13846 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13847 (frv_modes_tieable_p): New function.
13848 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13849 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13850 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13851 (TARGET_MODES_TIEABLE_P): Redefine.
13852 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13853 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13854 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13855 (TARGET_MODES_TIEABLE_P): Redefine.
13856 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13857 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13858 (ia64_modes_tieable_p): New function.
13859 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13860 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13861 (iq2000_modes_tieable_p): New function.
13862 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13863 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13864 (lm32_modes_tieable_p): New function.
13865 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13866 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13867 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13868 (TARGET_MODES_TIEABLE_P): Redefine.
13869 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13870 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13871 (m32r_modes_tieable_p): New function.
13872 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13873 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13874 (m68k_modes_tieable_p): New function.
13875 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13876 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13877 (mcore_modes_tieable_p): New function.
13878 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13879 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13881 (TARGET_MODES_TIEABLE_P): Redefine.
13882 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13883 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13884 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13885 (TARGET_MODES_TIEABLE_P): Redefine.
13886 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13887 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13888 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13889 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13890 (mn10300_modes_tieable_p): ...this and make static.
13891 (TARGET_MODES_TIEABLE_P): Redefine.
13892 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13893 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13894 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13895 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13896 (msp430_modes_tieable_p): Make static.
13897 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13898 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13899 (TARGET_MODES_TIEABLE_P): Redefine.
13900 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13901 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13902 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13903 (TARGET_MODES_TIEABLE_P): Redefine.
13904 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13905 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13906 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13907 (TARGET_MODES_TIEABLE_P): Redefine.
13908 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13909 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13910 (pdp11_modes_tieable_p): New function.
13911 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13912 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13913 (rs6000_modes_tieable_p): New function.
13914 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13915 * config/powerpcspe/powerpcspe.md: Update comment.
13916 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13917 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13918 (TARGET_MODES_TIEABLE_P): Redefine.
13919 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13920 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13921 (rl78_modes_tieable_p): New function.
13922 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13923 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13924 (rs6000_modes_tieable_p): New function.
13925 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13926 * config/rs6000/rs6000.md: Update comment.
13927 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13928 * config/rx/rx.c (rx_modes_tieable_p): New function.
13929 (TARGET_MODES_TIEABLE_P): Redefine.
13930 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13931 * config/s390/s390.c (s390_modes_tieable_p): New function.
13932 (TARGET_MODES_TIEABLE_P): Redefine.
13933 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13934 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13935 (sh_modes_tieable_p): New function.
13936 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13937 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13938 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13939 (sparc_modes_tieable_p): Make static.
13940 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13941 * config/spu/spu.c (spu_modes_tieable_p): New function.
13942 (TARGET_MODES_TIEABLE_P): Redefine.
13943 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13944 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13945 (TARGET_MODES_TIEABLE_P): Redefine.
13946 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13947 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13948 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13949 * config/v850/v850.c (v850_modes_tieable_p): New function.
13950 (TARGET_MODES_TIEABLE_P): Redefine.
13951 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13952 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13953 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13954 (visium_modes_tieable_p): New function.
13955 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13956 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13957 (xtensa_modes_tieable_p): New function.
13958 * system.h (MODES_TIEABLE_P): Poison.
13960 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13961 Alan Hayward <alan.hayward@arm.com>
13962 David Sherwood <david.sherwood@arm.com>
13964 * target.def (hard_regno_mode_ok): New hook.
13965 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13966 (TARGET_HARD_REGNO_MODE_OK): ...this.
13967 * doc/tm.texi.in: Regenerate.
13968 * hooks.h (hook_bool_uint_mode_true): Declare.
13969 * hooks.c (hook_bool_uint_mode_true): New function.
13970 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13971 HARD_REGNO_MODE_OK.
13972 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13973 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13974 instead of HARD_REGNO_MODE_OK.
13975 * caller-save.c: Include target.h.
13976 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13977 HARD_REGNO_MODE_OK.
13978 * combine.c (can_combine_p): Likewise.
13979 (combinable_i3pat): Likewise.
13980 (can_change_dest_mode): Likewise.
13981 * expr.c (init_expr_target): Likewise.
13982 (convert_move): Likewise.
13983 (convert_modes): Likewise.
13984 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13985 (setup_prohibited_mode_move_regs): Likewise.
13986 * ira.h (target_ira): Likewise.
13987 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13988 * lra-constraints.c (process_alt_operands): Likewise.
13989 (split_reg): Likewise.
13990 * recog.c (peep2_find_free_register): Likewise.
13991 * ree.c (combine_reaching_defs): Likewise.
13992 * regcprop.c (maybe_mode_change): Likewise.
13993 * reginfo.c (init_reg_sets_1): Likewise.
13994 (choose_hard_reg_mode): Likewise.
13995 (simplifiable_subregs): Likewise.
13996 * regrename.c (check_new_reg_p): Likewise.
13997 * reload.c (find_valid_class): Likewise.
13998 (find_valid_class_1): Likewise.
13999 (reload_inner_reg_of_subreg): Likewise.
14000 (push_reload): Likewise.
14001 (combine_reloads): Likewise.
14002 (find_dummy_reload): Likewise.
14003 (find_reloads): Likewise.
14004 * reload1.c (find_reg): Likewise.
14005 (set_reload_reg): Likewise.
14006 (allocate_reload_reg): Likewise.
14007 (choose_reload_regs): Likewise.
14008 (reload_adjust_reg_for_temp): Likewise.
14009 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14010 (simplify_subreg_regno): Likewise.
14011 * sel-sched.c (init_regs_for_mode): Likewise.
14012 * varasm.c (make_decl_rtl): Likewise.
14013 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14014 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14015 HARD_REGNO_MODE_OK.
14016 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14017 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14018 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14019 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14020 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14021 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14022 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14023 (arc_mode_class): Delete.
14024 (HARD_REGNO_MODE_OK): Delete.
14025 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14026 (arc_hard_regno_mode_ok): Rename old array to...
14027 (arc_hard_regno_mode_ok_modes): ...this.
14028 (arc_conditional_register_usage): Update accordingly.
14029 (arc_mode_class): Make static.
14030 (arc_hard_regno_mode_ok): New function.
14031 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14032 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14033 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14034 (arm_hard_regno_mode_ok): Make static.
14035 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14036 HARD_REGNO_MODE_OK.
14037 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14038 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14039 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14041 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14042 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14043 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14044 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14045 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14046 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14047 * config/bfin/predicates.md (valid_reg_operand): Use
14048 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14049 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14050 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14051 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14052 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14053 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14054 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14055 (cr16_hard_regno_mode_ok): Make static and return a bool.
14056 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14057 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14058 (cris_hard_regno_mode_ok): New function.
14059 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14060 (epiphany_mode_class): Delete.
14061 (HARD_REGNO_MODE_OK): Delete.
14062 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14063 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14064 (hard_regno_mode_ok): Rename to...
14065 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14066 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14067 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14068 HARD_REGNO_MODE_OK.
14069 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14070 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14071 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14072 (frv_hard_regno_mode_ok): Make static and return a bool.
14073 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14074 HARD_REGNO_MODE_OK.
14075 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14076 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14077 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14078 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14080 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14081 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14082 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14083 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14085 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14086 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14087 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14088 (ia64_hard_regno_mode_ok): New function.
14089 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14090 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14091 (iq2000_hard_regno_mode_ok): New function.
14092 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14093 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14094 (lm32_hard_regno_mode_ok): New function.
14095 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14096 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14097 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14098 instead of HARD_REGNO_MODE_OK.
14099 (m32c_hard_regno_ok): Rename to...
14100 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14101 (m32c_cannot_change_mode_class): Update accordingly.
14102 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14103 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14104 (m32r_mode_class): Delete.
14105 (HARD_REGNO_MODE_OK): Delete.
14106 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14107 (m32r_hard_regno_mode_ok): Rename to...
14108 (m32r_hard_regno_modes): ...this.
14109 (m32r_mode_class): Make static.
14110 (m32r_hard_regno_mode_ok): New function.
14111 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14112 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14113 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14114 (m68k_hard_regno_mode_ok): Make static.
14115 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14116 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14117 (mcore_hard_regno_mode_ok): New function.
14118 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14119 (HARD_REGNO_MODE_OK): Delete.
14120 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14122 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14123 (microblaze_hard_regno_mode_ok): New function.
14124 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14125 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14126 (mips_hard_regno_mode_ok): Delete.
14127 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14128 (mips_hard_regno_mode_ok_p): ...this and make static.
14129 (mips_hard_regno_mode_ok_p): Rename to...
14130 (mips_hard_regno_mode_ok_uncached): ...this.
14131 (mips_hard_regno_mode_ok): New function.
14132 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14133 of HARD_REGNO_MODE_OK.
14134 (mips_option_override): Update after above name changes.
14135 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14136 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14137 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14138 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14139 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14140 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14141 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14142 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14143 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14144 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14145 (msp430_hard_regno_mode_ok): Make static and return a bool.
14146 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14147 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14148 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14150 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14151 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14152 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14153 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14154 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14155 (PA_HARD_REGNO_MODE_OK): ...this
14156 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14157 (PA_HARD_REGNO_MODE_OK): ...this.
14158 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14159 (pa_hard_regno_mode_ok): New function.
14160 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14161 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14162 (pdp11_hard_regno_mode_ok): New function.
14163 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14164 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14166 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14168 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14169 (rs6000_hard_regno_mode_ok): Rename to...
14170 (rs6000_hard_regno_mode_ok_uncached): ...this.
14171 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14172 (rs6000_hard_regno_mode_ok): New function.
14173 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14174 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14175 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14176 (riscv_hard_regno_mode_ok): ...this and make static.
14177 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14178 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14179 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14180 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14181 (rl78_hard_regno_mode_ok): Make static and return bool.
14182 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14183 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14185 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14186 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14187 (rs6000_hard_regno_mode_ok): Rename to...
14188 (rs6000_hard_regno_mode_ok_uncached): ...this.
14189 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14190 (rs6000_hard_regno_mode_ok): New function.
14191 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14192 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14193 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14194 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14195 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14196 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14197 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14198 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14199 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14200 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14201 (sh_hard_regno_mode_ok): Make static.
14202 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14203 instead of HARD_REGNO_MODE_OK.
14204 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14205 (sparc_mode_class): Delete.
14206 (HARD_REGNO_MODE_OK): Delete.
14207 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14208 (hard_regno_mode_classes): Make static.
14209 (sparc_mode_class): Likewise.
14210 (sparc_hard_regno_mode_ok): New function.
14211 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14212 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14213 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14215 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14216 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14217 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14218 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14219 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14220 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14221 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14222 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14223 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14224 (visium_hard_regno_mode_ok): New function.
14225 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14226 instead of HARD_REGNO_MODE_OK.
14227 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14228 (HARD_REGNO_MODE_OK): Delete.
14229 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14230 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14231 (xtensa_option_override): Update accordingly.
14232 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14233 (xtensa_hard_regno_mode_ok): New function.
14234 * system.h (HARD_REGNO_MODE_OK): Poison.
14236 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14237 Alan Hayward <alan.hayward@arm.com>
14238 David Sherwood <david.sherwood@arm.com>
14240 * target.def (hard_regno_call_part_clobbered): New hook.
14241 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14242 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14243 * doc/tm.texi: Regenerate.
14244 * hooks.h (hook_bool_uint_mode_false): Declare.
14245 * hooks.c (hook_bool_uint_mode_false): New function.
14246 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14247 * cselib.c (cselib_process_insn): Use
14248 targetm.hard_regno_call_part_clobbered instead of
14249 HARD_REGNO_CALL_PART_CLOBBERED.
14250 * ira-conflicts.c (ira_build_conflicts): Likewise.
14251 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14252 * lra-constraints.c (need_for_call_save_p): Likewise.
14253 * lra-lives.c: Include target.h.
14254 (check_pseudos_live_through_calls): Use
14255 targetm.hard_regno_call_part_clobbered instead of
14256 HARD_REGNO_CALL_PART_CLOBBERED.
14257 * regcprop.c: Include target.h.
14258 (copyprop_hardreg_forward_1): Use
14259 targetm.hard_regno_call_part_clobbered instead of
14260 HARD_REGNO_CALL_PART_CLOBBERED.
14261 * reginfo.c (choose_hard_reg_mode): Likewise.
14262 * regrename.c (check_new_reg_p): Likewise.
14263 * reload.c (find_equiv_reg): Likewise.
14264 * reload1.c (emit_reload_insns): Likewise.
14265 * sched-deps.c (deps_analyze_insn): Likewise.
14266 * sel-sched.c (init_regs_for_mode): Likewise.
14267 (mark_unavailable_hard_regs): Likewise.
14268 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14269 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14270 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14272 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14273 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14274 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14276 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14278 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14279 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14280 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14282 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14283 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14284 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14286 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14287 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14289 * config/powerpcspe/powerpcspe.c
14290 (rs6000_hard_regno_call_part_clobbered): New function.
14291 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14292 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14293 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14295 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14296 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14297 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14299 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14300 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14301 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14303 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14304 Alan Hayward <alan.hayward@arm.com>
14305 David Sherwood <david.sherwood@arm.com>
14307 * rtl.h (subreg_memory_offset): Declare.
14308 * emit-rtl.c (subreg_memory_offset): New function.
14309 * expmed.c (store_bit_field_1): Use it.
14310 * expr.c (undefined_operand_subword_p): Likewise.
14311 * simplify-rtx.c (simplify_subreg): Likewise.
14313 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14315 PR rtl-optimization/57448
14318 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14319 using atomic_load pattern.
14320 (expand_atomic_store): Likewise.
14322 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14325 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14326 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14327 handling. Use replace_call_with_value with NULL instead of
14328 gsi_replace, unlink_stmt_vdef and release_defs.
14330 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14333 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14335 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14338 * rtl.h (emit_library_call_value_1): Declare.
14339 (emit_library_call): Replace declaration with a series of overloads.
14340 Remove the parameter count argument.
14341 (emit_library_call_value): Likewise.
14342 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14343 with an "rtx_mode_t *".
14344 (emit_library_call_value): Delete.
14345 (emit_library_call): Likewise.
14346 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14347 (asan_emit_allocas_unpoison): Likewise.
14348 * builtins.c (expand_builtin_powi): Likewise.
14349 (expand_asan_emit_allocas_unpoison): Likewise.
14350 * cfgexpand.c (expand_main_function): Likewise.
14351 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14352 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14353 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14354 * config/arm/arm.c (arm_trampoline_init): Likewise.
14355 (arm_call_tls_get_addr): Likewise.
14356 (arm_expand_divmod_libfunc): Likewise.
14357 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14358 (smulsi3_highpart): Likewise.
14359 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14360 (c6x_expand_compare): Likewise.
14361 (c6x_expand_movmem): Likewise.
14362 * config/frv/frv.c (frv_trampoline_init): Likewise.
14363 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14364 (ix86_expand_divmod_libfunc): Likewise.
14365 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14366 (ia64_expand_compare): Likewise.
14367 (ia64_profile_hook): Likewise.
14368 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14369 (nonlocal_goto): Likewise.
14370 (restore_stack_nonlocal): Likewise.
14371 * config/m32r/m32r.c (block_move_call): Likewise.
14372 (m32r_trampoline_init): Likewise.
14373 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14374 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14375 (m68k_call_m68k_read_tp): Likewise.
14376 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14377 (microblaze_expand_divide): Likewise.
14378 * config/mips/mips.h (mips_args): Likewise.
14379 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14380 (MIPS_ICACHE_SYNC): Likewise.
14381 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14382 (nios2_trampoline_init): Likewise.
14383 * config/pa/pa.c (hppa_tls_call): Likewise.
14384 (pa_trampoline_init): Likewise.
14385 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14386 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14387 (expand_strn_compare): Likewise.
14388 (rs6000_generate_compare): Likewise.
14389 (rs6000_expand_float128_convert): Likewise.
14390 (output_profile_hook): Likewise.
14391 (rs6000_trampoline_init): Likewise.
14392 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14393 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14394 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14395 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14396 (rs6000_generate_compare): Likewise.
14397 (rs6000_expand_float128_convert): Likewise.
14398 (output_profile_hook): Likewise.
14399 (rs6000_trampoline_init): Likewise.
14400 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14401 * config/sh/sh.c (sh_trampoline_init): Likewise.
14402 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14403 (sparc_emit_float_lib_cmp): Likewise.
14404 (sparc32_initialize_trampoline): Likewise.
14405 (sparc64_initialize_trampoline): Likewise.
14406 (sparc_profile_hook): Likewise.
14407 * config/spu/spu.c (ea_load_store): Likewise.
14408 * config/spu/spu.md (floatunssidf2): Likewise.
14409 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14410 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14411 * config/visium/visium.c (expand_block_move_4): Likewise.
14412 (expand_block_move_2): Likewise.
14413 (expand_block_move_1): Likewise.
14414 (expand_block_set_4): Likewise.
14415 (expand_block_set_2): Likewise.
14416 (expand_block_set_1): Likewise.
14417 (visium_trampoline_init): Likewise.
14418 (visium_profile_hook): Likewise.
14419 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14420 (xtensa_setup_frame_addresses): Likewise.
14421 (xtensa_trampoline_init): Likewise.
14422 * except.c (sjlj_emit_function_enter): Likewise.
14423 (sjlj_emit_function_exit): Likewise.
14424 * explow.c (allocate_dynamic_stack_space): Likewise.
14425 (probe_stack_range): Likewise.
14426 * expr.c (convert_mode_scalar): Likewise.
14427 * optabs.c (expand_binop): Likewise.
14428 (expand_twoval_binop_libfunc): Likewise.
14429 (expand_unop): Likewise.
14430 (prepare_cmp_insn): Likewise.
14431 (prepare_float_lib_cmp): Likewise.
14432 (expand_float): Likewise.
14433 (expand_fix): Likewise.
14434 (expand_fixed_convert): Likewise.
14435 (maybe_emit_sync_lock_test_and_set): Likewise.
14436 (expand_atomic_compare_and_swap): Likewise.
14437 (expand_mem_thread_fence): Likewise.
14438 (expand_atomic_fetch_op): Likewise.
14440 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14442 * doc/generic.texi (OpenACC): Adjust URL.
14443 * doc/invoke.texi (C Dialect Options): Ditto.
14445 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14447 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14448 predicate for operand 1. Add (m,<S>) constraint.
14449 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14450 Prevent memory operand 1 with register operand 2.
14452 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14454 PR rtl-optimization/82024
14455 * combine.c (try_combine): If the combination result is a PARALLEL,
14456 and we only need to retain the SET in there that would be placed
14457 at I2, check that we can place that at I3 instead, before doing so.
14459 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14462 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14464 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14465 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14468 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14469 Jeff Law <law@redhat.com>
14471 * varasm.c (bss_initializer_p): Do not put constants into .bss
14472 (categorize_decl_for_section): Handle bss_initializer_p returning
14473 false when DECL_INITIAL is NULL.
14475 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14478 * config/s390/s390.c (s390_can_inline_p): New function.
14480 2017-09-01 Jeff Law <law@redhat.com>
14482 PR tree-optimization/82052
14483 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14484 Always initialize the returned slot after a hash table miss
14485 when INSERT is true.
14487 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14489 * config/s390/s390.md (mem_signal_fence): Remove.
14490 * doc/md.texi (mem_signal_fence): Remove.
14491 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14493 * target-insns.def (mem_signal_fence): Remove.
14495 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14498 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14501 * asan.c (create_odr_indicator): Strip name encoding from assembler
14502 name before appending it after __odr_asan_.
14504 2017-09-01 Martin Liska <mliska@suse.cz>
14506 PR tree-optimization/82059
14507 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14508 frequency only when an edge is redirected.
14510 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14512 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14513 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14514 (arc_conditional_register_usage): Remove ARC600 lp_count
14516 (arc_file_start): Emit Tag_ARC_CPU_variation.
14517 (arc_can_use_doloop_p): New conditions to use ZOLs.
14518 (hwloop_fail): New function.
14519 (hwloop_optimize): Likewise.
14520 (hwloop_pattern_reg): Likewise.
14521 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14522 (arc_reorg_loops): New function, calls reorg_loops.
14523 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14524 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14526 (arc_loop_hazard): Remove function, functionality moved into
14528 (arc_hazard): Remove arc_loop_hazard call.
14529 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14530 into hwloop_optimize.
14531 (arc_label_align): Remove ZOL handling.
14532 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14533 * config/arc/arc.md (doloop_begin): Remove pattern.
14534 (doloop_begin_i): Likewise.
14535 (doloop_end_i): Likewise.
14536 (doloop_fallback): Likewise.
14537 (doloop_fallback_m): Likewise.
14538 (doloop_end): Reimplement expand.
14539 (arc_lp): New pattern for LP instruction.
14540 (loop_end): New pattern.
14541 (loop_fail): Likewise.
14542 (decrement_and_branch_until_zero): Likewise.
14543 * config/arc/arc.opt (mlpc-width): New option.
14544 * doc/invoke.texi (mlpc-width): Document option.
14546 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14548 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14549 (arc_ccfsm_advance): Fix checking for delay slots.
14550 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14552 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14554 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14556 (movhi_insn): Update store instruction constraint which are saving
14557 6-bit short immediates.
14558 (movsi_insn): Consider also short scaled load operations.
14559 (zero_extendhisi2_i): Use Usd constraint instead of T.
14560 (extendhisi2_i): Add q constraint.
14561 (arc_clzsi2): Add type and length attributes.
14562 (arc_ctzsi2): Likewise.
14563 * config/arc/constraints.md (Usc): Update constraint, the
14564 assembler can parse two relocations for a single instruction.
14566 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14568 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14569 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14571 2017-08-31 Olivier Hainque <hainque@adacore.com>
14573 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14574 match as powerpc-wrs-vxworks*.
14576 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14578 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14579 register constraint for by-element operand.
14580 (aarch64_mls_elt_merge<mode>): Likewise.
14582 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14584 * config/arc/arc.c (arc_can_follow_jump): Check for short
14587 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14589 * config.gcc: Use g.opt for arc.
14590 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14591 functionality moved to ...
14592 (legitimate_scaled_address_p): New function, ...here.
14593 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14594 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14595 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14597 (arc_override_options): Handle G option.
14598 (arc_output_pic_addr_const): Correct function definition.
14599 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14600 (arc_decl_anon_ns_mem_p): Delete.
14601 (arc_in_small_data_p): Overhaul this function to take into
14602 consideration the value given via G option.
14603 (arc_rewrite_small_data_1): Renamed and corrected old
14604 arc_rewrite_small_data function.
14605 (arc_rewrite_small_data): New function.
14606 (small_data_pattern): Don't use pic_offset_table_rtx.
14607 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14608 * config/arc/simdext.md (movmisalignv2hi): Use
14609 prepare_move_operands function.
14611 (movmisalign*): Likewise.
14612 * doc/invoke.texi (ARC options): Document -G option.
14614 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14616 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14618 * config/arc/arc.c (arc_print_operand): Output scalled address for
14619 sdata whenever is possible.
14620 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14621 load/stores are available.
14622 (compact_sda_memory_operand): Check for the alignment required by
14623 code density instructions.
14624 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14626 * config/arc/constraints.md (Usd): Update constraint.
14627 (Us0): New constraint.
14628 (Usc): Update constraint.
14630 2017-08-31 Richard Biener <rguenther@suse.de>
14632 PR middle-end/82054
14633 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14634 function only once.
14636 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14638 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14639 Resize type_signature.
14641 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14642 Alan Hayward <alan.hayward@arm.com>
14643 David Sherwood <david.sherwood@arm.com>
14645 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14646 subregs whose inner modes can be stored in GPRs.
14647 (aarch64_classify_index): Likewise.
14649 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14650 Alan Hayward <alan.hayward@arm.com>
14651 David Sherwood <david.sherwood@arm.com>
14653 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14654 (V_INT_EQUIV): ...this.
14655 (v_cmp_result): Rename to...
14656 (v_int_equiv): ...this.
14657 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14658 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14659 (copysign<mode>3): Likewise.
14660 (aarch64_simd_bsl<mode>_internal): Likewise.
14661 (aarch64_simd_bsl<mode>): Likewise.
14662 (vec_cmp<mode><mode>): Likewise.
14663 (vcond<mode><mode>): Likewise.
14664 (vcond<v_cmp_mixed><mode>): Likewise.
14665 (vcondu<mode><v_cmp_mixed>): Likewise.
14666 (aarch64_cm<optab><mode>): Likewise.
14667 (aarch64_cmtst<mode>): Likewise.
14668 (aarch64_fac<optab><mode>): Likewise.
14669 (vec_perm_const<mode>): Likewise.
14670 (vcond_mask_<mode><v_cmp_result>): Rename to...
14671 (vcond_mask_<mode><v_int_equiv>): ...this.
14672 (vec_cmp<mode><v_cmp_result>): Rename to...
14673 (vec_cmp<mode><v_int_equiv>): ...this.
14675 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14676 Alan Hayward <alan.hayward@arm.com>
14677 David Sherwood <david.sherwood@arm.com>
14679 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14681 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14682 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14683 (UNSPEC_LD4_DREG): New unspecs.
14684 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14685 (aarch64_ld2<mode>_dreg_be): Replace with...
14686 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14688 (aarch64_ld3<mode>_dreg_le)
14689 (aarch64_ld3<mode>_dreg_be): Replace with...
14690 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14692 (aarch64_ld4<mode>_dreg_le)
14693 (aarch64_ld4<mode>_dreg_be): Replace with...
14694 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14697 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14699 PR tree-optimization/81987
14700 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14701 insert an initializer in a location not dominated by the stride
14704 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14706 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14707 on the entire header of the finally block in the fallthru case.
14709 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14711 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14713 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14715 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14716 rs6000_emit_move_from_cr and call renamed function.
14717 (rs6000_emit_prologue): Call renamed functions.
14718 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14719 movesi_from_cr, remove volatile CRs.
14721 2017-08-30 Jon Beniston <jon@beniston.com>
14722 Richard Biener <rguenther@suse.de>
14724 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14725 of VECTOR_MODE_P check.
14726 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14727 element vector types.
14729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14731 * df.h (df_read_modify_subreg_p): Remove in favor of...
14732 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14733 const_rtx instead of an rtx.
14734 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14735 * df-problems.c (df_word_lr_mark_ref): Likewise.
14736 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14737 (mark_pseudo_reg_dead): Likewise.
14738 (mark_ref_dead): Likewise.
14739 * reginfo.c (init_subregs_of_mode): Likewise.
14740 * sched-deps.c (sched_analyze_1): Likewise.
14741 * df-scan.c (df_def_record_1): Likewise.
14742 (df_uses_record): Likewise.
14743 (df_read_modify_subreg_p): Remove in favor of...
14744 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14745 const_rtx instead of an rtx.
14747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14748 Alan Hayward <alan.hayward@arm.com>
14749 David Sherwood <david.sherwood@arm.com>
14751 * rtl.h (partial_subreg_p): New function.
14752 * caller-save.c (save_call_clobbered_regs): Use it.
14753 * calls.c (expand_call): Likewise.
14754 * combine.c (combinable_i3pat): Likewise.
14755 (simplify_set): Likewise.
14756 (make_extraction): Likewise.
14757 (make_compound_operation_int): Likewise.
14758 (gen_lowpart_or_truncate): Likewise.
14759 (force_to_mode): Likewise.
14760 (make_field_assignment): Likewise.
14761 (reg_truncated_to_mode): Likewise.
14762 (record_truncated_value): Likewise.
14763 (move_deaths): Likewise.
14764 * cse.c (record_jump_cond): Likewise.
14765 (cse_insn): Likewise.
14766 * cselib.c (cselib_lookup_1): Likewise.
14767 * expmed.c (extract_bit_field_using_extv): Likewise.
14768 * function.c (assign_parm_setup_reg): Likewise.
14769 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14770 * ira-build.c (create_insn_allocnos): Likewise.
14771 * lra-coalesce.c (merge_pseudos): Likewise.
14772 * lra-constraints.c (match_reload): Likewise.
14773 (simplify_operand_subreg): Likewise.
14774 (curr_insn_transform): Likewise.
14775 * lra-lives.c (process_bb_lives): Likewise.
14776 * lra.c (new_insn_reg): Likewise.
14777 (lra_substitute_pseudo): Likewise.
14778 * regcprop.c (mode_change_ok): Likewise.
14779 (maybe_mode_change): Likewise.
14780 (copyprop_hardreg_forward_1): Likewise.
14781 * reload.c (push_reload): Likewise.
14782 (find_reloads): Likewise.
14783 (find_reloads_subreg_address): Likewise.
14784 * reload1.c (alter_reg): Likewise.
14785 (eliminate_regs_1): Likewise.
14786 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14788 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14790 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14791 back to if statements, including unpack.
14793 2017-08-30 Martin Liska <mliska@suse.cz>
14795 PR inline-asm/82001
14796 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14798 (func_checker::compare_asm_inputs_outputs): ... this function.
14799 (func_checker::compare_gimple_asm): Use the function to compare
14800 also ASM constrains.
14801 * ipa-icf-gimple.h: Rename the function.
14803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14804 Alan Hayward <alan.hayward@arm.com>
14805 David Sherwood <david.sherwood@arm.com>
14807 * coretypes.h (complex_mode): New type.
14808 * gdbhooks.py (build_pretty_printer): Handle it.
14809 * machmode.h (complex_mode): New class.
14810 (complex_mode::includes_p): New function.
14811 (is_complex_int_mode): Likewise.
14812 (is_complex_float_mode): Likewise.
14813 * genmodes.c (get_mode_class): Handle complex mode classes.
14814 * function.c (expand_function_end): Use is_complex_int_mode.
14816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14817 Alan Hayward <alan.hayward@arm.com>
14818 David Sherwood <david.sherwood@arm.com>
14820 * coretypes.h (scalar_mode_pod): New typedef.
14821 * gdbhooks.py (build_pretty_printer): Handle it.
14822 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14823 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14824 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14825 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14826 as_a <scalar_mode>.
14828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14829 Alan Hayward <alan.hayward@arm.com>
14830 David Sherwood <david.sherwood@arm.com>
14832 * machmode.h (mode_for_vector): Take a scalar_mode instead
14834 * stor-layout.c (mode_for_vector): Likewise.
14835 * explow.c (promote_mode): Use as_a <scalar_mode>.
14836 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14839 Alan Hayward <alan.hayward@arm.com>
14840 David Sherwood <david.sherwood@arm.com>
14842 * target.def (preferred_simd_mode): Take a scalar_mode
14843 instead of a machine_mode.
14844 * targhooks.h (default_preferred_simd_mode): Likewise.
14845 * targhooks.c (default_preferred_simd_mode): Likewise.
14846 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14847 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14848 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14849 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14850 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14851 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14852 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14853 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14855 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14856 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14857 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14858 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14859 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14860 * doc/tm.texi: Regenerate.
14861 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14865 Alan Hayward <alan.hayward@arm.com>
14866 David Sherwood <david.sherwood@arm.com>
14868 * target.def (scalar_mode_supported_p): Take a scalar_mode
14869 instead of a machine_mode.
14870 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14871 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14872 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14873 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14874 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14875 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14876 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14877 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14878 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14879 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14880 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14881 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14882 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14883 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14885 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14886 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14887 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14888 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14889 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14891 * doc/tm.texi: Regenerate.
14893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14894 Alan Hayward <alan.hayward@arm.com>
14895 David Sherwood <david.sherwood@arm.com>
14897 * coretypes.h (opt_scalar_mode): New typedef.
14898 * gdbhooks.py (build_pretty_printers): Handle it.
14899 * machmode.h (mode_iterator::get_2xwider): Add overload for
14901 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14903 * expr.c (convert_mode_scalar): Likewise.
14904 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14905 * optabs.c (expand_float): Likewise.
14906 (expand_fix): Likewise.
14907 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14910 Alan Hayward <alan.hayward@arm.com>
14911 David Sherwood <david.sherwood@arm.com>
14913 * optabs.c (expand_float): Explicitly check for scalars before
14914 using a branching expansion.
14915 (expand_fix): Likewise.
14917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14918 Alan Hayward <alan.hayward@arm.com>
14919 David Sherwood <david.sherwood@arm.com>
14921 * expr.c (convert_mode): Split scalar handling out into...
14922 (convert_mode_scalar): ...this new function. Treat the modes
14925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14926 Alan Hayward <alan.hayward@arm.com>
14927 David Sherwood <david.sherwood@arm.com>
14929 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14931 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14934 Alan Hayward <alan.hayward@arm.com>
14935 David Sherwood <david.sherwood@arm.com>
14937 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14938 rather than a machine_mode.
14939 (fixed_from_string): Likewise.
14940 (fixed_convert): Likewise.
14941 (fixed_convert_from_int): Likewise.
14942 (fixed_convert_from_real): Likewise.
14943 (real_convert_from_fixed): Likewise.
14944 * fixed-value.c (fixed_from_double_int): Likewise.
14945 (fixed_from_string): Likewise.
14946 (fixed_convert): Likewise.
14947 (fixed_convert_from_int): Likewise.
14948 (fixed_convert_from_real): Likewise.
14949 (real_convert_from_fixed): Likewise.
14950 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14953 Alan Hayward <alan.hayward@arm.com>
14954 David Sherwood <david.sherwood@arm.com>
14956 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14957 of separate mode class checks. Do not allow vector modes here.
14958 (immed_wide_int_const): Use as_a <scalar_mode>.
14959 * explow.c (trunc_int_for_mode): Likewise.
14960 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14961 (wi::shwi): Likewise.
14962 (wi::min_value): Likewise.
14963 (wi::max_value): Likewise.
14964 * dwarf2out.c (loc_descriptor): Likewise.
14965 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14966 for CONST_WIDE_INT.
14968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14969 Alan Hayward <alan.hayward@arm.com>
14970 David Sherwood <david.sherwood@arm.com>
14972 * tree.h (SCALAR_TYPE_MODE): New macro.
14973 * expr.c (expand_expr_addr_expr_1): Use it.
14974 (expand_expr_real_2): Likewise.
14975 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14976 (fold_convert_const_fixed_from_int): Likewise.
14977 (fold_convert_const_fixed_from_real): Likewise.
14978 (native_encode_fixed): Likewise
14979 (native_encode_complex): Likewise
14980 (native_encode_vector): Likewise.
14981 (native_interpret_fixed): Likewise.
14982 (native_interpret_real): Likewise.
14983 (native_interpret_complex): Likewise.
14984 (native_interpret_vector): Likewise.
14985 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14986 (simd_clone_adjust_argument_types): Likewise.
14987 (simd_clone_init_simd_arrays): Likewise.
14988 (simd_clone_adjust): Likewise.
14989 * stor-layout.c (layout_type): Likewise.
14990 * tree.c (build_minus_one_cst): Likewise.
14991 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14992 * tree-inline.c (estimate_move_cost): Likewise.
14993 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14994 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14995 (vectorizable_reduction): Likewise.
14996 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14997 (vect_recog_mixed_size_cond_pattern): Likewise.
14998 (check_bool_pattern): Likewise.
14999 (adjust_bool_pattern): Likewise.
15000 (search_type_for_mask_1): Likewise.
15001 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15002 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15003 (vectorizable_load): Likewise.
15004 (vectorizable_store): Likewise.
15005 * ubsan.c (ubsan_encode_value): Likewise.
15006 * varasm.c (output_constant): Likewise.
15008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15009 Alan Hayward <alan.hayward@arm.com>
15010 David Sherwood <david.sherwood@arm.com>
15012 * coretypes.h (scalar_mode): New class.
15013 * machmode.h (scalar_mode): Likewise.
15014 (scalar_mode::includes_p): New function.
15015 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15016 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15017 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15018 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15019 * expmed.c (store_bit_field_1): Likewise.
15020 (extract_bit_field_1): Likewise.
15021 * expr.c (write_complex_part): Likewise.
15022 (read_complex_part): Likewise.
15023 (emit_move_complex_push): Likewise.
15024 (expand_expr_real_2): Likewise.
15025 * function.c (assign_parm_setup_reg): Likewise.
15026 (assign_parms_unsplit_complex): Likewise.
15027 * optabs.c (expand_binop): Likewise.
15028 * rtlanal.c (subreg_get_info): Likewise.
15029 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15030 * varasm.c (output_constant_pool_2): Likewise.
15032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15033 Alan Hayward <alan.hayward@arm.com>
15034 David Sherwood <david.sherwood@arm.com>
15036 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15038 (expmed_mult_highpart_optab): Likewise.
15039 (expmed_mult_highpart): Likewise.
15041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15042 Alan Hayward <alan.hayward@arm.com>
15043 David Sherwood <david.sherwood@arm.com>
15045 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15046 instead of a machine_mode.
15047 (builtin_memset_read_str): Likewise.
15048 * builtins.c (c_readstr): Likewise.
15049 (builtin_memcpy_read_str): Likewise.
15050 (builtin_strncpy_read_str): Likewise.
15051 (builtin_memset_read_str): Likewise.
15052 (builtin_memset_gen_str): Likewise.
15053 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15054 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15055 instead of a machine_mode.
15056 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15058 (make_extraction): Likewise.
15059 (try_widen_shift_mode): Take and return scalar_int_modes instead
15061 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15062 a scalar_int_mode instead of a machine_mode.
15063 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15064 (avr_addr_space_pointer_mode): Likewise.
15065 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15066 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15067 (msp430_unwind_word_mode): Likewise.
15068 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15069 (spu_addr_space_pointer_mode): Likewise.
15070 (spu_addr_space_address_mode): Likewise.
15071 (spu_libgcc_cmp_return_mode): Likewise.
15072 (spu_libgcc_shift_count_mode): Likewise.
15073 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15074 (rl78_addr_space_pointer_mode): Likewise.
15075 (fl78_unwind_word_mode): Likewise.
15076 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15078 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15079 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15080 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15081 (mips_valid_pointer_mode): Likewise.
15082 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15083 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15084 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15086 (ft32_addr_space_address_mode): Likewise.
15087 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15088 scalar_int_mode instead of a machine_mode.
15089 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15091 (m32c_addr_space_address_mode): Likewise.
15092 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15093 (rs6000_eh_return_filter_mode): Likewise.
15094 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15095 (rs6000_eh_return_filter_mode): Likewise.
15096 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15097 (s390_libgcc_shift_count_mode): Likewise.
15098 (s390_unwind_word_mode): Likewise.
15099 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15101 * target.def (mode_rep_extended): Likewise.
15102 (valid_pointer_mode): Likewise.
15103 (addr_space.valid_pointer_mode): Likewise.
15104 (eh_return_filter_mode): Return a scalar_int_mode rather than
15106 (libgcc_cmp_return_mode): Likewise.
15107 (libgcc_shift_count_mode): Likewise.
15108 (unwind_word_mode): Likewise.
15109 (addr_space.pointer_mode): Likewise.
15110 (addr_space.address_mode): Likewise.
15111 * doc/tm.texi: Regenerate.
15112 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15114 (do_jump): Use scalar_int_mode for local variables.
15115 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15116 rather than a machine_mode.
15117 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15118 (scompare_loc_descriptor_wide): Likewise.
15119 (scompare_loc_descriptor_narrow): Likewise.
15120 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15122 * except.c (sjlj_emit_dispatch_table): Likewise.
15123 (expand_builtin_eh_copy_values): Likewise.
15124 * explow.c (convert_memory_address_addr_space_1): Likewise.
15125 Take a scalar_int_mode rather than a machine_mode.
15126 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15127 than a machine_mode.
15128 (memory_address_addr_space): Use scalar_int_mode for local variables.
15129 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15130 rather than a machine_mode.
15131 * expmed.c (mask_rtx): Likewise.
15132 (init_expmed_one_conv): Likewise.
15133 (expand_mult_highpart_adjust): Likewise.
15134 (extract_high_half): Likewise.
15135 (expmed_mult_highpart_optab): Likewise.
15136 (expmed_mult_highpart): Likewise.
15137 (expand_smod_pow2): Likewise.
15138 (expand_sdiv_pow2): Likewise.
15139 (emit_store_flag_int): Likewise.
15140 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15142 (extract_low_bits): Likewise.
15143 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15145 * expr.c (pieces_addr::adjust): Likewise.
15146 (can_store_by_pieces): Likewise.
15147 (store_by_pieces): Likewise.
15148 (clear_by_pieces_1): Likewise.
15149 (expand_expr_addr_expr_1): Likewise.
15150 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15151 (expand_expr_real_1): Likewise.
15152 (try_casesi): Likewise.
15153 * final.c (shorten_branches): Likewise.
15154 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15155 type of "mode" to machine_mode.
15156 * internal-fn.c (expand_arith_overflow_result_store): Take a
15157 scalar_int_mode rather than a machine_mode.
15158 (expand_mul_overflow): Use scalar_int_mode for local variables.
15159 * loop-doloop.c (doloop_modify): Likewise.
15160 (doloop_optimize): Likewise.
15161 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15162 than a machine_mode.
15163 (expand_doubleword_shift_condmove): Likewise.
15164 (expand_doubleword_shift): Likewise.
15165 (expand_doubleword_clz): Likewise.
15166 (expand_doubleword_popcount): Likewise.
15167 (expand_doubleword_parity): Likewise.
15168 (expand_absneg_bit): Use scalar_int_mode for local variables.
15169 (prepare_float_lib_cmp): Likewise.
15170 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15171 rather than a machine_mode.
15172 (convert_memory_address_addr_space): Likewise.
15173 (get_mode_bounds): Likewise.
15174 (get_address_mode): Return a scalar_int_mode rather than a
15176 * rtlanal.c (get_address_mode): Likewise.
15177 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15178 than a machine_mode.
15179 * targhooks.c (default_mode_rep_extended): Likewise.
15180 (default_valid_pointer_mode): Likewise.
15181 (default_addr_space_valid_pointer_mode): Likewise.
15182 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15183 than a machine_mode.
15184 (default_libgcc_cmp_return_mode): Likewise.
15185 (default_libgcc_shift_count_mode): Likewise.
15186 (default_unwind_word_mode): Likewise.
15187 (default_addr_space_pointer_mode): Likewise.
15188 (default_addr_space_address_mode): Likewise.
15189 * targhooks.h (default_eh_return_filter_mode): Likewise.
15190 (default_libgcc_cmp_return_mode): Likewise.
15191 (default_libgcc_shift_count_mode): Likewise.
15192 (default_unwind_word_mode): Likewise.
15193 (default_addr_space_pointer_mode): Likewise.
15194 (default_addr_space_address_mode): Likewise.
15195 (default_mode_rep_extended): Take a scalar_int_mode rather than
15197 (default_valid_pointer_mode): Likewise.
15198 (default_addr_space_valid_pointer_mode): Likewise.
15199 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15201 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15202 rather than a machine_mode.
15203 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15204 for local variables.
15205 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15206 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15207 than a machine_mode.
15209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15210 Alan Hayward <alan.hayward@arm.com>
15211 David Sherwood <david.sherwood@arm.com>
15213 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15214 the mode argument to scalar_int_mode.
15215 (do_jump_by_parts_zero_rtx): Likewise.
15216 (do_jump_by_parts_equality_rtx): Likewise.
15217 (do_jump_by_parts_greater): Take a mode argument.
15218 (do_jump_by_parts_equality): Likewise.
15219 (do_jump_1): Update calls accordingly.
15221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15222 Alan Hayward <alan.hayward@arm.com>
15223 David Sherwood <david.sherwood@arm.com>
15225 * is-a.h (safe_dyn_cast): New function.
15226 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15227 (jump_table_for_label): Likewise.
15228 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15229 instead of an rtx_insn *.
15230 (shorten_branches): Use dyn_cast instead of LABEL_P and
15231 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15232 rtx_jump_table_data::get_data_mode.
15233 (final_scan_insn): Likewise.
15235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15236 Alan Hayward <alan.hayward@arm.com>
15237 David Sherwood <david.sherwood@arm.com>
15239 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15240 trying to combine a full-register integer set with a subreg
15243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15244 Alan Hayward <alan.hayward@arm.com>
15245 David Sherwood <david.sherwood@arm.com>
15247 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15248 that is always either address_mode or pointer_mode.
15250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15251 Alan Hayward <alan.hayward@arm.com>
15252 David Sherwood <david.sherwood@arm.com>
15254 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15255 when the two are known to be equal.
15257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15258 Alan Hayward <alan.hayward@arm.com>
15259 David Sherwood <david.sherwood@arm.com>
15261 * simplify-rtx.c (simplify_const_unary_operation): Use
15262 is_a <scalar_int_mode> instead of checking for a nonzero
15263 precision. Forcibly convert op_mode to a scalar_int_mode
15264 in that case. More clearly differentiate the operand and
15265 result modes and use the former when deciding what the value
15266 of a count-bits operation should be. Use is_int_mode instead
15267 of checking for a MODE_INT. Remove redundant check for whether
15268 this mode has a zero precision.
15270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15271 Alan Hayward <alan.hayward@arm.com>
15272 David Sherwood <david.sherwood@arm.com>
15274 * optabs.c (widen_leading): Change the type of the mode argument
15275 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15276 (widen_bswap): Likewise.
15277 (expand_parity): Likewise.
15278 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15279 (expand_ffs): Likewise.
15280 (epand_unop): Check for scalar integer modes before calling the
15283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15284 Alan Hayward <alan.hayward@arm.com>
15285 David Sherwood <david.sherwood@arm.com>
15287 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15289 (expand_expr_real_1): Update call accordingly.
15291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15292 Alan Hayward <alan.hayward@arm.com>
15293 David Sherwood <david.sherwood@arm.com>
15295 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15296 value_mode arguments. Use scalar_int_mode internally.
15297 (store_bit_field_1): Rename the new integer mode from imode
15298 to op0_mode and use it instead of GET_MODE (op0). Update calls
15299 to store_split_bit_field, store_bit_field_using_insv and
15300 store_fixed_bit_field.
15301 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15302 Use scalar_int_mode internally. Use a bit count rather than a mode
15303 when calculating the largest bit size for get_best_mode.
15304 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15305 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15306 Remove assertion that OP0 has a scalar integer mode.
15307 (store_split_bit_field): Add op0_mode and value_mode arguments.
15308 Update calls to extract_fixed_bit_field.
15309 (extract_bit_field_using_extv): Add an op0_mode argument.
15310 Use scalar_int_mode internally.
15311 (extract_bit_field_1): Rename the new integer mode from imode to
15312 op0_mode and use it instead of GET_MODE (op0). Update calls to
15313 extract_split_bit_field, extract_bit_field_using_extv and
15314 extract_fixed_bit_field.
15315 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15316 to extract_split_bit_field and extract_fixed_bit_field_1.
15317 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15318 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15319 on the target mode.
15320 (extract_split_bit_field): Add an op0_mode argument. Update call
15321 to extract_fixed_bit_field.
15323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15324 Alan Hayward <alan.hayward@arm.com>
15325 David Sherwood <david.sherwood@arm.com>
15327 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15328 * explow.c (hard_function_value): Likewise.
15329 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15330 convert_to_mode call outside the loop.
15331 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15332 for the mode iterator. Require the mode specified by max_pieces
15334 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15336 (copy_blkmode_to_reg): Likewise.
15337 (set_storage_via_setmem): Likewise.
15338 * optabs.c (prepare_cmp_insn): Likewise.
15339 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15340 * stor-layout.c (finish_bitfield_representative): Likewise.
15342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15343 Alan Hayward <alan.hayward@arm.com>
15344 David Sherwood <david.sherwood@arm.com>
15346 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15347 * expr.c (convert_move): Use them.
15348 (convert_modes): Likewise.
15349 (store_expr_with_bounds): Likewise.
15351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15352 Alan Hayward <alan.hayward@arm.com>
15353 David Sherwood <david.sherwood@arm.com>
15355 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15356 parameter for the mode of "x". Remove the "known_x", "known_mode"
15357 and "known_ret" arguments. Change the type of the mode argument
15358 to scalar_int_mode.
15359 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15360 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15361 (reg_num_sign_bit_copies_for_combine): Likewise.
15362 * rtlanal.c (nonzero_bits1): Likewise.
15363 (num_sign_bit_copies1): Likewise.
15364 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15365 (reg_num_sign_bit_copies_general): Likewise.
15366 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15367 (reg_nonzero_bits_general): Likewise.
15369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15370 Alan Hayward <alan.hayward@arm.com>
15371 David Sherwood <david.sherwood@arm.com>
15373 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15374 than in subroutines. Return 1 for non-integer modes.
15375 (cached_num_sign_bit_copies): Change the type of the mode parameter
15376 to scalar_int_mode.
15377 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15378 classes. Handle CONST_INT_P first and then check whether X also
15379 has a scalar integer mode. Check the same thing for inner registers
15380 of a SUBREG and for values that are being extended or truncated.
15382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15383 Alan Hayward <alan.hayward@arm.com>
15384 David Sherwood <david.sherwood@arm.com>
15386 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15387 in subroutines. Return the mode mask for non-integer modes.
15388 (cached_nonzero_bits): Change the type of the mode parameter
15389 to scalar_int_mode.
15390 (nonzero_bits1): Likewise. Remove early exit for other mode
15391 classes. Handle CONST_INT_P first and then check whether X
15392 also has a scalar integer mode.
15394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15395 Alan Hayward <alan.hayward@arm.com>
15396 David Sherwood <david.sherwood@arm.com>
15398 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15399 Return a scalar_int_mode and assert that the size is greater than
15401 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15402 (op_by_pieces_d::op_by_pieces_d): Likewise.
15403 (op_by_pieces_d::run): Likewise.
15404 (can_store_by_pieces): Likewise.
15406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15407 Alan Hayward <alan.hayward@arm.com>
15408 David Sherwood <david.sherwood@arm.com>
15410 * combine.c (extract_left_shift): Add a mode argument and update
15412 (make_compound_operation_int): Change the type of the mode parameter
15413 to scalar_int_mode and update the call to extract_left_shift.
15415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15416 Alan Hayward <alan.hayward@arm.com>
15417 David Sherwood <david.sherwood@arm.com>
15419 * combine.c (simplify_and_const_int): Change the type of the mode
15420 parameter to scalar_int_mode.
15421 (simplify_and_const_int_1): Likewise. Update recursive call.
15423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15424 Alan Hayward <alan.hayward@arm.com>
15425 David Sherwood <david.sherwood@arm.com>
15427 * combine.c (simplify_compare_const): Check that the mode is a
15428 scalar_int_mode (rather than VOIDmode) before testing its
15430 (simplify_comparison): Move COMPARISON_P handling out of the
15431 loop and restrict the latter part of the loop to scalar_int_modes.
15432 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15433 and when considering SUBREG_REGs. Use is_int_mode instead of
15434 checking GET_MODE_CLASS against MODE_INT.
15436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15437 Alan Hayward <alan.hayward@arm.com>
15438 David Sherwood <david.sherwood@arm.com>
15440 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15441 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15442 shift_unit_mode and for modes involved in scalar shifts.
15444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15445 Alan Hayward <alan.hayward@arm.com>
15446 David Sherwood <david.sherwood@arm.com>
15448 * combine.c (force_int_to_mode): New function, split out from...
15449 (force_to_mode): ...here. Keep xmode up-to-date and use it
15450 instead of GET_MODE (x).
15452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15453 Alan Hayward <alan.hayward@arm.com>
15454 David Sherwood <david.sherwood@arm.com>
15456 * optabs-query.h (extraction_insn::struct_mode): Change type to
15457 opt_scalar_int_mode and update comment.
15458 (extraction_insn::field_mode): Change type to scalar_int_mode.
15459 (extraction_insn::pos_mode): Likewise.
15460 * combine.c (make_extraction): Update accordingly.
15461 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15462 (get_optab_extraction_insn): Likewise.
15463 * recog.c (simplify_while_replacing): Likewise.
15464 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15465 parameter to opt_scalar_int_mode.
15467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15468 Alan Hayward <alan.hayward@arm.com>
15469 David Sherwood <david.sherwood@arm.com>
15471 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15472 to a scalar_int_mode instead of a machine_mode.
15473 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15474 (get_best_mode): Return a boolean and use a pointer argument to store
15475 the selected mode. Replace the limit mode parameter with a bit limit.
15476 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15477 for the values returned by bit_field_mode_iterator::next_mode.
15478 (store_bit_field): Update call to get_best_mode.
15479 (store_fixed_bit_field): Likewise.
15480 (extract_fixed_bit_field): Likewise.
15481 * expr.c (optimize_bitfield_assignment_op): Likewise.
15482 * fold-const.c (optimize_bit_field_compare): Likewise.
15483 (fold_truth_andor_1): Likewise.
15484 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15485 Update for new type of m_mode.
15486 (get_best_mode): As above.
15488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15489 Alan Hayward <alan.hayward@arm.com>
15490 David Sherwood <david.sherwood@arm.com>
15492 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15493 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15494 (store_bit_field): Check is_a <scalar_int_mode> before calling
15495 strict_volatile_bitfield_p.
15496 (extract_bit_field): Likewise.
15498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15499 Alan Hayward <alan.hayward@arm.com>
15500 David Sherwood <david.sherwood@arm.com>
15502 * target.def (cstore_mode): Return a scalar_int_mode.
15503 * doc/tm.texi: Regenerate.
15504 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15505 * targhooks.h (default_cstore_mode): Likewise.
15506 * targhooks.c (default_cstore_mode): Likewise, using a forced
15508 * expmed.c (emit_cstore): Expect the target of the cstore to be
15511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15512 Alan Hayward <alan.hayward@arm.com>
15513 David Sherwood <david.sherwood@arm.com>
15515 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15517 (niter_desc): Likewise mode.
15518 (iv_analyze): Add a mode parameter.
15520 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15521 and change its type to scalar_int_mode.
15522 * loop-iv.c: Update commentary at head of file.
15523 (iv_constant): Pass the mode paraeter before the rtx it describes
15524 and change its type to scalar_int_mode. Remove VOIDmode handling.
15525 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15526 (iv_extend): Likewise.
15527 (shorten_into_mode): Likewise.
15528 (iv_add): Use scalar_int_mode.
15529 (iv_mult): Likewise.
15530 (iv_shift): Likewise.
15531 (canonicalize_iv_subregs): Likewise.
15532 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15533 it describes and change its mode to scalar_int_mode. Also change
15534 the type of the returned inner_mode to scalar_int_mode.
15535 (get_biv_step): Likewise, turning outer_mode from a pointer
15536 into a direct parameter. Update call to get_biv_step_1.
15537 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15538 iv_constant and get_biv_step.
15539 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15540 and change its type to scalar_int_mode. Don't initialise iv->mode
15541 to VOIDmode and remove later checks for its still being VOIDmode.
15542 Update calls to iv_analyze_op and iv_analyze_expr. Check
15543 is_a <scalar_int_mode> when changing the mode under consideration.
15544 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15545 Update call to iv_analyze_expr.
15546 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15547 inner register is not also a scalar_int_mode. Update call to
15549 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15550 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15551 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15552 separate mode class checks. Update calls to iv_analyze. Remove
15553 fix-up of VOIDmodes after iv_analyze_biv.
15554 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15555 don't have a scalar_int_mode. Update call to biv_p.
15557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15558 Alan Hayward <alan.hayward@arm.com>
15559 David Sherwood <david.sherwood@arm.com>
15561 * cfgexpand.c (convert_debug_memory_address): Use
15562 as_a <scalar_int_mode>.
15563 * combine.c (expand_compound_operation): Likewise.
15564 (make_extraction): Likewise.
15565 (change_zero_ext): Likewise.
15566 (simplify_comparison): Likewise.
15567 * cse.c (cse_insn): Likewise.
15568 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15569 (mem_loc_descriptor): Likewise.
15570 (loc_descriptor): Likewise.
15571 * expmed.c (init_expmed_one_mode): Likewise.
15572 (synth_mult): Likewise.
15573 (emit_store_flag_1): Likewise.
15574 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15575 of a comparison with size.
15576 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15577 (reduce_to_bit_field_precision): Likewise.
15578 * function.c (expand_function_end): Likewise.
15579 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15580 * loop-doloop.c (doloop_modify): Likewise.
15581 * optabs.c (expand_binop): Likewise.
15582 (expand_unop): Likewise.
15583 (expand_copysign_absneg): Likewise.
15584 (prepare_cmp_insn): Likewise.
15585 (maybe_legitimize_operand): Likewise.
15586 * recog.c (const_scalar_int_operand): Likewise.
15587 * rtlanal.c (get_address_mode): Likewise.
15588 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15589 (simplify_cond_clz_ctz): Likewise.
15590 * tree-nested.c (get_nl_goto_field): Likewise.
15591 * tree.c (build_vector_type_for_mode): Likewise.
15592 * var-tracking.c (use_narrower_mode): Likewise.
15594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15595 Alan Hayward <alan.hayward@arm.com>
15596 David Sherwood <david.sherwood@arm.com>
15598 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15599 * builtins.c (expand_builtin_signbit): Use it.
15600 * cfgexpand.c (expand_debug_expr): Likewise.
15601 * dojump.c (do_jump): Likewise.
15602 (do_compare_and_jump): Likewise.
15603 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15604 * expmed.c (make_tree): Likewise.
15605 * expr.c (expand_expr_real_2): Likewise.
15606 (expand_expr_real_1): Likewise.
15607 (try_casesi): Likewise.
15608 * fold-const-call.c (fold_const_call_ss): Likewise.
15609 * fold-const.c (unextend): Likewise.
15610 (extract_muldiv_1): Likewise.
15611 (fold_single_bit_test): Likewise.
15612 (native_encode_int): Likewise.
15613 (native_encode_string): Likewise.
15614 (native_interpret_int): Likewise.
15615 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15616 * internal-fn.c (expand_addsub_overflow): Likewise.
15617 (expand_neg_overflow): Likewise.
15618 (expand_mul_overflow): Likewise.
15619 (expand_arith_overflow): Likewise.
15620 * match.pd: Likewise.
15621 * stor-layout.c (layout_type): Likewise.
15622 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15623 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15624 * tree-ssanames.c (get_range_info): Likewise.
15625 * tree-switch-conversion.c (array_value_type) Likewise.
15626 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15627 (vect_recog_divmod_pattern): Likewise.
15628 (vect_recog_mixed_size_cond_pattern): Likewise.
15629 * tree-vrp.c (extract_range_basic): Likewise.
15630 (simplify_float_conversion_using_ranges): Likewise.
15631 * tree.c (int_fits_type_p): Likewise.
15632 * ubsan.c (instrument_bool_enum_load): Likewise.
15633 * varasm.c (mergeable_string_section): Likewise.
15634 (narrowing_initializer_constant_valid_p): Likewise.
15635 (output_constant): Likewise.
15637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15638 Alan Hayward <alan.hayward@arm.com>
15639 David Sherwood <david.sherwood@arm.com>
15641 * machmode.h (NARROWEST_INT_MODE): New macro.
15642 * expr.c (alignment_for_piecewise_move): Use it instead of
15643 GET_CLASS_NARROWEST_MODE (MODE_INT).
15644 (push_block): Likewise.
15645 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15647 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15650 Alan Hayward <alan.hayward@arm.com>
15651 David Sherwood <david.sherwood@arm.com>
15653 * postreload.c (move2add_valid_value_p): Change the type of the
15654 mode parameter to scalar_int_mode.
15655 (move2add_use_add2_insn): Add a mode parameter and use it instead
15657 (move2add_use_add3_insn): Likewise.
15658 (reload_cse_move2add): Update accordingly.
15660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15661 Alan Hayward <alan.hayward@arm.com>
15662 David Sherwood <david.sherwood@arm.com>
15664 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15666 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15667 * optabs.c (expand_unop): Likewise.
15669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15670 Alan Hayward <alan.hayward@arm.com>
15671 David Sherwood <david.sherwood@arm.com>
15673 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15674 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15675 (popcount_loc_descriptor): Likewise.
15676 (bswap_loc_descriptor): Likewise.
15677 (rotate_loc_descriptor): Likewise.
15678 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15679 calling the functions above.
15681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15682 Alan Hayward <alan.hayward@arm.com>
15683 David Sherwood <david.sherwood@arm.com>
15685 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15687 (try_combine): Likewise.
15688 (simplify_if_then_else): Likewise.
15689 * cse.c (cse_insn): Likewise.
15690 * dwarf2out.c (mem_loc_descriptor): Likewise.
15691 * emit-rtl.c (gen_lowpart_common): Likewise.
15692 * simplify-rtx.c (simplify_truncation): Likewise.
15693 (simplify_binary_operation_1): Likewise.
15694 (simplify_const_relational_operation): Likewise.
15695 (simplify_ternary_operation): Likewise.
15696 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15699 Alan Hayward <alan.hayward@arm.com>
15700 David Sherwood <david.sherwood@arm.com>
15702 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15703 * reload.c (push_reload): Likewise.
15704 (find_reloads): Likewise.
15706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15707 Alan Hayward <alan.hayward@arm.com>
15708 David Sherwood <david.sherwood@arm.com>
15710 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15711 (make_compound_operation_int): Likewise.
15712 (change_zero_ext): Likewise.
15713 * expr.c (convert_move): Likewise.
15714 (convert_modes): Likewise.
15715 * fwprop.c (forward_propagate_subreg): Likewise.
15716 * loop-iv.c (get_biv_step_1): Likewise.
15717 * optabs.c (widen_operand): Likewise.
15718 * postreload.c (move2add_valid_value_p): Likewise.
15719 * recog.c (simplify_while_replacing): Likewise.
15720 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15721 (simplify_binary_operation_1): Likewise. Remove redundant
15722 mode equality check.
15724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15725 Alan Hayward <alan.hayward@arm.com>
15726 David Sherwood <david.sherwood@arm.com>
15728 * combine.c (combine_simplify_rtx): Add checks for
15729 is_a <scalar_int_mode>.
15730 (simplify_if_then_else): Likewise.
15731 (make_field_assignment): Likewise.
15732 (simplify_comparison): Likewise.
15733 * ifcvt.c (noce_try_bitop): Likewise.
15734 * loop-invariant.c (canonicalize_address_mult): Likewise.
15735 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15738 Alan Hayward <alan.hayward@arm.com>
15739 David Sherwood <david.sherwood@arm.com>
15741 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15742 is_a <scalar_int_mode> instead of != BLKmode.
15744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15745 Alan Hayward <alan.hayward@arm.com>
15746 David Sherwood <david.sherwood@arm.com>
15748 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15749 instead of != VOIDmode.
15750 * combine.c (if_then_else_cond): Likewise.
15751 (change_zero_ext): Likewise.
15752 * dwarf2out.c (mem_loc_descriptor): Likewise.
15753 (loc_descriptor): Likewise.
15754 * rtlanal.c (canonicalize_condition): Likewise.
15755 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15758 Alan Hayward <alan.hayward@arm.com>
15759 David Sherwood <david.sherwood@arm.com>
15761 * simplify-rtx.c (simplify_binary_operation_1): Use
15762 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15765 Alan Hayward <alan.hayward@arm.com>
15766 David Sherwood <david.sherwood@arm.com>
15768 * wide-int.h (int_traits<unsigned char>) New class.
15769 (int_traits<unsigned short>) Likewise.
15770 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15771 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15773 * combine.c (set_nonzero_bits_and_sign_copies): Use
15774 is_a <scalar_int_mode>.
15775 (find_split_point): Likewise.
15776 (combine_simplify_rtx): Likewise.
15777 (simplify_logical): Likewise.
15778 (expand_compound_operation): Likewise.
15779 (expand_field_assignment): Likewise.
15780 (make_compound_operation): Likewise.
15781 (extended_count): Likewise.
15782 (change_zero_ext): Likewise.
15783 (simplify_comparison): Likewise.
15784 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15785 (ucompare_loc_descriptor): Likewise.
15786 (minmax_loc_descriptor): Likewise.
15787 (mem_loc_descriptor): Likewise.
15788 (loc_descriptor): Likewise.
15789 * expmed.c (init_expmed_one_mode): Likewise.
15790 * lra-constraints.c (lra_constraint_offset): Likewise.
15791 * optabs.c (prepare_libcall_arg): Likewise.
15792 * postreload.c (move2add_note_store): Likewise.
15793 * reload.c (operands_match_p): Likewise.
15794 * rtl.h (load_extend_op): Likewise.
15795 * rtlhooks.c (gen_lowpart_general): Likewise.
15796 * simplify-rtx.c (simplify_truncation): Likewise.
15797 (simplify_unary_operation_1): Likewise.
15798 (simplify_binary_operation_1): Likewise.
15799 (simplify_const_binary_operation): Likewise.
15800 (simplify_const_relational_operation): Likewise.
15801 (simplify_subreg): Likewise.
15802 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15803 * var-tracking.c (adjust_mems): Likewise.
15804 (prepare_call_arguments): Likewise.
15806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15807 Alan Hayward <alan.hayward@arm.com>
15808 David Sherwood <david.sherwood@arm.com>
15810 * machmode.h (is_int_mode): New fuction.
15811 * combine.c (find_split_point): Use it.
15812 (combine_simplify_rtx): Likewise.
15813 (simplify_if_then_else): Likewise.
15814 (simplify_set): Likewise.
15815 (simplify_shift_const_1): Likewise.
15816 (simplify_comparison): Likewise.
15817 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15818 * cse.c (notreg_cost): Likewise.
15819 (cse_insn): Likewise.
15820 * cselib.c (cselib_lookup_1): Likewise.
15821 * dojump.c (do_jump_1): Likewise.
15822 (do_compare_rtx_and_jump): Likewise.
15823 * dse.c (get_call_args): Likewise.
15824 * dwarf2out.c (rtl_for_decl_init): Likewise.
15825 (native_encode_initializer): Likewise.
15826 * expmed.c (emit_store_flag_1): Likewise.
15827 (emit_store_flag): Likewise.
15828 * expr.c (convert_modes): Likewise.
15829 (store_field): Likewise.
15830 (expand_expr_real_1): Likewise.
15831 * fold-const.c (fold_read_from_constant_string): Likewise.
15832 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15833 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15834 * optabs.c (expand_binop): Likewise.
15835 (expand_unop): Likewise.
15836 (expand_abs_nojump): Likewise.
15837 (expand_one_cmpl_abs_nojump): Likewise.
15838 * simplify-rtx.c (mode_signbit_p): Likewise.
15839 (val_signbit_p): Likewise.
15840 (val_signbit_known_set_p): Likewise.
15841 (val_signbit_known_clear_p): Likewise.
15842 (simplify_relational_operation_1): Likewise.
15843 * tree.c (vector_type_mode): Likewise.
15845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15846 Alan Hayward <alan.hayward@arm.com>
15847 David Sherwood <david.sherwood@arm.com>
15849 * machmode.h (smallest_mode_for_size): Fix formatting.
15850 (smallest_int_mode_for_size): New function.
15851 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15852 instead of smallest_mode_for_size.
15853 * combine.c (make_extraction): Likewise.
15854 * config/arc/arc.c (arc_expand_movmem): Likewise.
15855 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15856 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15857 * config/s390/s390.c (s390_expand_insv): Likewise.
15858 * config/sparc/sparc.c (assign_int_registers): Likewise.
15859 * config/spu/spu.c (spu_function_value): Likewise.
15860 (spu_function_arg): Likewise.
15861 * coverage.c (get_gcov_type): Likewise.
15862 (get_gcov_unsigned_t): Likewise.
15863 * dse.c (find_shift_sequence): Likewise.
15864 * expmed.c (store_bit_field_1): Likewise.
15865 * expr.c (convert_move): Likewise.
15866 (store_field): Likewise.
15867 * internal-fn.c (expand_arith_overflow): Likewise.
15868 * optabs-query.c (get_best_extraction_insn): Likewise.
15869 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15870 * stor-layout.c (layout_type): Likewise.
15871 (initialize_sizetypes): Likewise.
15872 * targhooks.c (default_get_mask_mode): Likewise.
15873 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15876 Alan Hayward <alan.hayward@arm.com>
15877 David Sherwood <david.sherwood@arm.com>
15879 * machmode.h (opt_mode::else_blk): New function.
15880 (int_mode_for_mode): Declare.
15881 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15882 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15884 * cfgexpand.c (expand_debug_expr): Likewise.
15885 * combine.c (gen_lowpart_or_truncate): Likewise.
15886 (gen_lowpart_for_combine): Likewise.
15887 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15888 * config/avr/avr.c (avr_to_int_mode): Likewise.
15889 (avr_out_plus_1): Likewise.
15890 (avr_out_plus): Likewise.
15891 (avr_out_round): Likewise.
15892 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15893 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15894 (s390_expand_vcond): Likewise.
15895 * config/spu/spu.c (spu_split_immediate): Likewise.
15896 (spu_expand_mov): Likewise.
15897 * dse.c (get_stored_val): Likewise.
15898 * expmed.c (store_bit_field_1): Likewise.
15899 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15901 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15902 (extract_low_bits): Likewise.
15903 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15904 handling rather than repeating the check.
15905 (emit_group_store): Likewise.
15906 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15907 * optabs.c (expand_absneg_bit): Likewise.
15908 (expand_copysign_absneg): Likewise.
15909 (expand_copysign_bit): Likewise.
15910 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15911 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15912 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15913 * var-tracking.c (prepare_call_arguments): Likewise.
15914 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15915 int_mode_for_mode instead of mode_for_size.
15916 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15919 Alan Hayward <alan.hayward@arm.com>
15920 David Sherwood <david.sherwood@arm.com>
15922 * machmode.h (int_mode_for_size): New function.
15923 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15924 instead of mode_for_size.
15925 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15927 * combine.c (expand_field_assignment): Use int_mode_for_size
15928 instead of mode_for_size.
15929 (make_extraction): Likewise.
15930 (simplify_shift_const_1): Likewise.
15931 (simplify_comparison): Likewise.
15932 * dojump.c (do_jump): Likewise.
15933 * dwarf2out.c (mem_loc_descriptor): Likewise.
15934 * emit-rtl.c (init_derived_machine_modes): Likewise.
15935 * expmed.c (flip_storage_order): Likewise.
15936 (convert_extracted_bit_field): Likewise.
15937 * expr.c (copy_blkmode_from_reg): Likewise.
15938 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15939 * internal-fn.c (expand_mul_overflow): Likewise.
15940 * lower-subreg.c (simple_move): Likewise.
15941 * optabs-libfuncs.c (init_optabs): Likewise.
15942 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15943 * tree.c (vector_type_mode): Likewise.
15944 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15945 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15946 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15947 * tree-vect-stmts.c (vectorizable_load): Likewise.
15948 (vectorizable_store): Likewise.
15950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15951 Alan Hayward <alan.hayward@arm.com>
15952 David Sherwood <david.sherwood@arm.com>
15954 * coretypes.h (pod_mode): New type.
15955 (scalar_int_mode_pod): New typedef.
15956 * machmode.h (pod_mode): New class.
15957 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15958 * genmodes.c (emit_mode_int_n): Update accordingly.
15959 * lower-subreg.h (target_lower_subreg): Change type to
15960 scalar_int_mode_pod.
15961 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15962 scalar_int_mode_pod.
15964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15965 Alan Hayward <alan.hayward@arm.com>
15966 David Sherwood <david.sherwood@arm.com>
15968 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15969 machine_mode to scalar_int_mode.
15970 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15971 (rs6000_option_override_internal): Remove cast to int.
15972 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15973 machine_mode to scalar_int_mode.
15974 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15975 (rs6000_option_override_internal): Remove cast to int.
15976 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15977 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15979 * config/s390/s390.c (s390_expand_builtin): Likewise.
15980 * coretypes.h (scalar_int_mode): New type.
15981 (opt_scalar_int_mode): New typedef.
15982 * machmode.h (scalar_int_mode): New class.
15983 (scalar_int_mode::includes_p): New function.
15984 (byte_mode): Change type to scalar_int_mode.
15985 (word_mode): Likewise.
15986 (ptr_mode): Likewise.
15987 * emit-rtl.c (byte_mode): Likewise.
15988 (word_mode): Likewise.
15989 (ptr_mode): Likewise.
15990 (init_derived_machine_modes): Update accordingly.
15991 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15992 and MODE_PARTIAL_INT.
15993 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15994 opt_scalar_int_mode.
15996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15997 Alan Hayward <alan.hayward@arm.com>
15998 David Sherwood <david.sherwood@arm.com>
16000 * target.def (libgcc_floating_mode_supported_p): Take a
16002 * doc/tm.texi: Regenerate.
16003 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16005 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16006 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16010 Alan Hayward <alan.hayward@arm.com>
16011 David Sherwood <david.sherwood@arm.com>
16013 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16014 * doc/tm.texi: Regenerate.
16015 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16016 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16017 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16018 * targhooks.h (default_floatn_mode): Likewise.
16019 * targhooks.c (default_floatn_mode): Likewise.
16020 * tree.c (build_common_tree_nodes): Update accordingly.
16022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16023 Alan Hayward <alan.hayward@arm.com>
16024 David Sherwood <david.sherwood@arm.com>
16026 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16027 (mode_iterator::iterate_p): Likewise.
16028 (mode_iterator::get_wider): Likewise.
16029 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16032 Alan Hayward <alan.hayward@arm.com>
16033 David Sherwood <david.sherwood@arm.com>
16035 * coretypes.h (opt_scalar_float_mode): New typedef.
16036 * machmode.h (float_mode_for_size): New function.
16037 * emit-rtl.c (double_mode): Delete.
16038 (init_emit_once): Use float_mode_for_size.
16039 * stor-layout.c (layout_type): Likewise.
16040 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16043 Alan Hayward <alan.hayward@arm.com>
16044 David Sherwood <david.sherwood@arm.com>
16046 * output.h (assemble_real): Take a scalar_float_mode.
16047 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16048 * config/arm/arm.md (consttable_4): Likewise.
16049 (consttable_8): Likewise.
16050 (consttable_16): Likewise.
16051 * config/mips/mips.md (consttable_float): Likewise.
16052 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16053 * varasm.c (assemble_real): Take a scalar_float_mode.
16054 (output_constant_pool_2): Update accordingly.
16055 (output_constant): Likewise.
16057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16058 Alan Hayward <alan.hayward@arm.com>
16059 David Sherwood <david.sherwood@arm.com>
16061 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16062 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16063 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16064 (native_encode_real): Likewise.
16065 (native_interpret_real): Likewise.
16066 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16067 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16070 Alan Hayward <alan.hayward@arm.com>
16071 David Sherwood <david.sherwood@arm.com>
16073 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16074 <scalar_float_mode>. Simplify.
16075 (gen_extend_conv_libfunc): Likewise.
16077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16078 Alan Hayward <alan.hayward@arm.com>
16079 David Sherwood <david.sherwood@arm.com>
16081 * coretypes.h (scalar_float_mode): New type.
16082 * machmode.h (mode_traits::from_int): Use machine_mode if
16083 USE_ENUM_MODES is defined.
16084 (is_a): New function.
16086 (dyn_cast): Likewise.
16087 (scalar_float_mode): New class.
16088 (scalar_float_mode::includes_p): New function.
16089 (is_float_mode): Likewise.
16090 * gdbhooks.py (MachineModePrinter): New class.
16091 (build_pretty_printer): Use it for scalar_float_mode.
16092 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16093 (format_helper::format_helper): Turn into a template.
16094 * genmodes.c (get_mode_class): New function.
16095 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16096 or machine_mode if none.
16097 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16098 as_a <scalar_float_mode>.
16099 * dwarf2out.c (mem_loc_descriptor): Likewise.
16100 (insert_float): Likewise.
16101 (add_const_value_attribute): Likewise.
16102 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16103 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16104 (expand_unop): Update accordingly.
16105 (expand_abs_nojump): Likewise.
16106 (expand_copysign_absneg): Take a scalar_float_mode.
16107 (expand_copysign_bit): Likewise.
16108 (expand_copysign): Update accordingly.
16110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16111 Alan Hayward <alan.hayward@arm.com>
16112 David Sherwood <david.sherwood@arm.com>
16114 * coretypes.h (opt_mode): New class.
16115 * machmode.h (opt_mode): Likewise.
16116 (opt_mode::else_void): New function.
16117 (opt_mode::require): Likewise.
16118 (opt_mode::exists): Likewise.
16119 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16120 (GET_MODE_2XWIDER_MODE): Likewise.
16121 (mode_iterator::get_wider): Update accordingly.
16122 (mode_iterator::get_2xwider): Likewise.
16123 (mode_iterator::get_known_wider): Likewise, turning into a template.
16124 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16125 forcing a wider mode to exist.
16126 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16127 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16128 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16129 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16130 * lower-subreg.c (init_lower_subreg): Likewise.
16131 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16132 on the final iteration.
16133 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16134 a wider mode exists before asking for a move pattern.
16135 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16136 forcing a wider mode to exist.
16137 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16138 returning false if no such mode exists.
16139 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16140 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16141 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16142 Avoid checking for a MODE_INT if we already know the mode is not a
16144 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16145 forcing a wider mode to exist.
16146 (expmed_mult_highpart_optab): Likewise.
16147 (expmed_mult_highpart): Likewise.
16148 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16150 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16151 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16152 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16153 * internal-fn.c (expand_mul_overflow): Update use of
16154 GET_MODE_2XWIDER_MODE.
16155 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16156 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16157 GET_MODE_WIDER_MODE.
16158 (convert_plusminus_to_widen): Likewise.
16159 * tree-switch-conversion.c (array_value_type): Likewise.
16160 * var-tracking.c (emit_note_insn_var_location): Likewise.
16161 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16162 Return false inside rather than outside the loop if no wider mode
16164 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16165 and GET_MODE_2XWIDER_MODE
16166 (can_compare_p): Use else_void.
16167 * gdbhooks.py (OptMachineModePrinter): New class.
16168 (build_pretty_printer): Use it for opt_mode.
16170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16171 Alan Hayward <alan.hayward@arm.com>
16172 David Sherwood <david.sherwood@arm.com>
16174 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16175 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16178 Alan Hayward <alan.hayward@arm.com>
16179 David Sherwood <david.sherwood@arm.com>
16181 * machmode.h (mode_traits): New structure.
16182 (get_narrowest_mode): New function.
16183 (mode_iterator::start): Likewise.
16184 (mode_iterator::iterate_p): Likewise.
16185 (mode_iterator::get_wider): Likewise.
16186 (mode_iterator::get_known_wider): Likewise.
16187 (mode_iterator::get_2xwider): Likewise.
16188 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16189 (FOR_EACH_MODE): Likewise.
16190 (FOR_EACH_MODE_FROM): Likewise.
16191 (FOR_EACH_MODE_UNTIL): Likewise.
16192 (FOR_EACH_WIDER_MODE): Likewise.
16193 (FOR_EACH_2XWIDER_MODE): Likewise.
16194 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16195 * combine.c (simplify_comparison): Likewise
16196 * config/i386/i386.c (type_natural_mode): Likewise.
16197 * cse.c (cse_insn): Likewise.
16198 * dse.c (find_shift_sequence): Likewise.
16199 * emit-rtl.c (init_derived_machine_modes): Likewise.
16200 (init_emit_once): Likewise.
16201 * explow.c (hard_function_value): Likewise.
16202 * expmed.c (extract_fixed_bit_field_1): Likewise.
16203 (extract_bit_field_1): Likewise.
16204 (expand_divmod): Likewise.
16205 (emit_store_flag_1): Likewise.
16206 * expr.c (init_expr_target): Likewise.
16207 (convert_move): Likewise.
16208 (alignment_for_piecewise_move): Likewise.
16209 (widest_int_mode_for_size): Likewise.
16210 (emit_block_move_via_movmem): Likewise.
16211 (copy_blkmode_to_reg): Likewise.
16212 (set_storage_via_setmem): Likewise.
16213 (compress_float_constant): Likewise.
16214 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16215 * optabs-query.c (get_best_extraction_insn): Likewise.
16216 * optabs.c (expand_binop): Likewise.
16217 (expand_twoval_unop): Likewise.
16218 (expand_twoval_binop): Likewise.
16219 (widen_leading): Likewise.
16220 (widen_bswap): Likewise.
16221 (expand_parity): Likewise.
16222 (expand_unop): Likewise.
16223 (prepare_cmp_insn): Likewise.
16224 (prepare_float_lib_cmp): Likewise.
16225 (expand_float): Likewise.
16226 (expand_fix): Likewise.
16227 (expand_sfix_optab): Likewise.
16228 * postreload.c (move2add_use_add2_insn): Likewise.
16229 * reg-stack.c (reg_to_stack): Likewise.
16230 * reginfo.c (choose_hard_reg_mode): Likewise.
16231 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16232 * stor-layout.c (mode_for_size): Likewise.
16233 (smallest_mode_for_size): Likewise.
16234 (mode_for_vector): Likewise.
16235 (finish_bitfield_representative): Likewise.
16236 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16237 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16238 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16239 * var-tracking.c (prepare_call_arguments): Likewise.
16241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16242 Alan Hayward <alan.hayward@arm.com>
16243 David Sherwood <david.sherwood@arm.com>
16245 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16246 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16247 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16248 * machmode.h (mode_size): Move earlier in file.
16249 (mode_precision): Likewise.
16250 (mode_inner): Likewise.
16251 (mode_nunits): Likewise.
16252 (mode_unit_size): Likewise.
16253 (unit_unit_precision): Likewise.
16254 (mode_wider): Likewise.
16255 (mode_2xwider): Likewise.
16256 (machine_mode): New class.
16257 (mode_to_bytes): New function.
16258 (mode_to_bits): Likewise.
16259 (mode_to_precision): Likewise.
16260 (mode_to_inner): Likewise.
16261 (mode_to_unit_size): Likewise.
16262 (mode_to_unit_precision): Likewise.
16263 (mode_to_nunits): Likewise.
16264 (GET_MODE_SIZE): Use mode_to_bytes.
16265 (GET_MODE_BITSIZE): Use mode_to_bits.
16266 (GET_MODE_PRECISION): Use mode_to_precision.
16267 (GET_MODE_INNER): Use mode_to_inner.
16268 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16269 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16270 (GET_MODE_NUNITS): Use mode_to_nunits.
16271 * system.h (ALWAYS_INLINE): New macro.
16272 * config/powerpcspe/powerpcspe-c.c
16273 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16274 int for arg1_mode and arg2_mode.
16276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16277 Alan Hayward <alan.hayward@arm.com>
16278 David Sherwood <david.sherwood@arm.com>
16280 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16281 Prefix mode names with E_ in case statements.
16282 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16283 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16284 (aarch64_split_simd_move): Likewise.
16285 (aarch64_gen_storewb_pair): Likewise.
16286 (aarch64_gen_loadwb_pair): Likewise.
16287 (aarch64_gen_store_pair): Likewise.
16288 (aarch64_gen_load_pair): Likewise.
16289 (aarch64_get_condition_code_1): Likewise.
16290 (aarch64_constant_pool_reload_icode): Likewise.
16291 (get_rsqrte_type): Likewise.
16292 (get_rsqrts_type): Likewise.
16293 (get_recpe_type): Likewise.
16294 (get_recps_type): Likewise.
16295 (aarch64_gimplify_va_arg_expr): Likewise.
16296 (aarch64_simd_container_mode): Likewise.
16297 (aarch64_emit_load_exclusive): Likewise.
16298 (aarch64_emit_store_exclusive): Likewise.
16299 (aarch64_expand_compare_and_swap): Likewise.
16300 (aarch64_gen_atomic_cas): Likewise.
16301 (aarch64_emit_bic): Likewise.
16302 (aarch64_emit_atomic_swap): Likewise.
16303 (aarch64_emit_atomic_load_op): Likewise.
16304 (aarch64_evpc_trn): Likewise.
16305 (aarch64_evpc_uzp): Likewise.
16306 (aarch64_evpc_zip): Likewise.
16307 (aarch64_evpc_ext): Likewise.
16308 (aarch64_evpc_rev): Likewise.
16309 (aarch64_evpc_dup): Likewise.
16310 (aarch64_gen_ccmp_first): Likewise.
16311 (aarch64_gen_ccmp_next): Likewise.
16312 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16313 (alpha_emit_xfloating_libcall): Likewise.
16314 (emit_insxl): Likewise.
16315 (alpha_arg_type): Likewise.
16316 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16317 (arc_preferred_simd_mode): Likewise.
16318 (arc_secondary_reload): Likewise.
16319 (get_arc_condition_code): Likewise.
16320 (arc_print_operand): Likewise.
16321 (arc_legitimate_constant_p): Likewise.
16322 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16323 * config/arc/arc.md (casesi_load): Likewise.
16324 (casesi_compact_jump): Likewise.
16325 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16326 (cc_use_register): Likewise.
16327 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16328 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16329 (arm_init_iwmmxt_builtins): Likewise.
16330 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16331 (neon_expand_vector_init): Likewise.
16332 (arm_attr_length_move_neon): Likewise.
16333 (maybe_get_arm_condition_code): Likewise.
16334 (arm_emit_vector_const): Likewise.
16335 (arm_preferred_simd_mode): Likewise.
16336 (arm_output_iwmmxt_tinsr): Likewise.
16337 (thumb1_output_casesi): Likewise.
16338 (thumb2_output_casesi): Likewise.
16339 (arm_emit_load_exclusive): Likewise.
16340 (arm_emit_store_exclusive): Likewise.
16341 (arm_expand_compare_and_swap): Likewise.
16342 (arm_evpc_neon_vuzp): Likewise.
16343 (arm_evpc_neon_vzip): Likewise.
16344 (arm_evpc_neon_vrev): Likewise.
16345 (arm_evpc_neon_vtrn): Likewise.
16346 (arm_evpc_neon_vext): Likewise.
16347 (arm_validize_comparison): Likewise.
16348 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16349 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16350 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16351 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16352 (c6x_preferred_simd_mode): Likewise.
16353 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16354 (epiphany_rtx_costs): Likewise.
16355 * config/epiphany/predicates.md (proper_comparison_operator):
16357 * config/frv/frv.c (condexec_memory_operand): Likewise.
16358 (frv_emit_move): Likewise.
16359 (output_move_single): Likewise.
16360 (output_condmove_single): Likewise.
16361 (frv_hard_regno_mode_ok): Likewise.
16362 (frv_matching_accg_mode): Likewise.
16363 * config/h8300/h8300.c (split_adds_subs): Likewise.
16364 (h8300_rtx_costs): Likewise.
16365 (h8300_print_operand): Likewise.
16366 (compute_mov_length): Likewise.
16367 (output_logical_op): Likewise.
16368 (compute_logical_op_length): Likewise.
16369 (compute_logical_op_cc): Likewise.
16370 (h8300_shift_needs_scratch_p): Likewise.
16371 (output_a_shift): Likewise.
16372 (compute_a_shift_length): Likewise.
16373 (compute_a_shift_cc): Likewise.
16374 (expand_a_rotate): Likewise.
16375 (output_a_rotate): Likewise.
16376 * config/i386/i386.c (classify_argument): Likewise.
16377 (function_arg_advance_32): Likewise.
16378 (function_arg_32): Likewise.
16379 (function_arg_64): Likewise.
16380 (function_value_64): Likewise.
16381 (ix86_gimplify_va_arg): Likewise.
16382 (ix86_legitimate_constant_p): Likewise.
16383 (put_condition_code): Likewise.
16384 (split_double_mode): Likewise.
16385 (ix86_avx256_split_vector_move_misalign): Likewise.
16386 (ix86_expand_vector_logical_operator): Likewise.
16387 (ix86_split_idivmod): Likewise.
16388 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16389 (ix86_build_const_vector): Likewise.
16390 (ix86_build_signbit_mask): Likewise.
16391 (ix86_match_ccmode): Likewise.
16392 (ix86_cc_modes_compatible): Likewise.
16393 (ix86_expand_branch): Likewise.
16394 (ix86_expand_sse_cmp): Likewise.
16395 (ix86_expand_sse_movcc): Likewise.
16396 (ix86_expand_int_sse_cmp): Likewise.
16397 (ix86_expand_vec_perm_vpermi2): Likewise.
16398 (ix86_expand_vec_perm): Likewise.
16399 (ix86_expand_sse_unpack): Likewise.
16400 (ix86_expand_int_addcc): Likewise.
16401 (ix86_split_to_parts): Likewise.
16402 (ix86_vectorize_builtin_gather): Likewise.
16403 (ix86_vectorize_builtin_scatter): Likewise.
16404 (avx_vpermilp_parallel): Likewise.
16405 (inline_memory_move_cost): Likewise.
16406 (ix86_tieable_integer_mode_p): Likewise.
16407 (x86_maybe_negate_const_int): Likewise.
16408 (ix86_expand_vector_init_duplicate): Likewise.
16409 (ix86_expand_vector_init_one_nonzero): Likewise.
16410 (ix86_expand_vector_init_one_var): Likewise.
16411 (ix86_expand_vector_init_concat): Likewise.
16412 (ix86_expand_vector_init_interleave): Likewise.
16413 (ix86_expand_vector_init_general): Likewise.
16414 (ix86_expand_vector_set): Likewise.
16415 (ix86_expand_vector_extract): Likewise.
16416 (emit_reduc_half): Likewise.
16417 (ix86_emit_i387_round): Likewise.
16418 (ix86_mangle_type): Likewise.
16419 (ix86_expand_round_sse4): Likewise.
16420 (expand_vec_perm_blend): Likewise.
16421 (canonicalize_vector_int_perm): Likewise.
16422 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16423 (expand_vec_perm_1): Likewise.
16424 (expand_vec_perm_interleave3): Likewise.
16425 (expand_vec_perm_even_odd_pack): Likewise.
16426 (expand_vec_perm_even_odd_1): Likewise.
16427 (expand_vec_perm_broadcast_1): Likewise.
16428 (ix86_vectorize_vec_perm_const_ok): Likewise.
16429 (ix86_expand_vecop_qihi): Likewise.
16430 (ix86_expand_mul_widen_hilo): Likewise.
16431 (ix86_expand_sse2_abs): Likewise.
16432 (ix86_expand_pextr): Likewise.
16433 (ix86_expand_pinsr): Likewise.
16434 (ix86_preferred_simd_mode): Likewise.
16435 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16436 * config/i386/sse.md (*andnot<mode>3): Likewise.
16437 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16438 (*<code><mode>3): Likewise.
16439 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16440 (ia64_expand_atomic_op): Likewise.
16441 (ia64_arg_type): Likewise.
16442 (ia64_mode_to_int): Likewise.
16443 (ia64_scalar_mode_supported_p): Likewise.
16444 (ia64_vector_mode_supported_p): Likewise.
16445 (expand_vec_perm_broadcast): Likewise.
16446 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16447 (iq2000_function_arg_advance): Likewise.
16448 (iq2000_function_arg): Likewise.
16449 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16450 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16451 (m68k_libcall_value): Likewise.
16452 (m68k_function_value): Likewise.
16453 (sched_attr_op_type): Likewise.
16454 * config/mcore/mcore.c (mcore_output_move): Likewise.
16455 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16457 (microblaze_function_arg): Likewise.
16458 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16459 (mips_print_operand): Likewise.
16460 (mips_mode_ok_for_mov_fmt_p): Likewise.
16461 (mips_vector_mode_supported_p): Likewise.
16462 (mips_preferred_simd_mode): Likewise.
16463 (mips_expand_vpc_loongson_even_odd): Likewise.
16464 (mips_expand_vec_unpack): Likewise.
16465 (mips_expand_vi_broadcast): Likewise.
16466 (mips_expand_vector_init): Likewise.
16467 (mips_expand_vec_reduc): Likewise.
16468 (mips_expand_msa_cmp): Likewise.
16469 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16470 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16471 (cc_flags_for_mode): Likewise.
16472 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16473 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16474 (nds32_output_casesi_pc_relative): Likewise.
16475 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16476 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16477 (nvptx_gen_unpack): Likewise.
16478 (nvptx_gen_pack): Likewise.
16479 (nvptx_gen_shuffle): Likewise.
16480 (nvptx_gen_wcast): Likewise.
16481 (nvptx_preferred_simd_mode): Likewise.
16482 * config/pa/pa.c (pa_secondary_reload): Likewise.
16483 * config/pa/predicates.md (base14_operand): Likewise.
16484 * config/powerpcspe/powerpcspe-c.c
16485 (altivec_resolve_overloaded_builtin): Likewise.
16486 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16488 (rs6000_preferred_simd_mode): Likewise.
16489 (output_vec_const_move): Likewise.
16490 (rs6000_expand_vector_extract): Likewise.
16491 (rs6000_split_vec_extract_var): Likewise.
16492 (reg_offset_addressing_ok_p): Likewise.
16493 (rs6000_legitimate_offset_address_p): Likewise.
16494 (rs6000_legitimize_address): Likewise.
16495 (rs6000_emit_set_const): Likewise.
16496 (rs6000_const_vec): Likewise.
16497 (rs6000_emit_move): Likewise.
16498 (spe_build_register_parallel): Likewise.
16499 (rs6000_darwin64_record_arg_recurse): Likewise.
16500 (swap_selector_for_mode): Likewise.
16501 (spe_init_builtins): Likewise.
16502 (paired_init_builtins): Likewise.
16503 (altivec_init_builtins): Likewise.
16504 (do_load_for_compare): Likewise.
16505 (rs6000_generate_compare): Likewise.
16506 (rs6000_expand_float128_convert): Likewise.
16507 (emit_load_locked): Likewise.
16508 (emit_store_conditional): Likewise.
16509 (rs6000_output_function_epilogue): Likewise.
16510 (rs6000_handle_altivec_attribute): Likewise.
16511 (rs6000_function_value): Likewise.
16512 (emit_fusion_gpr_load): Likewise.
16513 (emit_fusion_p9_load): Likewise.
16514 (emit_fusion_p9_store): Likewise.
16515 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16516 (fusion_gpr_mem_load): Likewise.
16517 (fusion_addis_mem_combo_load): Likewise.
16518 (fusion_addis_mem_combo_store): Likewise.
16519 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16520 (fusion_gpr_mem_load): Likewise.
16521 (fusion_addis_mem_combo_load): Likewise.
16522 (fusion_addis_mem_combo_store): Likewise.
16523 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16525 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16526 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16527 (rs6000_preferred_simd_mode): Likewise.
16528 (output_vec_const_move): Likewise.
16529 (rs6000_expand_vector_extract): Likewise.
16530 (rs6000_split_vec_extract_var): Likewise.
16531 (reg_offset_addressing_ok_p): Likewise.
16532 (rs6000_legitimate_offset_address_p): Likewise.
16533 (rs6000_legitimize_address): Likewise.
16534 (rs6000_emit_set_const): Likewise.
16535 (rs6000_const_vec): Likewise.
16536 (rs6000_emit_move): Likewise.
16537 (rs6000_darwin64_record_arg_recurse): Likewise.
16538 (swap_selector_for_mode): Likewise.
16539 (paired_init_builtins): Likewise.
16540 (altivec_init_builtins): Likewise.
16541 (rs6000_expand_float128_convert): Likewise.
16542 (emit_load_locked): Likewise.
16543 (emit_store_conditional): Likewise.
16544 (rs6000_output_function_epilogue): Likewise.
16545 (rs6000_handle_altivec_attribute): Likewise.
16546 (rs6000_function_value): Likewise.
16547 (emit_fusion_gpr_load): Likewise.
16548 (emit_fusion_p9_load): Likewise.
16549 (emit_fusion_p9_store): Likewise.
16550 * config/rx/rx.c (rx_gen_move_template): Likewise.
16551 (flags_from_mode): Likewise.
16552 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16553 (s390_slb_comparison): Likewise.
16554 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16555 (s390_vector_mode_supported_p): Likewise.
16556 (s390_cc_modes_compatible): Likewise.
16557 (s390_match_ccmode_set): Likewise.
16558 (s390_canonicalize_comparison): Likewise.
16559 (s390_emit_compare_and_swap): Likewise.
16560 (s390_branch_condition_mask): Likewise.
16561 (s390_rtx_costs): Likewise.
16562 (s390_secondary_reload): Likewise.
16563 (__SECONDARY_RELOAD_CASE): Likewise.
16564 (s390_expand_cs): Likewise.
16565 (s390_preferred_simd_mode): Likewise.
16566 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16567 * config/sh/sh.c (sh_print_operand): Likewise.
16568 (dump_table): Likewise.
16569 (sh_secondary_reload): Likewise.
16570 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16571 * config/sh/sh.md (casesi_worker_1): Likewise.
16572 (casesi_worker_2): Likewise.
16573 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16574 (fcc_comparison_operator): Likewise.
16575 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16576 (emit_soft_tfmode_cvt): Likewise.
16577 (sparc_preferred_simd_mode): Likewise.
16578 (output_cbranch): Likewise.
16579 (sparc_print_operand): Likewise.
16580 (sparc_expand_vec_perm_bmask): Likewise.
16581 (vector_init_bshuffle): Likewise.
16582 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16583 (spu_vector_mode_supported_p): Likewise.
16584 (spu_expand_insv): Likewise.
16585 (spu_emit_branch_or_set): Likewise.
16586 (spu_handle_vector_attribute): Likewise.
16587 (spu_builtin_splats): Likewise.
16588 (spu_builtin_extract): Likewise.
16589 (spu_builtin_promote): Likewise.
16590 (spu_expand_sign_extend): Likewise.
16591 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16592 (tilegx_simd_int): Likewise.
16593 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16594 (tilepro_simd_int): Likewise.
16595 * config/v850/v850.c (const_double_split): Likewise.
16596 (v850_print_operand): Likewise.
16597 (ep_memory_offset): Likewise.
16598 * config/vax/vax.c (vax_rtx_costs): Likewise.
16599 (vax_output_int_move): Likewise.
16600 (vax_output_int_add): Likewise.
16601 (vax_output_int_subtract): Likewise.
16602 * config/visium/predicates.md (visium_branch_operator): Likewise.
16603 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16604 (visium_print_operand_address): Likewise.
16605 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16606 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16607 (xtensa_expand_conditional_branch): Likewise.
16608 (xtensa_copy_incoming_a7): Likewise.
16609 (xtensa_output_literal): Likewise.
16610 * dfp.c (decimal_real_maxval): Likewise.
16611 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16614 Alan Hayward <alan.hayward@arm.com>
16615 David Sherwood <david.sherwood@arm.com>
16617 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16618 (mode_nunits_inline): Likewise.
16619 (mode_inner_inline): Likewise.
16620 (mode_unit_size_inline): Likewise.
16621 (mode_unit_precision_inline): Likewise.
16622 (emit_insn_modes_h): Likewise. Also emit a #define of the
16624 (emit_mode_wider): Add an E_ prefix to mode names.
16625 (emit_mode_complex): Likewise.
16626 (emit_mode_inner): Likewise.
16627 (emit_mode_adjustments): Likewise.
16628 (emit_mode_int_n): Likewise.
16629 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16630 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16631 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16632 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16633 (CRC32_BUILTIN, ENTRY): Likewise.
16634 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16635 (aarch64_pop_regs): Likewise.
16636 (aarch64_process_components): Likewise.
16637 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16638 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16639 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16640 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16641 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16642 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16644 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16645 (pcmpistr): Likewise.
16646 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16647 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16648 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16650 * config/rl78/rl78.c (mduc_regs): Likewise.
16651 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16652 (htm_expand_builtin): Likewise.
16653 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16654 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16655 * config/xtensa/xtensa.c (print_operand): Likewise.
16656 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16657 (NUM_MODE_VECTOR_INT): Likewise.
16658 * genoutput.c (null_operand): Likewise.
16659 (output_operand_data): Likewise.
16660 * genrecog.c (print_parameter_value): Likewise.
16661 * lra.c (debug_operand_data): Likewise.
16663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16665 * dwarf2out.c (scompare_loc_descriptor_wide)
16666 (scompare_loc_descriptor_narrow): New functions, split out from...
16667 (scompare_loc_descriptor): ...here.
16668 * expmed.c (emit_store_flag_int): New function, split out from...
16669 (emit_store_flag): ...here.
16671 2017-08-30 Richard Biener <rguenther@suse.de>
16673 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16674 (dwarf2out_early_finish): Move setting of AT_pubnames from
16675 early debug output to early finish.
16677 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16679 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16680 and -mdata-region to the assembler.
16682 2017-08-30 Richard Biener <rguenther@suse.de>
16684 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16686 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16687 (dwarf2out_early_global_decl): Only generate a DIE for the
16688 abstract origin if it doesn't already exist or is a declaration DIE.
16689 (resolve_addr): Do not add the linkage name twice when
16690 generating a stub DIE for the DW_TAG_GNU_call_site target.
16692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16694 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16695 Use machine_mode rather than int for arg1_mode.
16697 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16700 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16701 that the second argument of the built-in functions to unpack
16702 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16703 switch statement instead a lot of if statements.
16704 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16705 Allow 64-bit values to be in Altivec registers as well as
16706 traditional floating point registers.
16707 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16709 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16711 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16712 MAX_REGS_PER_ADDRESS == 1.
16714 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16716 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16717 * config/i386/i386.c (ix86_option_override_internal): Simplify
16718 setting of opts->x_flag_entry.
16720 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16721 Jakub Jelinek <jakub@redhat.com>
16722 Richard Biener <rguenther@suse.de>
16724 PR tree-optimization/81503
16725 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16726 folded constant fits in the target type; reorder tests for clarity.
16728 2017-08-29 Martin Liska <mliska@suse.cz>
16730 * passes.def: Include pass_lower_switch.
16731 * stmt.c (dump_case_nodes): Remove and move to
16732 tree-switch-conversion.
16733 (case_values_threshold): Likewise.
16734 (expand_switch_as_decision_tree_p): Likewise.
16735 (emit_case_decision_tree): Likewise.
16736 (expand_case): Likewise.
16737 (balance_case_nodes): Likewise.
16738 (node_has_low_bound): Likewise.
16739 (node_has_high_bound): Likewise.
16740 (node_is_bounded): Likewise.
16741 (emit_case_nodes): Likewise.
16742 (struct simple_case_node): New struct.
16743 (add_case_node): Remove.
16744 (emit_case_dispatch_table): Use vector instead of case_list.
16745 (reset_out_edges_aux): Remove.
16746 (compute_cases_per_edge): Likewise.
16747 (expand_case): Build list of simple_case_node.
16748 (expand_sjlj_dispatch_table): Use it.
16749 * tree-switch-conversion.c (struct case_node): Moved from
16750 stmt.c and adjusted.
16751 (emit_case_nodes): Likewise.
16752 (node_has_low_bound): Likewise.
16753 (node_has_high_bound): Likewise.
16754 (node_is_bounded): Likewise.
16755 (case_values_threshold): Likewise.
16756 (reset_out_edges_aux): Likewise.
16757 (compute_cases_per_edge): Likewise.
16758 (add_case_node): Likewise.
16759 (dump_case_nodes): Likewise.
16760 (balance_case_nodes): Likewise.
16761 (expand_switch_as_decision_tree_p): Likewise.
16762 (emit_jump): Likewise.
16763 (emit_case_decision_tree): Likewise.
16764 (try_switch_expansion): Likewise.
16765 (do_jump_if_equal): Likewise.
16766 (emit_cmp_and_jump_insns): Likewise.
16767 (fix_phi_operands_for_edge): New function.
16768 (record_phi_operand_mapping): Likewise.
16769 (class pass_lower_switch): New pass.
16770 (pass_lower_switch::execute): New function.
16771 (make_pass_lower_switch): Likewise.
16772 (conditional_probability):
16773 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16774 * tree-pass.h: Add make_pass_lower_switch.
16776 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16779 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16782 2017-08-29 Richard Biener <rguenther@suse.de>
16784 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16785 we do not add a DW_AT_inline attribute twice.
16786 (gen_subprogram_die): Remove code setting DW_AT_inline on
16787 DECL_ABSTRACT_P nodes.
16789 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16791 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16792 calls to internal functions.
16793 (gimplify_modify_expr): Likewise.
16794 * tree-call-cdce.c (use_internal_fn): Likewise.
16795 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16796 (convert_to_divmod): Set the nothrow flag.
16797 * tree-if-conv.c (predicate_mem_writes): Likewise.
16798 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16799 (vectorizable_call): Likewise.
16800 (vectorizable_store): Likewise.
16801 (vectorizable_load): Likewise.
16802 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16803 (vect_recog_mask_conversion_pattern): Likewise.
16805 2017-08-29 Martin Liska <mliska@suse.cz>
16808 * gcc.c (driver_handle_option): Add new argument.
16809 * opts-common.c (handle_option): Pass
16810 target_option_override_hook.
16811 * opts-global.c (lang_handle_option): Add new option.
16812 (set_default_handlers): Add new argument.
16813 (decode_options): Likewise.
16814 * opts.c (target_handle_option): Likewise.
16815 (common_handle_option): Call target_option_override_hook.
16816 * opts.h (struct cl_option_handler_func): Add hook for
16817 target option override.
16818 (struct cl_option_handlers): Likewise.
16819 (set_default_handlers): Add new argument.
16820 (decode_options): Likewise.
16821 (common_handle_option): Likewise.
16822 (target_handle_option): Likewise.
16823 * toplev.c (toplev::main): Pass targetm.target_option.override
16826 2017-08-29 Richard Biener <rguenther@suse.de>
16827 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16829 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16830 life to the active subtree.
16832 2017-08-28 Jeff Law <law@redhat.com>
16834 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16835 derive_equivalences.
16836 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16837 Code moved into....
16838 (edge_info::derive_equivalences): New private member function
16840 * tree-ssa-dom.c (class edge_info): Changed from a struct
16841 to a class. Add ctor/dtor, methods and data members.
16842 (edge_info::edge_info): Renamed from allocate_edge_info.
16843 Initialize additional members.
16844 (edge_info::~edge_info): New.
16845 (free_dom_edge_info): Delete the edge info.
16846 (record_edge_info): Use new class & associated member functions.
16847 Tighten forms for testing for edge equivalences.
16848 (record_temporary_equivalences): Iterate over the simple
16849 equivalences rather than assuming there's only one per edge.
16850 (cprop_into_successor_phis): Iterate over the simple
16851 equivalences rather than assuming there's only one per edge.
16852 (optimize_stmt): Use operand_equal_p rather than pointer
16853 equality for mini-DSE code.
16855 2017-08-28 Nathan Sidwell <nathan@acm.org>
16857 * gcc.c (execute): Fold SIGPIPE handling into switch
16858 statement. Adjust internal error message.
16860 2017-08-28 Richard Biener <rguenther@suse.de>
16863 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16864 Do nothing for removed DIEs.
16866 2017-08-28 Richard Biener <rguenther@suse.de>
16868 PR tree-optimization/81977
16869 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16872 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16875 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16876 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16877 using targetm.gen_mem_thread_fence.
16879 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16882 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16883 predicate to register_operand. Reorder operands.
16884 (*btr<mode>): Ditto.
16885 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16886 (*btr<mode>_mask): Ditto.
16888 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16890 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16891 * config/rs6000/xmmintrin.h: New file.
16892 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16894 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16897 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16898 parameter and_insn and return it.
16899 (recombine_lvx_pattern): Insert a copy to ensure availability of
16900 the base register of the copied masking operation at the point of
16901 the instruction replacement.
16902 (recombine_stvx_pattern): Likewise.
16904 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16906 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16907 undocumented switches.
16908 (-mpower9-dform-vector): Likewise.
16909 (-mpower9-dform): Likewise.
16910 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16911 comments to delete references to -mpower9-dform* switches.
16912 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16913 Delete reference to -mpower9-dform* switches, test for
16914 -mpower9-vector instead.
16915 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16916 (OTHER_P9_VECTOR_MASKS): Likewise.
16917 (POWERPC_MASKS): Likewise.
16918 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16919 tests against -mpower9-dform* to -mpower9-vector. Delete code
16920 that checked for -mpower9-dform* consistancy with other options.
16921 Add test for -mpower9-misc to enable other power9 switches.
16922 (rs6000_init_hard_regno_mode_ok): Likewise.
16923 (rs6000_option_override_internal): Likewise.
16924 (rs6000_emit_prologue): Likewise.
16925 (rs6000_emit_epilogue): Likewise.
16926 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16927 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16928 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16930 (emit_fusion_p9_store): Likewise.
16931 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16932 resetting these macros if the assembler does not support ISA 3.0
16934 (TARGET_P9_DFORM_VECTOR): Likewise.
16935 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16936 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16938 2017-08-25 Alan Modra <amodra@gmail.com>
16940 PR rtl-optimization/81747
16941 * cse.c (cse_extended_basic_block): Don't attempt to record
16942 equivalences for degenerate conditional branches that branch
16943 to their fall-through.
16945 2017-08-24 Martin Sebor <msebor@redhat.com>
16947 PR middle-end/81908
16948 * gimple-fold.c (size_must_be_zero_p): New function.
16949 (gimple_fold_builtin_memory_op): Call it.
16951 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16953 * config/rs6000/mm_malloc.h: New file.
16955 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16957 PR tree-optimization/81913
16958 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16959 analysis when either IVs in condition can wrap.
16961 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16963 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16964 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16966 2017-08-24 Richard Biener <rguenther@suse.de>
16969 * targhooks.c (default_target_can_inline_p): Properly
16970 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16971 is present and always compare.
16972 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16973 infer -mfpmath=sse from TARGET_SSE_P.
16974 (ix86_can_inline_p): Properly use target_option_default_node when
16975 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16977 2017-08-24 Richard Biener <rguenther@suse.de>
16980 * dwarf2out.c (output_die): Handle flag_generate_offload like
16982 (output_comp_unit): Likewise.
16983 (gen_array_type_die): Likewise.
16984 (dwarf2out_early_finish): Likewise.
16985 (note_variable_value_in_expr): Likewise.
16986 (dwarf2out_finish): Likewise. Adjust assert.
16987 * cgraphunit.c (symbol_table::compile): Move setting of
16988 flag_generate_offload earlier ...
16989 (symbol_table::finalize_compilation_unit): ... here, before
16990 early debug finalization.
16992 2017-08-24 Richard Biener <rguenther@suse.de>
16994 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16995 and ipa-fnsummary.h.
16996 (ix86_can_inline_p): When ix86_fpmath flags do not match
16997 check whether the callee uses FP math at all.
16999 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17001 PR middle-end/81931
17002 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17003 instead of TYPE_PRECISION.
17005 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17006 Alan Hayward <alan.hayward@arm.com>
17007 David Sherwood <david.sherwood@arm.com>
17009 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17011 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17013 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17015 (ix86_frame::stack_realign_allocate): New field.
17016 (struct machine_frame_state): Modify comments.
17017 (machine_frame_state::sp_realigned_fp_end): New field.
17018 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17019 layout calculation.
17020 (sp_valid_at): Add assertion to assure no attempt to access invalid
17021 offset of a realigned stack.
17022 (fp_valid_at): Likewise.
17023 (choose_baseaddr): Modify comments.
17024 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17025 ix86_expand_prologue.
17026 (ix86_expand_prologue): Modify stack realignment and allocation.
17027 (ix86_expand_epilogue): Modify comments.
17028 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17029 avx2_runtime, avx512f, and avx512f_runtime.
17031 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17033 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17034 (mstackrealign): Do not init to -1.
17035 * config/i386/i386.c (ix86_option_override_internal):
17036 Check opts_set, not opts when setting default value of
17037 opts->x_ix86_force_align_arg_pointer.
17039 2017-08-23 Richard Biener <rguenther@suse.de>
17041 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17042 lang_hooks.decl_printable_name.
17043 * print-rtl-function.c (print_rtx_function): Likewise.
17044 * tree-pretty-print.c (dump_function_header): Likewise.
17046 2017-08-23 Richard Biener <rguenther@suse.de>
17049 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17050 -g0 at compile-time.
17052 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17054 PR middle-end/19706
17055 * doc/sourcebuild.texi (Other hardware attributes):
17058 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17060 PR middle-end/19706
17061 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17062 Fix single-use check.
17064 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17066 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17068 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17070 * doc/install.texi: Modify to add more details on running selected
17073 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17075 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17076 is combined with -mabi=ms.
17077 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17080 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17082 PR tree-optimization/81488
17083 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17084 and cached_basis fields.
17085 (MAX_SPREAD): New constant.
17086 (alloc_cand_and_find_basis): Initialize new fields.
17087 (clear_visited): New function.
17088 (create_phi_basis_1): Rename from create_phi_basis, set visited
17089 and cached_basis fields.
17090 (create_phi_basis): New wrapper function.
17091 (phi_add_costs_1): Rename from phi_add_costs, add spread
17092 parameter, set visited field, short-circuit when limits reached.
17093 (phi_add_costs): New wrapper function.
17094 (record_phi_increments_1): Rename from record_phi_increments, set
17096 (record_phi_increments): New wrapper function.
17097 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17098 (phi_incr_cost): New wrapper function.
17099 (all_phi_incrs_profitable_1): Rename from
17100 all_phi_incrs_profitable, set visited field.
17101 (all_phi_incrs_profitable): New wrapper function.
17103 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17104 Alan Hayward <alan.hayward@arm.com>
17105 David Sherwood <david.sherwood@arm.com>
17107 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17108 that takes the outer and inner modes.
17109 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17110 comparison as the canonical test for a paradoxical subreg.
17111 * combine.c (simplify_set): Use paradoxical_subreg_p.
17112 (make_extraction): Likewise.
17113 (force_to_mode): Likewise.
17114 (rtx_equal_for_field_assignment_p): Likewise.
17115 (gen_lowpart_for_combine): Likewise.
17116 (simplify_comparison): Likewise.
17117 * cse.c (equiv_constant): Likewise.
17118 * expmed.c (store_bit_field_1): Likewise.
17119 * final.c (alter_subreg): Likewise.
17120 * fwprop.c (propagate_rtx): Likewise.
17121 (forward_propagate_subreg): Likewise.
17122 * ira-conflicts.c (ira_build_conflicts): Likewise.
17123 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17124 * lra-constraints.c (curr_insn_transform): Likewise.
17125 (split_reg): Likewise.
17126 * lra-eliminations.c (move_plus_up): Likewise.
17127 (lra_eliminate_regs_1): Likewise.
17128 * recog.c (general_operand): Likewise.
17129 * ree.c (combine_reaching_defs): Likewise.
17130 * reload.c (push_reload): Likewise.
17131 (find_reloads): Likewise.
17132 * reload1.c (elimination_effects): Likewise.
17133 (compute_reload_subreg_offset): Likewise.
17134 (choose_reload_regs): Likewise.
17135 * rtlanal.c (subreg_lsb_1): Likewise.
17136 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17137 (simplify_subreg): Likewise.
17138 * var-tracking.c (track_loc_p): Likewise.
17139 * emit-rtl.c (byte_lowpart_offset): Likewise.
17140 (paradoxical_subreg_p): Delete out-of-line definition.
17142 2017-08-22 Jeff Law <law@redhat.com>
17144 PR tree-optimization/81741
17145 PR tree-optimization/71947
17146 * tree-ssa-dom.c: Include tree-inline.h.
17147 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17148 equivalences if one is more expensive to compute than the other.
17149 * tree-ssa-scopedtables.h (class const_or_copies): Make
17150 record_const_or_copy_raw method private.
17151 (class avail_exprs_stack): New method simplify_binary_operation.
17152 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17153 avail_exprs_stack::simplify_binary_operation as needed.
17154 (avail_exprs_stack::simplify_binary_operation): New function.
17156 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17158 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17159 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17160 (DOT_SYMBOLS): Likewise.
17161 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17162 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17163 (RS6000_ABI_NAME): Likewise.
17164 (TARGET_CMODEL): Likewise.
17165 (TOC_SECTION_ASM_OP): Likewise.
17166 (SET_CMODEL): New macro.
17167 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17169 2017-08-22 Richard Biener <rguenther@suse.de>
17171 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17172 to allow for free-lang-data replacements similar to verify_type_variant.
17174 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17175 Alan Hayward <alan.hayward@arm.com>
17176 David Sherwood <david.sherwood@arm.com>
17178 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17179 VOIDmode for the LABEL_REF.
17181 2017-08-22 Richard Biener <rguenther@suse.de>
17183 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17185 2017-08-22 Richard Biener <rguenther@suse.de>
17187 * common.opt (feliminate-dwarf2-dups): Ignore.
17188 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17189 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17190 same_die_p_wrap, compute_section_prefix,
17191 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17192 (comdat_symbol_id, comdat_symbol_number): Likewise.
17193 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17195 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17196 (output_die): Mark unreachable path unreachable.
17197 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17198 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17199 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17200 (dwarf2out_early_finish): Likewise.
17202 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17204 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17206 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17209 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17210 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17211 (avr_attribute_table) <io, io_low, address>: Initialize
17212 .decl_required with true.
17214 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17216 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17217 undocumented debugging options.
17218 (-mvsx-scalar-double): Likewise.
17219 (-mallow-df-permute): Likewise.
17220 (-mvectorize-builtins): Likewise.
17221 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17222 (rs6000_builtin_vectorized_function): Likewise.
17223 (rs6000_builtin_md_vectorized_function): Likewise.
17224 (rs6000_opt_vars): Likewise.
17226 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17229 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17230 (*btrq_imm): Rename from *btrq.
17231 (*btcq_imm): Rename from *btcq.
17232 (btsc): New code attribute.
17233 (*<btsc><mode>): New insn pattern.
17234 (*btr<mode>): Ditto.
17235 (*<btsc><mode>_mask): New insn_and_split pattern.
17236 (*btr<mode>_mask): Ditto.
17238 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17239 Alan Hayward <alan.hayward@arm.com>
17240 David Sherwood <david.sherwood@arm.com>
17242 * function.c (pad_below): Simplify padding calculation.
17244 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17245 Alan Hayward <alan.hayward@arm.com>
17246 David Sherwood <david.sherwood@arm.com>
17248 * target.def (function_prologue): Remove frame size argument.
17249 (function_epilogue): Likewise.
17250 * doc/tm.texi: Regenerate.
17251 * final.c (final_start_function): Update call to function_prologue.
17252 (final_end_function): Update call to function_epilogue.
17253 (default_function_pro_epilogue): Remove frame size argument.
17254 * output.h (default_function_pro_epilogue): Likewise.
17255 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17256 (arm_output_function_prologue): Likewise.
17257 * config/frv/frv.c (frv_function_prologue): Likewise.
17258 (frv_function_epilogue): Likewise.
17259 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17260 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17261 (ia64_output_function_epilogue): Likewise.
17262 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17263 (m32r_output_function_epilogue): Likewise.
17264 * config/microblaze/microblaze.c (microblaze_function_prologue)
17265 (microblaze_function_epilogue): Likewise.
17266 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17267 (mips_output_function_epilogue): Likewise.
17268 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17269 (mmix_target_asm_function_epilogue): Likewise.
17270 * config/msp430/msp430.c (msp430_start_function): Likewise.
17271 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17272 (nds32_asm_function_epilogue): Likewise.
17273 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17274 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17275 (pa_output_function_epilogue): Likewise.
17276 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17277 (rs6000_output_function_epilogue): Likewise.
17278 * config/rl78/rl78.c (rl78_start_function): Likewise.
17279 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17280 (rs6000_output_function_epilogue): Likewise.
17281 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17282 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17283 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17284 (sparc_asm_function_epilogue): Likewise.
17286 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17288 * tree.h (type_has_mode_precision_p): New function.
17289 * convert.c (convert_to_integer_1): Use it.
17290 * expr.c (expand_expr_real_2): Likewise.
17291 (expand_expr_real_1): Likewise.
17292 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17293 * match.pd: Likewise.
17294 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17295 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17296 * tree-tailcall.c (process_assignment): Likewise.
17297 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17298 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17299 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17300 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17301 (vectorizable_assignment): Likewise.
17302 (vectorizable_shift): Likewise.
17303 (vectorizable_operation): Likewise.
17304 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17306 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17308 * match.pd: Add pow (C, x) simplification.
17310 2017-08-21 Richard Biener <rguenther@suse.de>
17312 PR tree-optimization/81900
17313 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17314 for blocks with abnormal predecessors.
17315 (compute_antic): Do not set visited flag prematurely.
17317 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17320 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17322 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17324 * stor-layout.h (vector_type_mode): Move to...
17325 * tree.h (vector_type_mode): ...here.
17326 * stor-layout.c (vector_type_mode): Move to...
17327 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17329 2017-08-21 Richard Biener <rguenther@suse.de>
17331 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17332 register_external_die hooks.
17333 (debug_false_tree_charstarstar_uhwistar): Declare.
17334 (debug_nothing_tree_charstar_uhwi): Likewise.
17335 * debug.c (do_nothing_debug_hooks): Adjust.
17336 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17337 (debug_nothing_tree_charstar_uhwi): Likewise.
17338 * dbxout.c (dbx_debug_hooks): Adjust.
17339 (xcoff_debug_hooks): Likewise.
17340 * sdbout.c (sdb_debug_hooks): Likewise.
17341 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17342 * dwarf2out.c (macinfo_label_base): New global.
17343 (dwarf2out_register_external_die): New function for the
17344 register_external_die hook.
17345 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17346 (dwarf2_debug_hooks): Use them.
17347 (dwarf2_lineno_debug_hooks): Adjust.
17348 (struct die_struct): Add with_offset flag.
17349 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17350 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17351 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17352 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17353 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17354 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17355 defining section names for the early LTO debug variants.
17356 (reset_indirect_string): New helper.
17357 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17358 (print_dw_val): Add support for offsetted symbol references.
17359 (get_ultimate_context): Split out from is_cxx.
17360 (is_cxx): Use get_ultimate_context.
17361 (is_fortran): Add decl overload.
17362 (compute_comp_unit_symbol): Split out worker from
17363 compute_section_prefix.
17364 (compute_section_prefix): Call compute_comp_unit_symbol and
17365 set comdat_type_p here.
17366 (output_die): Skip DIE symbol output for the LTO added one.
17367 Handle DIE symbol references with offset.
17368 (output_comp_unit): Guard section name mangling properly.
17369 For LTO debug sections emit a symbol at the section beginning
17370 which we use to refer to its DIEs.
17371 (add_abstract_origin_attribute): For DIEs registered via
17372 dwarf2out_register_external_die directly refer to the early
17373 DIE rather than indirectly through the shadow one we created.
17374 Remove obsolete call to dwarf2out_abstract_function for
17375 non-function/block origins.
17376 (gen_array_type_die): When generating early LTO debug do
17377 not emit DW_AT_string_length.
17378 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17379 late when in LTO. As suggested place a gcc_unreachable for
17380 the DECL_ABSTRACT_P case.
17381 (gen_subprogram_die): Avoid another specification DIE
17382 for early built declarations/definitions for the late LTO case.
17383 (gen_variable_die): Add type references for late duplicated VLA dies
17385 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17386 we have the abstract instance already.
17387 (process_scope_var): Adjust decl DIE contexts in LTO which
17388 first puts them in limbo.
17389 (gen_decl_die): Do not generate type DIEs late apart from
17390 types for VLAs or for decls we do not yet have a DIE. Do not
17391 call dwarf2out_abstract_function late.
17392 (dwarf2out_early_global_decl): Make sure to create DIEs
17393 for abstract instances of a decl first.
17394 (dwarf2out_late_global_decl): Adjust comment.
17395 (output_macinfo_op): With multiple macro sections use
17396 macinfo_label_base to distinguish labels.
17397 (output_macinfo): Likewise. Update macinfo_label_base.
17398 Pass in the line info label.
17399 (note_variable_value_in_expr): When generating LTO resolve
17400 all variable values here by generating DIEs as needed.
17401 (init_sections_and_labels): Add early LTO debug flag parameter
17402 and generate different sections and names if set. Add generation
17403 counter for the labels so we can have multiple of them.
17404 (reset_dies): Helper to allow DIEs to be output multiple times.
17405 (dwarf2out_finish): When outputting DIEs to the fat part of an
17406 LTO object first reset DIEs.
17407 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17408 (modified_type_die): Check for decl_ultimate_origin being self
17410 (gen_type_die_with_usage): Likewise.
17411 (gen_typedef_die): Allow decl_ultimate_origin being self.
17412 (set_decl_abstract_flags): Remove.
17413 (set_block_abstract_flags): Likewise.
17414 (dwarf2out_abstract_function): Treat the early generated DIEs
17415 as the abstract copy and only add DW_AT_inline and
17416 DW_AT_artificial here and call set_decl_origin_self.
17417 If the DIE has an abstract origin don't do anything.
17418 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17419 if we have none yet (Go fails to build one, PR78628).
17420 (variably_modified_type_p): Prevent endless recursion for Ada
17421 cyclic pointer types.
17422 * lto-streamer-in.c: Include debug.h.
17423 (dref_queue): New global.
17424 (lto_read_tree_1): Stream in DIE references.
17425 (lto_input_tree): Register DIE references.
17426 (input_function): Stream DECL_DEBUG_ARGS.
17427 * lto-streamer-out.c: Include debug.h.
17428 (lto_write_tree_1): Output DIE references.
17429 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17430 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17431 (output_function): Stream DECL_DEBUG_ARGS.
17432 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17433 Stream DECL_ABSTRACT_ORIGIN.
17434 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17435 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17436 DECL_CONTEXT for file-scope decls.
17437 * lto-streamer.h (struct dref_entry): Declare.
17438 (dref_queue): Likewise.
17439 * cfgexpand.c (pass_expand::execute): Do not call the
17440 outlining_inline_function hook here.
17441 * lto-wrapper.c (debug_obj): New global.
17442 (tool_cleanup): Unlink it if required.
17443 (debug_objcopy): New function.
17444 (run_gcc): Handle early debug sections in the IL files by
17445 extracting them to separate files, partially linkin them and
17446 feeding the result back as result to the linker.
17447 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17448 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17449 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17450 sections into a separate segment.
17451 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17453 (darwin_asm_dwarf_section): Likewise.
17454 (darwin_asm_output_dwarf_offset): Likewise.
17455 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17457 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17458 Alan Hayward <alan.hayward@arm.com>
17459 David Sherwood <david.sherwood@arm.com>
17461 * read-md.h (md_reader::record_potential_iterator_use): Replace
17462 pointer argument with an rtx and an index.
17463 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17464 (apply_mode_iterator): Likewise.
17465 (apply_code_iterator): Likewise.
17466 (apply_int_iterator): Likewise.
17467 (apply_subst_iterator): Likewise.
17468 (record_iterator_use): Likewise.
17469 (record_attribute_use): Likewise.
17470 (md_reader::record_potential_iterator_use): Likewise. Update calls
17471 to record_iterator_use and apply_iterator.
17472 (iterator_use): Replace ptr with x and index.
17473 (attribute_use): Likewise.
17474 (apply_attribute_uses): Update calls to apply_iterator.
17475 (apply_iterators): Likewise. Update initialization of iterator_use.
17476 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17477 and record_potential_iterator_use.
17478 (rtx_reader::read_rtx_operand): Likewise.
17480 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17481 Alan Hayward <alan.hayward@arm.com>
17482 David Sherwood <david.sherwood@arm.com>
17484 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17487 2017-08-21 Richard Biener <rguenther@suse.de>
17489 PR middle-end/81884
17490 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17491 at struct end conservatively when comparing common bases.
17493 2017-08-21 Richard Biener <rguenther@suse.de>
17495 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17496 (mem_ref_in_stmt): Remove.
17497 (determine_max_movement): Use ref index to get at the reference.
17498 (invariantness_dom_walker::before_dom_children): Deal with
17499 lim data already initialized.
17500 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17502 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17504 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17505 (TARGET_ISA_ROUND): Ditto.
17506 (TARGET_ROUND): Ditto.
17507 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17508 * config/i386/i386.md: Ditto.
17509 * config/i386/sse.md: Ditto.
17510 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17511 with OPTION_MASK_ISA_SSE4_1.
17513 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17516 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17517 __builtin_ia32_lzcnt_u16.
17519 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17522 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17523 (rs6000_set_current_function): Rewrite function to use it.
17525 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17528 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17529 and TYPE_WARN_IF_NOT_ALIGN.
17530 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17531 (handle_warn_if_not_align): New.
17532 (place_union_field): Call handle_warn_if_not_align.
17533 (place_field): Call handle_warn_if_not_align.
17534 Copy TYPE_WARN_IF_NOT_ALIGN.
17535 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17536 (layout_type): Likewise.
17537 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17539 (tree_decl_common): Add warn_if_not_align.
17540 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17541 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17542 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17543 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17544 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17545 * doc/extend.texi: Document warn_if_not_aligned attribute.
17546 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17548 2017-08-17 Martin Liska <mliska@suse.cz>
17551 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17552 (get_data_dependence): Use it as pointer type.
17553 (distribute_loop): Likewise.
17555 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17557 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17558 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17559 (p8_vmrgow_<mode>_direct): New define_insn.
17560 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17561 handle endianness for vmrgew and vmrgow permute patterns.
17563 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17565 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17566 * config/rs6000/rs6000-cpus.def: Remove comment.
17567 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17568 (POWERPC_MASKS): Likewise.
17569 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17570 use of TARGET_VSX_TIMODE.
17571 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17572 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17573 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17574 (rs6000_option_override_internal): Remove dead code.
17575 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17576 (rs6000_legitimize_reload_address): Likewise.
17577 (rs6000_legitimate_address_p): Likewise.
17578 (rs6000_opt_masks): Delete "vsx-timode".
17579 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17580 from function comment.
17581 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17582 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17583 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17585 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17586 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17587 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17588 (VSX_TI): Likewise.
17590 (define_peephole2): Likewise.
17592 2017-08-17 Martin Sebor <msebor@redhat.com>
17595 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17596 past the end of an array.
17597 (test_pp_format): Add test cases.
17599 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17601 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17602 missing ECF_NOTHROW flags.
17604 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17607 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17608 operands residing in integer registers.
17609 (*vsx_le_perm_load_<mode>): Likewise.
17610 (*vsx_le_perm_store_<mode>): Likewise.
17611 (define_peephole2): Add peepholes to optimize the above.
17613 2017-08-17 Marek Polacek <polacek@redhat.com>
17615 PR middle-end/81814
17616 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17617 to mimic what shorten_compare did. Change the return type to bool.
17618 (fold_cond_expr_with_comparison): Update call to
17619 operand_equal_for_comparison_p.
17620 (fold_ternary_loc): Likewise.
17622 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17624 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17626 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17628 2017-08-17 Richard Biener <rguenther@suse.de>
17630 * tree-ssa-structalias.c (solve_graph): When propagating
17631 to successors update the graphs succ edges and avoid duplicate work.
17633 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17636 * config/i386/i386.c (ix86_option_override_internal): Save target
17637 specific options after ix86_stack_protector_guard_reg was changed.
17639 2017-08-17 Richard Biener <rguenther@suse.de>
17641 PR tree-optimization/81827
17642 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17643 (new_var_info): Initialize it conservatively.
17644 (get_call_vi): Mark register vars.
17645 (new_scalar_tmp_constraint_exp): Likewise.
17646 (handle_rhs_call): Likewise.
17647 (handle_const_call): Likewise.
17648 (create_function_info_for): Likewise.
17649 (solve_constraints): Sort varinfos to separate register from
17650 non-register vars to pack points-to solution bitmaps during
17653 2017-08-17 Marek Polacek <polacek@redhat.com>
17655 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17657 2017-08-17 Richard Biener <rguenther@suse.de>
17659 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17660 to true when overflow is undefined and we saturated the result.
17662 2017-08-17 Alan Modra <amodra@gmail.com>
17665 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17666 Don't use store multiple if only one reg needs saving.
17667 (interesting_frame_related_regno): New function.
17668 (rs6000_frame_related): Don't emit frame info for regs that
17670 (rs6000_emit_epilogue): Likewise.
17672 2017-08-16 Nathan Sidwell <nathan@acm.org>
17674 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17675 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17676 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17677 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17678 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17679 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17680 (verify_type): Adjust for TYPE_BINFO move.
17681 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17682 process TYPE_BINFO directly.
17683 (hash_tree): Likewise.
17684 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17686 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17689 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17691 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17693 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17696 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17697 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17698 Add ix86_binary_operator_ok to insn constraint.
17699 (*iordi_1_bts): Ditto.
17700 (*xordi_1_btc): Ditto.
17701 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17702 Update corresponding peephole2 pattern.
17706 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17708 PR tree-optimization/81832
17709 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17710 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17712 2017-08-16 Marek Polacek <polacek@redhat.com>
17715 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17716 regarding size_zero_node.
17718 2017-08-16 Martin Liska <mliska@suse.cz>
17721 * config.gcc: Respect previously set extra_objs in case
17724 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17726 PR tree-optimization/81835
17727 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17728 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17729 not depend on the phi.
17731 2017-08-16 Alan Modra <amodra@gmail.com>
17733 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17736 2017-08-16 Alan Modra <amodra@gmail.com>
17738 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17739 (save_reg_p): ..into this. Update all callers.
17740 (first_reg_to_save): Simplify.
17742 2017-08-16 Alan Modra <amodra@gmail.com>
17744 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17747 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17751 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17752 constant count if that count is less than 32.
17754 2017-08-15 Nathan Sidwell <nathan@acm.org>
17756 * gcc.c (execute): Emit friendlier message if inferior is killed
17757 by an external cause.
17759 2017-08-15 Richard Biener <rguenther@suse.de>
17761 PR tree-optimization/81790
17762 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17763 CONSTRUCTORs from simplifying and VN.
17765 2017-08-14 Martin Sebor <msebor@redhat.com>
17767 * builtin-attrs.def: Add comments.
17769 2017-08-14 Martin Sebor <msebor@redhat.com>
17772 * doc/extend.texi (attribute nonstring): Document new attribute.
17774 2017-08-14 Martin Sebor <msebor@redhat.com>
17777 * tree-diagnostic.c (default_tree_printer): Handle %G.
17778 * gimple-pretty-print.h (percent_G_format): Declare new function.
17779 * gimple-pretty-print.c (percent_G_format): Define.
17780 * tree-pretty-print.c (percent_K_format): Add argument.
17782 2017-08-14 Martin Sebor <msebor@redhat.com>
17784 PR translation/79998
17785 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17786 Remove a stray space.
17788 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17791 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17792 (*iordi_1_bts): Ditto.
17793 (*xordi_1_btc): Ditto.
17795 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17798 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17799 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17801 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17802 quoted strings, and make more translator-friendly.
17803 (darwin_rs6000_override_options): Likewise.
17804 (rs6000_option_override_internal): Likewise.
17805 (rs6000_return_in_memory): Fix overlong line.
17806 (init_cmulative_args): Use quoted strings, and make more
17807 translator-friendly.
17808 (rs6000_pass_by_reference): Fix overlong line.
17809 (def_builtin): Use quoted strings.
17810 (altivec_expand_predicate_builtin): Use quoted strings, and make
17811 more translator-friendly.
17812 (htm_expand_builtin): Use quoted strings.
17813 (cpu_expand_builtin): Use quoted strings, and make more
17814 translator-friendly.
17815 (altivec_expand_builtin): Likewise.
17816 (paired_expand_predicate_builtin): Likewise.
17817 (rs6000_invalid_builtin): Likewise.
17818 (builtin_function_type): Use quoted strings.
17819 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17820 more translator-friendly.
17821 (rs6000_trampoline_init): Likewise.
17822 (rs6000_handle_altivec_attribute): Likewise.
17823 (rs6000_inner_target_options): Use quoted strings.
17824 (rs6000_disable_incompatible_switches): Likewise.
17825 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17826 strings, and make more translator-friendly.
17827 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17829 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17831 PR tree-optimization/81799
17832 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17833 cond_expr to simple gimple operand.
17835 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17837 PR middle-end/46932
17838 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17840 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17844 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17845 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17846 TARGET_GASISR_PROLOGUES.
17847 * config/avr/avr.c (avr_option_override): Same.
17848 (avr_pass_pre_proep::execute): Same.
17850 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17853 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17854 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17856 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17858 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17859 define_insn_and_split. Split to a memory load from 0 in
17860 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17861 using PTR mode iterator.
17862 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17863 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17864 (*add_tp_<mode>): Redefine as define_insn_and_split.
17865 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17866 address space. Merge with *add_tp_x32 using PTR mode iterator.
17867 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17868 Split to an add with a memory load from 0 in
17869 DEFAULT_TLS_SEG_REG address space.
17871 2017-08-12 Andrew Pinski <apinski@cavium.com>
17873 * config/aarch64/aarch64-option-extensions.def (rdma):
17874 Fix feature string to what Linux prints out in /proc/cpuinfo.
17876 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17879 * dwarf2out.c (modified_type_die): For C typedef types that have
17880 an ultimate origin, process the ultimate origin instead of the
17882 (gen_typedef_die): Assert that input DECLs have no ultimate
17884 (gen_type_die_with_usage): For typedef variants that have an
17885 ultimate origin, just call gen_decl_die on the original DECL.
17886 (process_scope_var): Avoid creating DIEs for local typedefs and
17887 concrete static variables.
17889 2017-08-12 Alan Modra <amodra@gmail.com>
17893 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17894 match gnu-user.h startfile.
17895 (ENDFILE_LINUX_SPEC): Similarly.
17897 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17900 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17902 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17904 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17905 * config/aarch64/aarch64.md (mov<mode>): Change.
17906 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17907 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17908 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17910 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17912 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17913 for storage order barriers.
17915 2017-08-11 Martin Liska <mliska@suse.cz>
17918 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17919 variables of void type.
17921 2017-08-11 Martin Liska <mliska@suse.cz>
17923 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17924 TARGET_SUPPORTS_ALIASES.
17925 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17926 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17927 (optimize_weakref): Likewise.
17928 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17929 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17930 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17932 2017-08-11 Martin Liska <mliska@suse.cz>
17935 * config/i386/i386.c (make_resolver_func): Do complete
17936 refactoring of the function.
17938 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17941 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17942 * config/i386/i386.c (ix86_stack_protect_guard): Use
17943 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17944 * doc/invoke.texi (x86 Options): Document
17945 -mstack-protector-guard-symbol= option.
17947 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17949 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17950 * config/i386/i386.c (ix86_split_stack_guard): New function.
17951 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17952 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17953 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17954 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17955 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17956 (split_stack_space_check): Call ix86_split_stack_guard.
17958 2017-08-10 Martin Sebor <msebor@redhat.com>
17960 * print-tree.c (print_node): Print location using the established
17962 Replace spaces with colons.
17963 (debug_raw, debug): Ditto.
17965 2017-08-10 Martin Sebor <msebor@redhat.com>
17968 * pretty-print.c (pp_format): Correct the handling of %s precision.
17970 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17973 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17975 (ix86_finalize_stack_frame_flags): This. Also clear
17976 frame_pointer_needed if -fno-omit-frame-pointer is used without
17978 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17979 with ix86_finalize_stack_frame_flags.
17980 (ix86_expand_epilogue): Likewise.
17981 (ix86_expand_split_stack_prologue): Likewise.
17982 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17984 2017-08-10 Martin Liska <mliska@suse.cz>
17987 * c-attribs.c (handle_target_attribute):
17988 Report warning for an empty string argument of target attribute.
17990 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17993 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17995 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17996 or DECL_NONLOCAL labels.
17997 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17998 or DECL_NONLOCAL labels here.
18000 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18002 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18003 to indicate when early gimple folding has been disabled.
18004 (rs6000_gimple_fold_builtin): Add debug content.
18005 (rs6000_invalid_builtin): Fix whitespace.
18006 (rs6000_expand_builtin): Fix whitespace.
18007 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18009 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18012 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18013 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18015 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18017 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18018 (qdf24xx): Likewise.
18019 * config/aarch64/aarch64.md: Include falkor.md.
18020 * config/aarch64/falkor.md: New.
18022 2017-08-09 Marek Polacek <polacek@redhat.com>
18025 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18026 * diagnostic.c (emit_diagnostic): Add a comment.
18027 (emit_diagnostic_valist): New function.
18029 2017-08-09 Marek Polacek <polacek@redhat.com>
18032 * input.c (make_location): New overload.
18033 * input.h (make_location): Declare.
18035 2017-08-08 Alan Modra <amodra@gmail.com>
18036 H.J. Lu <hongjiu.lu@intel.com>
18039 * gcc.c (NO_PIE_SPEC): Delete.
18040 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18042 (LINK_PIE_SPEC): ..to here.
18043 (LINK_COMMAND_SPEC): Support -no-pie.
18044 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18045 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18047 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18048 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18049 (ENDFILE_CRTEND_SPEC): Similarly.
18051 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18054 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18055 (mstack-protector-guard-offset=): Ditto.
18056 * config/i386/i386.c (ix86_option_override): Handle
18057 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18059 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18060 ix86_stack_protect_guard_offset variables.
18061 (TARGET_STACK_PROTECT_GUARD): Always define.
18062 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18063 and -mstack-protector-guard-offset= options.
18065 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18067 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18068 boundary case for the last candidate.
18070 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18072 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18073 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18075 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18077 PR middle-end/19706
18078 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18079 * config/aarch64/aarch64-builtins.c
18080 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18081 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18082 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18084 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18085 Andrew Pinski <pinskia@gmail.com>
18087 PR middle-end/19706
18088 * internal-fn.def (XORSIGN): New.
18089 * optabs.def (xorsign_optab): New.
18090 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18091 (convert_expand_mult_copysign): New.
18092 (pass_optimize_widening_mul::execute): Call
18093 convert_expand_mult_copysign.
18095 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18097 PR tree-optimization/81354
18098 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18099 Insert on edges rather than explicitly creating landing pads.
18100 (analyze_candidates_and_replace): Commit edge inserts.
18102 2017-08-08 Richard Biener <rguenther@suse.de>
18104 PR middle-end/81719
18105 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18106 (expand_simple_operations): Also look through ADDR_EXPRs with
18107 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18109 2017-08-08 Richard Biener <rguenther@suse.de>
18111 PR tree-optimization/81723
18112 * tree-vect-slp.c (struct bst_traits): New hash traits.
18113 (bst_fail): New global.
18114 (vect_build_slp_tree_2): New worker, split out from ...
18115 (vect_build_slp_tree): ... this now wrapping it with using
18116 bst_fail set to cache SLP tree build fails. Properly handle
18118 (vect_analyze_slp_instance): Allocate and free bst_fail.
18120 2017-08-08 Martin Liska <mliska@suse.cz>
18123 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18124 LABEL_DECLs that can be from a different function.
18126 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18128 PR tree-optimization/81744
18129 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18130 loop's number of iterations.
18132 2017-08-08 Martin Liska <mliska@suse.cz>
18134 * asan.c: Include header files.
18135 * attribs.c (build_decl_attribute_variant): New function moved
18137 (build_type_attribute_qual_variant): Likewise.
18138 (cmp_attrib_identifiers): Likewise.
18139 (simple_cst_list_equal): Likewise.
18140 (omp_declare_simd_clauses_equal): Likewise.
18141 (attribute_value_equal): Likewise.
18142 (comp_type_attributes): Likewise.
18143 (build_type_attribute_variant): Likewise.
18144 (lookup_ident_attribute): Likewise.
18145 (remove_attribute): Likewise.
18146 (merge_attributes): Likewise.
18147 (merge_type_attributes): Likewise.
18148 (merge_decl_attributes): Likewise.
18149 (merge_dllimport_decl_attributes): Likewise.
18150 (handle_dll_attribute): Likewise.
18151 (attribute_list_equal): Likewise.
18152 (attribute_list_contained): Likewise.
18153 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18154 (lookup_attribute_by_prefix): Likewise.
18155 * bb-reorder.c: Include header files.
18156 * builtins.c: Likewise.
18157 * calls.c: Likewise.
18158 * cfgexpand.c: Likewise.
18159 * cgraph.c: Likewise.
18160 * cgraphunit.c: Likewise.
18161 * convert.c: Likewise.
18162 * dwarf2out.c: Likewise.
18163 * final.c: Likewise.
18164 * fold-const.c: Likewise.
18165 * function.c: Likewise.
18166 * gimple-expr.c: Likewise.
18167 * gimple-fold.c: Likewise.
18168 * gimple-pretty-print.c: Likewise.
18169 * gimple.c: Likewise.
18170 * gimplify.c: Likewise.
18171 * hsa-common.c: Likewise.
18172 * hsa-gen.c: Likewise.
18173 * internal-fn.c: Likewise.
18174 * ipa-chkp.c: Likewise.
18175 * ipa-cp.c: Likewise.
18176 * ipa-devirt.c: Likewise.
18177 * ipa-fnsummary.c: Likewise.
18178 * ipa-inline.c: Likewise.
18179 * ipa-visibility.c: Likewise.
18181 * lto-cgraph.c: Likewise.
18182 * omp-expand.c: Likewise.
18183 * omp-general.c: Likewise.
18184 * omp-low.c: Likewise.
18185 * omp-offload.c: Likewise.
18186 * omp-simd-clone.c: Likewise.
18187 * opts-global.c: Likewise.
18188 * passes.c: Likewise.
18189 * predict.c: Likewise.
18190 * sancov.c: Likewise.
18191 * sanopt.c: Likewise.
18192 * symtab.c: Likewise.
18193 * toplev.c: Likewise.
18194 * trans-mem.c: Likewise.
18195 * tree-chkp.c: Likewise.
18196 * tree-eh.c: Likewise.
18197 * tree-into-ssa.c: Likewise.
18198 * tree-object-size.c: Likewise.
18199 * tree-parloops.c: Likewise.
18200 * tree-profile.c: Likewise.
18201 * tree-ssa-ccp.c: Likewise.
18202 * tree-ssa-live.c: Likewise.
18203 * tree-ssa-loop.c: Likewise.
18204 * tree-ssa-sccvn.c: Likewise.
18205 * tree-ssa-structalias.c: Likewise.
18206 * tree-ssa.c: Likewise.
18207 * tree-streamer-in.c: Likewise.
18208 * tree-vectorizer.c: Likewise.
18209 * tree-vrp.c: Likewise.
18210 * tsan.c: Likewise.
18211 * ubsan.c: Likewise.
18212 * varasm.c: Likewise.
18213 * varpool.c: Likewise.
18214 * tree.c: Remove functions moved to attribs.[ch].
18215 * tree.h: Likewise.
18216 * config/aarch64/aarch64.c: Add attrs.h header file.
18217 * config/alpha/alpha.c: Likewise.
18218 * config/arc/arc.c: Likewise.
18219 * config/arm/arm.c: Likewise.
18220 * config/avr/avr.c: Likewise.
18221 * config/bfin/bfin.c: Likewise.
18222 * config/c6x/c6x.c: Likewise.
18223 * config/cr16/cr16.c: Likewise.
18224 * config/cris/cris.c: Likewise.
18225 * config/darwin.c: Likewise.
18226 * config/epiphany/epiphany.c: Likewise.
18227 * config/fr30/fr30.c: Likewise.
18228 * config/frv/frv.c: Likewise.
18229 * config/ft32/ft32.c: Likewise.
18230 * config/h8300/h8300.c: Likewise.
18231 * config/i386/winnt.c: Likewise.
18232 * config/ia64/ia64.c: Likewise.
18233 * config/iq2000/iq2000.c: Likewise.
18234 * config/lm32/lm32.c: Likewise.
18235 * config/m32c/m32c.c: Likewise.
18236 * config/m32r/m32r.c: Likewise.
18237 * config/m68k/m68k.c: Likewise.
18238 * config/mcore/mcore.c: Likewise.
18239 * config/microblaze/microblaze.c: Likewise.
18240 * config/mips/mips.c: Likewise.
18241 * config/mmix/mmix.c: Likewise.
18242 * config/mn10300/mn10300.c: Likewise.
18243 * config/moxie/moxie.c: Likewise.
18244 * config/msp430/msp430.c: Likewise.
18245 * config/nds32/nds32-isr.c: Likewise.
18246 * config/nds32/nds32.c: Likewise.
18247 * config/nios2/nios2.c: Likewise.
18248 * config/nvptx/nvptx.c: Likewise.
18249 * config/pa/pa.c: Likewise.
18250 * config/pdp11/pdp11.c: Likewise.
18251 * config/powerpcspe/powerpcspe.c: Likewise.
18252 * config/riscv/riscv.c: Likewise.
18253 * config/rl78/rl78.c: Likewise.
18254 * config/rx/rx.c: Likewise.
18255 * config/s390/s390.c: Likewise.
18256 * config/sh/sh.c: Likewise.
18257 * config/sol2.c: Likewise.
18258 * config/sparc/sparc.c: Likewise.
18259 * config/spu/spu.c: Likewise.
18260 * config/stormy16/stormy16.c: Likewise.
18261 * config/tilegx/tilegx.c: Likewise.
18262 * config/tilepro/tilepro.c: Likewise.
18263 * config/v850/v850.c: Likewise.
18264 * config/vax/vax.c: Likewise.
18265 * config/visium/visium.c: Likewise.
18266 * config/xtensa/xtensa.c: Likewise.
18268 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18271 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18272 constraints since the -mupper-regs-* switches have been
18274 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18275 into a vector from a double word element that was extracted from
18276 another vector, and eliminate extra XXPERMDI instructions.
18277 (vsx_concat_<mode>_2): Likewise.
18278 (vsx_concat_<mode>_3): Likewise.
18279 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18280 concat to allow optimizing inserts from previous extracts.
18282 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18284 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18285 memory reference to a SSP offset in TLS address space.
18286 (ix86_print_operand) <case '@'>: Remove.
18287 (ix86_print_operand_punct_valid_p): Remove '@' code.
18288 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18289 UNSPEC_SP_TLS_TEST.
18290 (stack_tls_protect_set_<mode>): Remove.
18291 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18292 (stack_tls_protect_test_<mode>): Remove.
18293 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18295 2017-08-07 Olivier Hainque <hainque@adacore.com>
18298 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18300 2017-08-07 Douglas Rupp <rupp@adacore.com>
18302 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18303 variable was referenced as multidir in command.
18305 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18308 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18311 2017-08-07 Martin Liska <mliska@suse.cz>
18313 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18314 * config/rl78/rl78.c: Add include of attribs.h.
18315 * config/sh/sh.c: Likewise.
18316 * config/v850/v850.c: Likewise.
18318 2017-08-07 Tom de Vries <tom@codesourcery.com>
18320 PR middle-end/78266
18321 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18323 2017-08-07 Martin Liska <mliska@suse.cz>
18325 * config/mips/mips.c: Include attribs.h.
18327 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18330 * doc/invoke.texi: Document change in behvaior for -Ofast for
18333 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18335 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18337 (aarch64_pop_regs): Likewise.
18338 (aarch64_gen_load_pair): Likewise.
18339 (aarch64_save_callee_saves): Likewise.
18340 (aarch64_restore_callee_saves): Likewise.
18342 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18344 * config/i386/i386.c: Revert the last change.
18346 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18349 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18351 (ix86_finalize_stack_frame_flags): This. Also clear
18352 frame_pointer_needed if -fno-omit-frame-pointer is used without
18354 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18355 with ix86_finalize_stack_frame_flags.
18356 (ix86_expand_epilogue): Likewise.
18357 (ix86_expand_split_stack_prologue): Likewise.
18359 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18362 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18363 to P_AES for Westmere.
18365 2017-08-07 Jonathan Yong <10walls@gmail.com>
18367 * config/i386/mingw.opt (fset-stack-executable): Removed.
18368 * config/i386/cygming.opt (fset-stack-executable): Moved
18370 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18372 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18374 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18376 2017-08-07 Marek Polacek <polacek@redhat.com>
18378 PR middle-end/81737
18379 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18381 2017-08-07 Martin Liska <mliska@suse.cz>
18383 * attribs.h (canonicalize_attr_name): New function.
18384 (cmp_attribs): Move from c-format.c and adjusted.
18385 (is_attribute_p): Moved from tree.h.
18386 * tree-inline.c: Add new includes.
18387 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18388 (private_is_attribute_p): Remove.
18389 (private_lookup_attribute): Likewise.
18390 (private_lookup_attribute_by_prefix): Simplify.
18391 (remove_attribute): Use is_attribute_p.
18392 * tree.h: Remove removed declarations.
18394 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18396 PR middle-end/81698
18397 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18398 instead of computing it in the function. Formatting fix.
18399 (expand_case): Don't rely on default_edge being the first edge,
18400 clear it if removing it, pass default_edge to
18401 emit_case_dispatch_table.
18402 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18405 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18407 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18408 insn in the function, emit NOP after the insn.
18410 2017-08-06 Tom de Vries <tom@codesourcery.com>
18412 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18415 2017-08-06 Tom de Vries <tom@codesourcery.com>
18417 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18420 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18422 PR tree-optimization/57371
18423 * match.pd: New pattern.
18425 2017-08-04 Marek Polacek <polacek@redhat.com>
18427 PR middle-end/81695
18428 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18429 perform the computation in offset_int.
18431 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18433 PR tree-optimization/81136
18434 * tree-vectorizer.h: Include tree-hash-traits.h.
18435 (vec_base_alignments): New typedef.
18436 (vec_info): Add a base_alignments field.
18437 (vect_record_base_alignments): Declare.
18438 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18440 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18441 (create_data_ref): Add an is_conditional_in_stmt argument.
18442 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18443 the is_conditional_in_stmt field.
18444 (data_ref_loc): Add an is_conditional_in_stmt field.
18445 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18446 (find_data_references_in_stmt): Update call to create_data_ref.
18447 (graphite_find_data_references_in_stmt): Likewise.
18448 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18449 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18450 (vect_record_base_alignment): New function.
18451 (vect_record_base_alignments): Likewise.
18452 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18453 for nested statements even if we fail to compute a misalignment.
18454 Use pooled base alignments for unconditional references.
18455 (vect_find_same_alignment_drs): Compare base addresses instead
18457 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18458 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18460 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18462 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18463 Add an explicit name for the enum. Use auto_vec for slp_instances
18464 and grouped_stores.
18465 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18467 (_bb_vec_info): Add a constructor and destructor.
18468 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18469 (destroy_loop_vec_info): Delete.
18470 (vect_destroy_datarefs): Likewise.
18471 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18472 (vec_info::vec_info): New function.
18473 (vec_info::~vec_info): Likewise.
18474 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18475 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18476 destroy_loop_vec_info.
18477 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18478 (_loop_vec_info::_loop_vec_info): ...this.
18479 (destroy_loop_vec_info): Replace with...
18480 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18481 the stmt_vec_infos. Leave handling of vec_info information to its
18482 destructor. Remove explicit vector releases.
18483 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18484 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18485 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18486 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18487 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18488 (destroy_bb_vec_info): Replace with...
18489 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18490 information to its destructor.
18491 (vect_slp_analyze_bb_1): Use new and delete instead of
18492 new_bb_vec_info and destroy_bb_vec_info.
18493 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18496 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18498 * tree-data-ref.h (subscript): Add access_fn field.
18499 (data_dependence_relation): Add could_be_independent_p.
18500 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18501 (same_access_functions): Move to tree-data-ref.c.
18502 * tree-data-ref.c (ref_contains_union_access_p): New function.
18503 (access_fn_component_p): Likewise.
18504 (access_fn_components_comparable_p): Likewise.
18505 (dr_analyze_indices): Add a reference to access_fn_component_p.
18506 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18508 (constant_access_functions): Likewise.
18509 (add_other_self_distances): Likewise.
18510 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18511 (initialize_data_dependence_relation): Use XCNEW and remove
18512 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18513 of access functions that have the same type. Allow the
18514 subsequence to end with different bases in some circumstances.
18515 Record the chosen access functions in SUB_ACCESS_FN.
18516 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18517 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18518 (subscript_dependence_tester_1): Likewise dra and drb.
18519 (build_classic_dist_vector): Update calls accordingly.
18520 (subscript_dependence_tester): Likewise.
18521 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18522 DDR_COULD_BE_INDEPENDENT_P.
18523 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18524 comp_alias_ddrs instead of may_alias_ddrs.
18525 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18527 (vect_analyze_data_ref_dependence): Use it if
18528 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18529 distance vectors if that fails.
18530 (dependence_distance_ge_vf): New function.
18531 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18532 LOOP_VINFO_MAY_ALIAS_DDRS.
18534 2017-08-04 Richard Biener <rguenther@suse.de>
18536 PR middle-end/81705
18537 * fold-const.c (fold_binary_loc): Properly restrict
18538 minus_var0 && minus_var1 case when associating undefined overflow
18541 2017-08-04 Tom de Vries <tom@codesourcery.com>
18543 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18545 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18547 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18548 Don't start diagnostic messages with a capital letter.
18549 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18551 (rs6000_invalid_builtin): Likewise.
18552 (rs6000_trampoline_init): Likewise.
18554 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18557 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18558 after setting changeable df flags.
18560 2017-08-03 Richard Biener <rguenther@suse.de>
18562 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18563 up if the use is in USE - X.
18565 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18567 * toplev.c (dumpfile.h): New include.
18568 (internal_error_reentered): New static function. Use it...
18569 (internal_error_function): ...here to handle reentered internal_error.
18571 2017-08-03 Richard Biener <rguenther@suse.de>
18573 PR middle-end/81148
18574 * fold-const.c (split_tree): Add minus_var and minus_con
18575 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18576 here but always use minus_*.
18577 (associate_trees): Assert we never associate with MINUS_EXPR
18578 and NULL first operand. Do not recurse for PLUS_EXPR operands
18579 when associating as MINUS_EXPR either.
18580 (fold_binary_loc): Track minus_var and minus_con.
18582 2017-08-03 Tom de Vries <tom@codesourcery.com>
18585 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18586 ACCEL_COMPILER, apply finish_options on
18587 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18589 2017-08-03 Tom de Vries <tom@codesourcery.com>
18592 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18593 function_entry_patch_area_size > 0.
18595 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18598 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18599 instead of 10??LU, perform unit multiplication in wide_int,
18600 don't change alloc_object_size_limit if the limit is larger
18603 PR tree-optimization/81655
18604 PR tree-optimization/81588
18605 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18606 the case when ranges[i].low and high are 1 for unsigned type with
18609 PR middle-end/81052
18610 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18611 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18613 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18615 * tree-vrp.h: Add include guard.
18617 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18620 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18621 (ud2): New insn pattern.
18622 * config/i386/i386.c (ix86_expand_epilogue):
18623 For naked functions, generate ud2 instead of trap insn.
18625 2017-08-02 Marek Polacek <polacek@redhat.com>
18628 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18630 2017-08-02 Tom de Vries <tom@codesourcery.com>
18631 Cesar Philippidis <cesar@codesourcery.com>
18633 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18634 Add missing edge probabilities.
18636 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18638 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18639 Correct endianness.
18641 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18643 PR middle-end/79499
18644 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18645 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18646 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18647 prologue_seq sequences - if any.
18649 2017-08-02 Richard Biener <rguenther@suse.de>
18651 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18652 via vectors if supported, integer extracts via punning if supported
18653 or otherwise vector extracts.
18655 2017-08-02 Richard Biener <rguenther@suse.de>
18657 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18659 (bitmap_insert_into_set): ... this.
18661 2017-08-02 Richard Biener <rguenther@suse.de>
18663 PR tree-optimization/81633
18665 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18667 PR tree-optimization/71752
18668 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18670 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18672 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18673 (machine_function::call_ms2sysv_pad_out): Remove field.
18674 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18675 (ix86_compute_frame_layout): Likewise.
18677 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18680 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18681 attribute with interrupt attribute.
18683 2017-08-01 Andrew Pinski <apinski@cavium.com>
18685 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18686 BIT_INSERT_EXPR's operand 1
18687 to see if the types precision matches.
18689 2017-08-01 Martin Liska <mliska@suse.cz>
18691 PR middle-end/70140
18692 * builtins.c (expand_builtin_memcpy_args): Remove.
18693 (expand_builtin_memcpy): Call newly added function
18694 expand_builtin_memory_copy_args.
18695 (expand_builtin_memcpy_with_bounds): Likewise.
18696 (expand_builtin_mempcpy): Remove last argument.
18697 (expand_builtin_mempcpy_with_bounds): Likewise.
18698 (expand_builtin_memory_copy_args): New function created from
18699 expand_builtin_mempcpy_args with small modifications.
18700 (expand_builtin_mempcpy_args): Remove.
18701 (expand_builtin_stpcpy): Remove unused argument.
18702 (expand_builtin): Likewise.
18703 (expand_builtin_with_bounds): Likewise.
18705 2017-08-01 Martin Liska <mliska@suse.cz>
18708 Make mempcpy more optimal (PR middle-end/70140).
18710 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18713 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18714 __builtin_vec_cmpne verify both arguments are compatible vectors
18715 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18716 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18717 move computation of aligned to after checking the argument types.
18721 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18722 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18725 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18726 Doug Rupp <rupp@adacore.com>
18727 Olivier Hainque <hainque@adacore.com>
18729 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18730 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18732 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18733 for TARGET_OS_CPP_BUILTIN.
18734 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18735 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18737 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18738 passing required abi options to the assembler for EABI configurations.
18739 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18740 of .text.hot and .text.unlikely sections for kernel modules when
18741 using ARM style exceptions.
18742 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18743 options. Add EXTRA_CC1_SPEC.
18744 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18746 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18748 (ARM_TARGET2_DWARF_FORMAT): Define.
18749 * config/arm/t-vxworks: Adjust multilib control to removal of the
18750 Diab command line options.
18752 2017-08-01 Martin Liska <mliska@suse.cz>
18754 PR gcov-profile/81561
18755 * gcov.c (unblock): Make unblocking safe as we need to preserve
18756 index correspondence of blocks and block_lists.
18758 2017-08-01 Richard Biener <rguenther@suse.de>
18760 PR tree-optimization/81181
18761 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18762 (compute_antic): ... end of iteration here.
18764 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18766 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18767 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18768 (ftree-slp-vectorize): Likewise.
18769 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18770 can no longer be set independent of flag_tree_loop_vectorize.
18771 * omp-general.c (emp_max_vf): Likewise.
18772 * opts.c (enable_fdo_optimizations): Remove references to
18773 flag_tree_vectorize, these are now implicit.
18774 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18775 and leave it for the options machinery.
18777 2017-08-01 Martin Liska <mliska@suse.cz>
18779 PR middle-end/70140
18780 * builtins.c (expand_builtin_memcpy_args): Remove.
18781 (expand_builtin_memcpy): Call newly added function
18782 expand_builtin_memory_copy_args.
18783 (expand_builtin_memcpy_with_bounds): Likewise.
18784 (expand_builtin_mempcpy): Remove last argument.
18785 (expand_builtin_mempcpy_with_bounds): Likewise.
18786 (expand_builtin_memory_copy_args): New function created from
18787 expand_builtin_mempcpy_args with small modifications.
18788 (expand_builtin_mempcpy_args): Remove.
18789 (expand_builtin_stpcpy): Remove unused argument.
18790 (expand_builtin): Likewise.
18791 (expand_builtin_with_bounds): Likewise.
18793 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18796 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18797 print "ds:" only for immediates in generic address space.
18799 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18802 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18803 (ix86_function_ok_for_sibcall): Return false for naked functions.
18805 2017-08-01 Richard Biener <rguenther@suse.de>
18807 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18808 (compute_antic): Seed worklist with exit block predecessors.
18809 * cfganal.c (dfs_find_deadend): For a cycle return the source
18810 of the edge closing it.
18812 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18814 * config/aarch64/aarch64.c
18815 (aarch64_can_const_movi_rtx_p): Move 0 check.
18817 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18819 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18820 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18822 * match.pd: Ditto in address comparison pattern.
18824 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18826 PR tree-optimization/81627
18827 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18828 closed ssa form for store-store chain.
18830 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18832 PR tree-optimization/81620
18833 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18834 for store-store chain.
18836 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18838 PR tree-optimization/81588
18839 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18840 ranges[i].in_p, invert comparison code ccode. For >/>=,
18841 swap rhs1 and rhs2 and comparison code unconditionally,
18842 for </<= don't do that. Don't swap rhs1/rhs2 again if
18843 ranges[i].in_p, instead invert comparison code ccode if
18844 opcode or oe->rank is BIT_IOR_EXPR.
18847 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18848 a direct optab to conversion optab.
18849 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18850 with GET_MODE_INNER as last argument instead of optab_handler.
18851 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18852 vector extraction if possible and optab is available.
18853 * expr.c (store_constructor): Use convert_optab_handler instead
18854 of optab_handler. Use vector initialization from smaller
18855 vectors if possible and optab is available.
18856 * tree-vect-stmts.c (vectorizable_load): Likewise.
18857 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18858 now have two modes.
18859 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18860 of vec_init from half-sized vectors with the same element mode.
18861 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18862 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18863 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18864 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18865 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18866 after mode in gen_vec_extract* calls.
18867 (vec_extract<mode>): Renamed to ...
18868 (vec_extract<mode><ssescalarmodelower>): ... this.
18869 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18870 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18871 element mode after mode in gen_vec_init* calls.
18872 (VEC_INIT_HALF_MODE): New mode iterator.
18873 (vec_init<mode>): Renamed to ...
18874 (vec_init<mode><ssescalarmodelower>): ... this.
18875 (vec_init<mode><ssehalfvecmodelower>): New expander.
18876 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18877 (vec_extractv2sfsf): ... this.
18878 (vec_initv2sf): Renamed to ...
18879 (vec_initv2sfsf): ... this.
18880 (vec_extractv2si): Renamed to ...
18881 (vec_extractv2sisi): ... this.
18882 (vec_initv2si): Renamed to ...
18883 (vec_initv2sisi): ... this.
18884 (vec_extractv4hi): Renamed to ...
18885 (vec_extractv4hihi): ... this.
18886 (vec_initv4hi): Renamed to ...
18887 (vec_initv4hihi): ... this.
18888 (vec_extractv8qi): Renamed to ...
18889 (vec_extractv8qiqi): ... this.
18890 (vec_initv8qi): Renamed to ...
18891 (vec_initv8qiqi): ... this.
18892 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18893 (vec_init<mode>): Renamed to ...
18894 (vec_init<mode><VEC_base_l>): ... this.
18895 (vec_extract<mode>): Renamed to ...
18896 (vec_extract<mode><VEC_base_l>): ... this.
18897 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18898 (vec_initv2sfsf): ... this.
18899 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18900 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18901 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18902 element mode after mode in gen_vec_init* calls.
18903 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18904 (vec_init<mode><Vel>): ... this.
18905 (vec_extract<mode>): Renamed to ...
18906 (vec_extract<mode><Vel>): ... this.
18907 * config/aarch64/iterators.md (Vel): New mode attribute.
18908 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18909 Add element mode after mode in gen_vec_extract* calls.
18910 * config/s390/vector.md (non_vec_l): New mode attribute.
18911 (vec_extract<mode>): Renamed to ...
18912 (vec_extract<mode><non_vec_l>): ... this.
18913 (vec_init<mode>): Renamed to ...
18914 (vec_init<mode><non_vec_l>): ... this.
18915 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18916 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18918 * config/arm/iterators.md (V_elem_l): New mode attribute.
18919 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18920 (vec_extract<mode><V_elem_l>): ... this.
18921 (vec_extractv2di): Renamed to ...
18922 (vec_extractv2didi): ... this.
18923 (vec_init<mode>): Renamed to ...
18924 (vec_init<mode><V_elem_l>): ... this.
18925 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18926 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18927 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18928 Add element mode after gen_vec_extract* calls.
18929 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18930 (vec_init<mode><unitmode>): ... this.
18931 (vec_extract<mode>): Renamed to ...
18932 (vec_extract<mode><unitmode>): ... this.
18933 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18934 (vec_init<mode><unitmode>): ... this.
18935 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18936 (vec_initv2sfsf): ... this.
18937 (vec_extractv2sf): Renamed to ...
18938 (vec_extractv2sfsf): ... this.
18939 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18940 Add element mode after gen_vec_extract* calls.
18941 * config/mips/mips.md (unitmode): New mode iterator.
18942 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18943 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18944 * config/spu/spu.md (inner_l): New mode attribute.
18945 (vec_init<mode>): Renamed to ...
18946 (vec_init<mode><inner_l>): ... this.
18947 (vec_extract<mode>): Renamed to ...
18948 (vec_extract<mode><inner_l>): ... this.
18949 * config/sparc/sparc.md (veltmode): New mode iterator.
18950 (vec_init<VMALL:mode>): Renamed to ...
18951 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18952 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18953 (vec_initv2sisi): ... this.
18954 (vec_initv2sf): Renamed to ...
18955 (vec_initv2sfsf): ... this.
18956 (vec_extractv2sf): Renamed to ...
18957 (vec_extractv2sfsf): ... this.
18958 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18959 (vec_init<mode>): Renamed to ...
18960 (vec_init<mode><VEC_base_l>): ... this.
18961 (vec_extract<mode>): Renamed to ...
18962 (vec_extract<mode><VEC_base_l>): ... this.
18963 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18964 (vec_initv2sfsf): ... this.
18965 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18966 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18967 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18968 gen_vec_init* calls.
18970 2017-08-01 Richard Biener <rguenther@suse.de>
18972 PR tree-optimization/81297
18973 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18974 TREE_OVERFLOW from INTEGER_CSTs.
18976 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18978 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18980 2017-07-31 Carl Love <cel@us.ibm.com>
18982 * config/rs6000/rs6000-c: Add support for built-in functions
18983 vector signed char vec_xl_be (signed long long, signed char *);
18984 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18985 vector signed int vec_xl_be (signed long long, signed int *);
18986 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18987 vector signed long long vec_xl_be (signed long long, signed long long *);
18988 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18989 vector signed short vec_xl_be (signed long long, signed short *);
18990 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18991 vector double vec_xl_be (signed long long, double *);
18992 vector float vec_xl_be (signed long long, float *);
18993 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18994 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18995 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18997 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18998 (altivec_expand_builtin): Add switch statement to call
18999 altivec_expand_xl_be for each builtin.
19000 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19001 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19002 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19003 __builtin_vsx_le_be_v16qi.
19004 * doc/extend.texi: Update the built-in documentation file for the
19005 new built-in functions.
19007 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19010 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19012 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19014 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19016 * config.gcc: Add z14.
19017 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19018 CPU model numbers for z13s and z14.
19019 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19021 * config/s390/s390-opts.h (enum processor_type): Rename
19022 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19023 * config/s390/s390.c (processor_table): Add field for CPU name to
19024 be passed to Binutils.
19025 (s390_asm_output_machine_for_arch): Use the new field in
19026 processor_table for Binutils.
19027 (s390_expand_builtin): Replace arch12 with z14.
19028 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19029 (s390_get_sched_attrmask): Likewise.
19030 (s390_get_unit_mask): Likewise.
19031 * config/s390/s390.opt: Add z14 to processor_type enum.
19033 2017-07-31 Martin Jambor <mjambor@suse.cz>
19036 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19037 regardless of optimization level.
19039 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19040 Martin Liska <mliska@suse.cz>
19042 * predict.def: Remove old comment and adjust probability.
19043 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19044 PREDICT statements.
19046 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19049 * config/i386/i386.c (ix86_function_naked): New function.
19050 (ix86_can_use_return_insn_p): Return false for naked functions.
19051 (ix86_expand_prologue): Skip prologue for naked functions.
19052 (ix86_expand_epilogue): Skip epilogue for naked functions
19053 and emit trap instruction.
19054 (ix86_warn_func_return): New function.
19055 (ix86_attribute_table): Add "naked" attribute specification.
19056 (TARGET_WARN_FUNC_RETURN): Define.
19057 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19059 2017-07-31 Martin Liska <mliska@suse.cz>
19061 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19062 (dump_gimple_bb_header): Always dump BB info.
19063 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19065 2017-07-31 Martin Liska <mliska@suse.cz>
19068 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19069 also with current_function_decl non-null equality.
19071 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19074 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19075 change type to the element type, instead add eltype variable and
19076 use it where we are interested in the element type.
19078 PR tree-optimization/81603
19079 * ipa-polymorphic-call.c
19080 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19081 offset arithmetic in offset_int, bail out if the resulting bit offset
19082 doesn't fit into shwi.
19084 2017-07-31 Martin Liska <mliska@suse.cz>
19086 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19087 (gimplify_save_expr): Fix comment.
19089 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19092 * config/i386/i386.c (ix86_function_arg): Update arguments for
19094 (ix86_compute_frame_layout): Set the initial stack offset to
19095 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19096 INCOMING_FRAME_SP_OFFSET.
19097 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19098 stack before exception handler returns.
19099 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19100 the 'ERROR_CODE' for exception handler.
19102 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19104 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19105 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19106 (ASM_OUTPUT_REG_POP): Ditto.
19107 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19108 instead of asm_fprintf to output pure string.
19110 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19112 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19113 to imported_module_or_decl hook.
19114 (debug_nothing_tree_tree_tree_bool): Remove.
19115 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19116 * debug.c (do_nothing_debug_hooks): Use
19117 debug_nothing_tree_tree_tree_bool_bool instead of
19118 debug_nothing_tree_tree_tree_bool.
19119 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19120 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19121 * sdbout.c (sdb_debug_hooks): Likewise.
19122 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19123 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19125 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19126 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19127 attribute, don't add anything.
19129 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19131 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19132 (fold_build2_stat_loc): Likewise.
19133 (fold_build3_stat_loc): Likewise.
19134 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19135 (fold_build1_loc): Remove macro.
19136 (fold_build2_loc): Likewise.
19137 (fold_build3_loc): Likewise.
19139 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19141 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19142 (gimple_build_debug_bind_source_stat): Likewise.
19143 * gimple.h (gimple_build_debug_bind): Remove macro.
19144 (gimple_build_debug_bind_source): Likewise.
19146 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19148 * bitmap.c (bitmap_alloc): Adjust.
19149 (bitmap_gc_alloc): Likewise.
19150 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19152 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19154 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19155 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19156 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19157 (bitmap_gc_alloc_stat): Likewise.
19158 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19160 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19162 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19163 * rtl.h (shallow_copy_rtx): Remove macro.
19165 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19167 * emit-rtl.c (gen_raw_REG): Adjust.
19168 * gengenrtl.c (gendef): Likewise.
19169 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19170 * rtl.h (rtx_alloc): Remove macro.
19172 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19174 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19175 (build_tree_list_stat): Likewise.
19176 * tree.h (build_tree_list): Remove macro.
19177 (build_tree_list_vec): Likewise.
19179 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19181 * tree.c (make_vector_stat): Remove _stat from name.
19182 (build_vector_stat): Likewise.
19183 * tree.h (make_vector_stat): Remove macro.
19184 (build_vector_stat): Likewise.
19186 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19188 * tree.h (build_var_debug_value): Remove prototype.
19190 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19192 * tree.c (tree_cons_stat): Remove _stat from name.
19193 * tree.h (tree_cons): Remove macro.
19195 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19197 * tree.c (build_vl_exp_stat): Remove _stat from name.
19198 * tree.h (build_vl_exp): Remove macro.
19200 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19202 * tree.c (build_decl_stat): Remove _stat from name.
19203 * tree.h (build_decl): Remove macro.
19205 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19207 * gimple.c (gimple_build_with_ops_stat): Adjust.
19208 (gimple_alloc_stat): Remove _stat from name.
19209 * gimple.h (gimple_alloc): Remove macro.
19211 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19213 * tree.c (make_tree_vec_stat): Remove _stat from name.
19214 (grow_tree_vec_stat): Likewise.
19215 * tree.h (make_tree_vec_stat): Adjust prototype.
19216 (grow_tree_vec_stat): Likewise.
19217 (make_tree_vec): Remove macro.
19218 (grow_tree_vec): Likewise.
19220 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19222 * fold-const.c (fold_build1_stat_loc): Adjust.
19223 (fold_build2_stat_loc): Likewise.
19224 (fold_build3_stat_loc): Likewise.
19225 * tree.c (build0_stat): Remove _stat from name.
19226 (build1_stat): Likewise.
19227 (build2_stat): Likewise.
19228 (build3_stat): Likewise.
19229 (build4_stat): Likewise.
19230 (build5_stat): Likewise.
19231 * tree.h (build1_loc): Remove macro, and rename _stat function
19233 (build2_loc): Likewise.
19234 (build3_loc): Likewise.
19235 (build4_loc): Likewise.
19236 (build5_loc): Likewise.
19238 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19240 * tree.c (make_int_cst_stat): Remove _stat from name.
19241 * tree.h (make_int_cst_stat): Adjust prototype.
19242 (make_int_cst): Remove macro.
19244 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19246 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19247 * tree.h (make_tree_binfo_stat): Adjust prototype.
19248 (make_tree_binfo): Remove.
19250 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19252 * tree.c (copy_node_stat): Rename to copy_node.
19253 (build_distinct_type_copy): Adjust.
19254 * tree.h (copy_node_stat): Adjust prototype.
19255 (copy_node): Remove macro.
19257 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19259 * tree.c (make_node_stat): rename to make_node.
19260 (build_tree_list_stat): Adjust.
19261 (build0_stat): Likewise.
19262 (build2_stat): Likewise.
19263 (build3_stat): Likewise.
19264 (build4_stat): Likewise.
19265 (build5_stat): Likewise.
19266 (build_decl_stat): Likewise.
19267 * tree.h (make_node_stat): Adjust prototype.
19268 (make_node): remove macro.
19270 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19272 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19273 (PPC_FEATURE2_SCV): Likewise.
19274 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19276 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19278 * config/aarch64/aarch64.c
19279 (aarch64_internal_mov_immediate): Add new special pattern.
19280 * config/aarch64/aarch64.md (*movdi_aarch64):
19281 Add reg/32bit const mov case.
19283 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19284 Richard Sandiford <richard.sandiford@linaro.org>
19286 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19287 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19288 Add integer and movi cases.
19289 (movi-split-hf-df-sf split, fp16): New.
19290 (enabled): Added TARGET_FP_F16INST.
19291 * config/aarch64/iterators.md (GPF_HF): New.
19292 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19294 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19296 * config/aarch64/aarch64.c
19297 (aarch64_simd_container_mode): Add prototype.
19298 (aarch64_expand_mov_immediate): Add HI support.
19299 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19300 (aarch64_can_const_movi_rtx_p): New.
19301 (aarch64_preferred_reload_class):
19302 Remove restrictions of using FP registers for certain SIMD operations.
19303 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19304 (aarch64_valid_floating_const): Add integer move validation.
19305 (aarch64_simd_imm_scalar_p): Remove.
19306 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19307 (aarch64_legitimate_constant_p): Expand list of supported cases.
19308 * config/aarch64/aarch64-protos.h
19309 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19310 (aarch64_reinterpret_float_as_int): New.
19311 (aarch64_simd_imm_scalar_p): Remove.
19312 * config/aarch64/constraints.md (Uvi): New.
19313 (Dd): Split into Ds and new Dd.
19314 * config/aarch64/aarch64.md (*movsi_aarch64):
19316 (*movdi_aarch64): Add SIMD mov case.
19318 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19320 * tree-predcom.c: (struct chain): Handle store-store chain in which
19321 stores for elimination only store loop invariant values.
19322 (execute_pred_commoning_chain): Ditto.
19323 (prepare_initializers_chain_store_elim): Ditto.
19324 (prepare_finalizers): Ditto.
19325 (is_inv_store_elimination_chain): New function.
19326 (initialize_root_vars_store_elim_1): New function.
19328 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19330 * tree-predcom.c: Revise general description of the pass.
19331 (enum chain_type): New enum type for store elimination.
19332 (struct chain): New field supporting store elimination.
19333 (struct component): Ditto.
19334 (dump_chain): Dump store-stores chain.
19335 (release_chain): Release resources.
19336 (split_data_refs_to_components): Compute and create component
19337 contains only stores for elimination.
19338 (get_chain_last_ref_at): New function.
19339 (make_invariant_chain): Initialization.
19340 (make_rooted_chain): Specify chain type in parameter and record it.
19341 (add_looparound_copies): Skip for store-stores chain.
19342 (determine_roots_comp): Compute type of chain and pass it to
19344 (initialize_root_vars_store_elim_2): New function.
19345 (finalize_eliminated_stores): New function.
19346 (remove_stmt): Handle store for elimination.
19347 (execute_pred_commoning_chain): Execute predictive commoning on
19348 store-store chains.
19349 (determine_unroll_factor): Skip unroll for store-stores chain.
19350 (prepare_initializers_chain_store_elim): New function.
19351 (prepare_initializers_chain): Hanlde store-store chain.
19352 (prepare_finalizers_chain, prepare_finalizers): New function.
19353 (tree_predictive_commoning_loop): Return integer value indicating
19354 if loop is unrolled or lcssa form is corrupted.
19355 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19357 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19359 * tree-predcom.c (initialize_root): Delete.
19360 (execute_pred_commoning_chain): Initialize root vars and replace
19361 reference of non-combined chain directly, rather than call above
19364 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19366 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19367 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19369 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19371 * tree-predcom.c (struct chain): New field init_seq.
19372 (release_chain): Release init_seq.
19373 (prepare_initializers_chain): Record intialization stmts in above
19375 (insert_init_seqs): New function.
19376 (tree_predictive_commoning_loop): Call insert_init_seqs.
19378 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19380 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19382 2017-07-28 Richard Biener <rguenther@suse.de>
19384 * match.pd: Remove superfluous :c.
19385 * genmatch.c (simplify::id): Add member.
19386 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19388 (current_id): New global.
19389 (dt_node::parent): Move from ...
19390 (dt_operand::parent): ... here. Add for_id member.
19391 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19392 (decision_tree::find_node): Relax order requirement when
19393 merging DT_TRUE nodes to ones inbetween the current simplify
19394 and the one we try to merge with. Add diagnostic whenever
19395 we need to enforce pattern order by not merging.
19396 (decision_tree::insert): Set current_id.
19397 (decision_tree::print_node): Dump parent node and for_id.
19398 (parser::last_id): Add member.
19399 (parser::push_simplify): Assign unique id.
19400 (parser::parser): Initialize last_id.
19402 2017-07-28 Martin Liska <mliska@suse.cz>
19405 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19406 gimple_build_debug_bind.
19408 2017-07-28 Richard Biener <rguenther@suse.de>
19410 PR tree-optimization/81502
19411 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19413 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19415 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19416 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19417 for BIT_FIELD_REF args.
19418 * fold-const.c (make_bit_field_ref): Likewise.
19419 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19421 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19424 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19425 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19426 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19427 Or it into SANITIZER_UNDEFINED.
19428 * ubsan.c: Include gimple-fold.h and varasm.h.
19429 (ubsan_expand_ptr_ifn): New function.
19430 (instrument_pointer_overflow): New function.
19431 (maybe_instrument_pointer_overflow): New function.
19432 (instrument_object_size): Formatting fix.
19433 (pass_ubsan::execute): Call instrument_pointer_overflow
19434 and maybe_instrument_pointer_overflow.
19435 * internal-fn.c (expand_UBSAN_PTR): New function.
19436 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19437 * sanitizer.def (__ubsan_handle_pointer_overflow,
19438 __ubsan_handle_pointer_overflow_abort): New builtins.
19439 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19440 * internal-fn.def (UBSAN_PTR): New internal function.
19441 * opts.c (sanitizer_opts): Add pointer-overflow.
19442 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19443 * fold-const.c (build_range_check): Compute pointer range check in
19444 integral type if pointer arithmetics would be needed. Formatting
19447 2017-07-28 Martin Liska <mliska@suse.cz>
19450 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19451 parameters that are of a variable-length.
19453 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19455 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19457 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19458 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19459 (CRT_CALL_STATIC_FUNCTION): Likewise.
19460 (ASM_DEFAULT_SPEC): New define.
19461 (ASM_SPEC32): Likewise.
19462 (ASM_SPEC64): Likewise.
19463 (ASM_SPEC_COMMON): Likewise.
19464 (ASM_SPEC): Likewise.
19465 (INVALID_64BIT): Likewise.
19466 (LINK_OS_DEFAULT_SPEC): Likewise.
19467 (LINK_OS_SPEC32): Likewise.
19468 (LINK_OS_SPEC64): Likewise.
19469 (POWERPC_LINUX): Likewise.
19470 (PTRDIFF_TYPE): Likewise.
19471 (RESTORE_FP_PREFIX): Likewise.
19472 (RESTORE_FP_SUFFIX): Likewise.
19473 (SAVE_FP_PREFIX): Likewise.
19474 (SAVE_FP_SUFFIX): Likewise.
19475 (SIZE_TYPE): Likewise.
19476 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19477 (TARGET_64BIT): Likewise.
19478 (TARGET_64BIT): Likewise.
19479 (TARGET_AIX): Likewise.
19480 (WCHAR_TYPE_SIZE): Likewise.
19481 (WCHAR_TYPE): Undefine.
19482 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19483 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19484 (CPP_OS_RTEMS_SPEC): Delete.
19485 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19486 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19488 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19490 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19492 PR tree-optimization/81578
19493 * tree-parloops.c (build_new_reduction): Bail out if
19494 reduction_code isn't one of the standard OpenMP reductions.
19495 Move the details printing after that decision.
19497 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19499 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19500 related to reload_in_progress.
19501 (splat_input_operand): Likewise.
19502 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19504 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19506 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19507 (TARGET_INSTANTIATE_DECLS): Likewise.
19508 (legitimate_indexed_address_p): Delete reload_in_progress code.
19509 (rs6000_debug_legitimate_address_p): Likewise.
19510 (rs6000_eliminate_indexed_memrefs): Likewise.
19511 (rs6000_emit_le_vsx_store): Likewise.
19512 (rs6000_emit_move_si_sf_subreg): Likewise.
19513 (rs6000_emit_move): Likewise.
19514 (register_to_reg_type): Likewise.
19515 (rs6000_pre_atomic_barrier): Likewise.
19516 (rs6000_machopic_legitimize_pic_address): Likewise.
19517 (rs6000_allocate_stack_temp): Likewise.
19518 (rs6000_address_for_fpconvert): Likewise.
19519 (rs6000_address_for_altivec): Likewise.
19520 (rs6000_secondary_memory_needed_rtx): Delete function.
19521 (rs6000_check_sdmode): Likewise.
19522 (rs6000_alloc_sdmode_stack_slot): Likewise.
19523 (rs6000_instantiate_decls): Likewise.
19524 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19525 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19526 Delete reload_in_progress.
19527 (*vec_reload_and_plus_<mptrsize>): Likewise.
19528 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19529 (vsx_div_v2di): Likewise.
19530 (vsx_udiv_v2di): Likewise.
19532 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19534 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19535 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19536 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19537 (rs6000_debug_reg_global): Delete print of LRA status.
19538 (rs6000_option_override_internal): Delete dead LRA related code.
19539 (rs6000_lra_p): Delete function.
19540 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19542 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19544 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19545 * config/riscv/rtems.h: New file.
19547 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19548 Sudakshina Das <sudi.das@arm.com>
19550 * config/aarch64/aarch64.md
19551 (define_split for and<mode>3nr_compare): Move
19552 non aarch64_logical_operand to a register.
19553 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19554 register immediate operand to a register.
19555 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19557 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19559 PR middle-end/81564
19560 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19562 2017-07-27 Richard Biener <rguenther@suse.de>
19564 PR tree-optimization/81573
19565 PR tree-optimization/81494
19566 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19567 multi defuse cycle case.
19569 2017-07-27 Richard Biener <rguenther@suse.de>
19571 PR tree-optimization/81571
19572 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19575 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19577 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19578 earlier and only if MASK_FPU is set. Adjust formatting.
19580 2017-07-27 Martin Liska <mliska@suse.cz>
19582 * opt-functions.awk: Add validation of value of Init.
19583 * optc-gen.awk: Pass new argument.
19585 2017-07-27 Martin Liska <mliska@suse.cz>
19587 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19588 Fix wrong condition.
19590 2017-07-27 Martin Liska <mliska@suse.cz>
19592 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19593 BBs and edges seen by autoFDO.
19595 2017-07-27 Richard Biener <rguenther@suse.de>
19597 PR tree-optimization/81502
19598 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19599 with incompatible but same sized type.
19600 (execute_update_addresses_taken): Likewise.
19602 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19604 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19605 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19607 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19610 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19611 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19612 Change s_operand to memory_operand.
19614 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19616 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19617 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19618 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19619 Emit instructions rather than returning an expression. Handle TFmode
19620 and KFmode by casting to TImode.
19621 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19622 (rs6000_emit_le_vsx_store): Likewise.
19623 * config/rs6000/vsx.md (VSX_TI): New iterator.
19624 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19625 (*vsx_le_undo_permute_<mode>): Likewise.
19626 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19627 emit the split sequence.
19628 (*vsx_le_perm_store_<mode>): Likewise.
19630 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19632 PR tree-optimization/81555
19633 PR tree-optimization/81556
19634 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19635 if true, force CHANGED for the recursive invocation.
19636 (reassociate_bb): Remember original length of ops array, pass
19637 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19639 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19640 attributes for noipa attribute. For naked attribute use
19641 lookup_attribute first before lookup_attribute_spec.
19642 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19644 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19645 for functions with noipa attribute.
19646 (cgraph_externally_visible_p): Return true for functions with noipa
19648 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19649 for functions with noipa attribute.
19650 * doc/extend.texi: Document noipa function attribute.
19651 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19652 also for functions with noipa attribute.
19653 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19655 2017-07-26 Andrew Pinski <apinski@cavium.com>
19657 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19658 vec_unalign_load_cost and vec_unalign_store_cost.
19660 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19662 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19663 -mvsx-small-integer option.
19664 (ISA_3_0_MASKS_IEEE): Likewise.
19665 (OTHER_VSX_VECTOR_MASKS): Likewise.
19666 (POWERPC_MASKS): Likewise.
19667 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19668 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19669 code, only testing for DImode being allowed in non-VSX floating
19671 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19672 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19674 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19675 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19676 TARGET_P8_VECTOR test.
19677 (rs6000_secondary_reload_simple_move): Likewise.
19678 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19679 since TARGET_P9_VECTOR was already tested.
19680 (rs6000_opt_masks): Remove -mvsx-small-integer.
19681 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19682 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19684 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19685 test for TARGET_VEXTRACTUB was used, and that uses
19687 (p9 extract splitter): Likewise.
19688 (vsx_extract_<mode>_di_p9): Likewise.
19689 (vsx_extract_<mode>_store_p9): Likewise.
19690 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19691 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19692 the elimination of TARGET_VSX_SMALL_INTEGER.
19693 (vsx_extract_<mode>_p8): Likewise.
19694 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19695 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19696 (vsx_set_<mode>_p9): Likewise.
19697 (vsx_set_v4sf_p9): Likewise.
19698 (vsx_set_v4sf_p9_zero): Likewise.
19699 (vsx_insert_extract_v4sf_p9): Likewise.
19700 (vsx_insert_extract_v4sf_p9_2): Likewise.
19701 * config/rs6000/rs6000.md (sign extend splitter): Change
19702 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19703 (floatsi<mode>2_lfiwax_mem): Likewise.
19704 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19705 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19706 since a test for TARGET_P9_VECTOR was used.
19707 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19708 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19709 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19710 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19711 TARGET_P8_VECTOR test.
19712 (fix_trunc<mode>si2_stfiwx): Likewise.
19713 (fix_trunc<mode>si2_internal): Likewise.
19714 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19715 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19717 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19718 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19719 TARGET_P8_VECTOR test.
19720 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19721 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19722 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19724 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19725 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19726 since a test for TARGET_P9_VECTOR was used.
19727 (splitter for loading small constants): Likewise.
19729 2017-07-26 Andrew Pinski <apinski@cavium.com>
19731 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19734 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19737 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19738 (fp_valid_at): Likewise.
19740 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19742 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19743 (qdf24xx_addrcost_table): Likewise.
19744 (cortexa57_tunings): Update to use generic_branch_cost.
19745 (cortexa72_tunings): Likewise.
19746 (cortexa73_tunings): Likewise.
19747 (qdf24xx_tunings): Likewise.
19749 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19751 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19752 (thunderx2t99_branch_cost): Likewise.
19753 (cortexa35_tunings): Update to use generic_branch_cost.
19754 (cortexa53_tunings): Likewise.
19755 (cortexa57_tunings): Likewise.
19756 (cortexa72_tunings): Likewise.
19757 (cortexa73_tunings): Likewise.
19758 (thunderx2t99_tunings): Likewise.
19760 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19762 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19763 (sparc_option_override): Honour MASK_FSMULD.
19764 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19765 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19766 * config/sparc/sparc.opt (mfsmuld): New option.
19767 * doc/invoke.texi (mfsmuld): Document option.
19769 2017-07-26 Marek Polacek <polacek@redhat.com>
19771 PR middle-end/70992
19772 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19774 2017-07-26 Richard Biener <rguenther@suse.de>
19776 * gimple-match-head.c (do_valueize): Return OP if valueize
19778 (get_def): New helper to get at the def stmt of a SSA name
19779 if valueize allows.
19780 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19781 do_valueize to get at the def stmt.
19782 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19784 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19786 PR middle-end/46932
19787 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19789 2017-07-26 Martin Liska <mliska@suse.cz>
19792 * function.c (expand_function_start): Make expansion of
19793 nonlocal_goto_save_area after parm_birth_insn.
19795 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19797 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19798 from all CPU target flags enable members.
19800 2017-07-26 Richard Biener <rguenther@suse.de>
19802 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19803 (decision_tree::gen): Make 'type' const.
19804 (write_predicate): Likewise.
19806 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19808 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19809 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19810 (rs6000_option_override_internal): Likewise.
19811 (rs6000_expand_vector_set): Likewise.
19812 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19813 (TARGET_UPPER_REGS_SF): Likewise.
19814 (TARGET_UPPER_REGS_DI): Likewise.
19815 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19816 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19817 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19818 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19819 (Splitters for DI constants in Altivec registers): Likewise.
19820 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19821 (vsx_set_v4sf_p9): Likewise.
19822 (vsx_set_v4sf_p9_zero): Likewise.
19823 (vsx_insert_extract_v4sf_p9): Likewise.
19824 (vsx_insert_extract_v4sf_p9_2): Likewise.
19826 2017-07-25 Carl Love <cel@us.ibm.com>
19828 * doc/extend.texi: Update the built-in documentation file for the
19829 existing built-in functions
19830 vector signed char vec_cnttz (vector signed char);
19831 vector unsigned char vec_cnttz (vector unsigned char);
19832 vector signed short vec_cnttz (vector signed short);
19833 vector unsigned short vec_cnttz (vector unsigned short);
19834 vector signed int vec_cnttz (vector signed int);
19835 vector unsigned int vec_cnttz (vector unsigned int);
19836 vector signed long long vec_cnttz (vector signed long long);
19837 vector unsigned long long vec_cnttz (vector unsigned long long);
19839 2017-07-25 Andrew Pinski <apinski@cavium.com>
19841 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19842 accesses where the use is for the first operand of a BIT_INSERT.
19844 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19847 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19848 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19850 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19852 * config/i386/gstabs.h: Delete.
19853 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19855 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19857 * config/i386/i386.c (ix86_decompose_address): Do not check for
19858 register RTX when looking at index_reg or base_reg.
19859 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19861 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19863 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19864 to update EH info here.
19866 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19868 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19870 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19872 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19874 2017-07-25 Torsten Duwe <duwe@suse.de>
19876 * common.opt: Introduce -fpatchable-function-entry
19877 command line option, and its variables function_entry_patch_area_size
19878 and function_entry_patch_area_start.
19879 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19880 including a two-value parser.
19881 * target.def (print_patchable_function_entry): New target hook.
19882 * targhooks.h (default_print_patchable_function_entry): New function.
19883 * targhooks.c (default_print_patchable_function_entry): Likewise.
19884 * toplev.c (process_options): Switch off IPA-RA if
19885 patchable function entries are being generated.
19886 * varasm.c (assemble_start_function): Look at the
19887 patchable-function-entry command line switch and current
19888 function attributes and maybe generate NOP instructions by
19889 calling the print_patchable_function_entry hook.
19890 * doc/extend.texi: Document patchable_function_entry attribute.
19891 * doc/invoke.texi: Document -fpatchable_function_entry
19892 command line option.
19893 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19895 * doc/tm.texi: Re-generate.
19897 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19900 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19901 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19903 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19905 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19907 2017-07-25 Richard Biener <rguenther@suse.de>
19909 PR tree-optimization/81455
19910 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19911 not walk in cycles when looking for guards.
19913 2017-07-25 Richard Biener <rguenther@suse.de>
19915 PR tree-optimization/81529
19916 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19917 when optimizing backedge uses.
19919 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19921 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19923 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19924 to dl_section_ref. On AIX, append an expression to subtract
19925 the size of the section length to dl_section_ref.
19927 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19929 * configure.ac: If any of the config.* scripts fail, exit 1.
19930 * configure: Regenerate.
19932 2017-07-25 Richard Biener <rguenther@suse.de>
19934 PR middle-end/81546
19935 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19936 of immediate uses, be more verbose on errors.
19938 2017-07-25 Richard Biener <rguenther@suse.de>
19940 PR tree-optimization/81510
19941 * tree-vect-loop.c (vect_is_simple_reduction): When the
19942 reduction stmt is not inside the loop bail out.
19944 2017-07-25 Richard Biener <rguenther@suse.de>
19946 PR tree-optimization/81303
19947 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19948 profitability check against LOOP_VINFO_NITERSM1.
19950 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19952 * domwalk.c (cmp_bb_postorder): Simplify.
19953 (sort_bbs_postorder): New function. Use it...
19954 (dom_walker::walk): ...here to optimize common cases.
19956 2017-07-25 Martin Liska <mliska@suse.cz>
19959 * ipa-visibility.c (function_and_variable_visibility): Make the
19960 redirection just on target that supports aliasing.
19961 Fix GNU coding style.
19963 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19966 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19968 * config/arm/rtems.h: Likewise.
19969 * config/bfin/rtems.h: Likewise.
19970 * config/i386/rtemself.h: Likewise.
19971 * config/lm32/rtems.h: Likewise.
19972 * config/m32c/rtems.h: Likewise.
19973 * config/m68k/rtemself.h: Likewise.
19974 * config/microblaze/rtems.h: Likewise.
19975 * config/mips/rtems.h: Likewise.
19976 * config/moxie/rtems.h: Likewise.
19977 * config/nios2/rtems.h: Likewise.
19978 * config/powerpcspe/rtems.h: Likewise.
19979 * config/rs6000/rtems.h: Likewise.
19980 * config/rtems.h: Likewise.
19981 * config/sh/rtems.h: Likewise.
19982 * config/sh/rtemself.h: Likewise.
19983 * config/sparc/rtemself.h: Likewise.
19985 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19988 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19989 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19990 * tree-ssa-structalias.c (alias_get_name): Same.
19992 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19995 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19996 instructions if no du chain is found.
19998 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20000 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20002 2017-07-25 Richard Biener <rguenther@suse.de>
20004 PR middle-end/81505
20005 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20008 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20010 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20011 upper-regs options.
20012 (ISA_2_7_MASKS_SERVER): Likewise.
20013 (ISA_3_0_MASKS_IEEE): Likewise.
20014 (OTHER_P8_VECTOR_MASKS): Likewise.
20015 (OTHER_VSX_VECTOR_MASKS): Likewise.
20016 (POWERPC_MASKS): Likewise.
20017 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20018 duplicate list of options.
20019 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20020 explicit -mupper-regs options.
20021 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20022 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20023 alias for -mupper-regs-df.
20024 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20025 (rs6000_init_hard_regno_mode_ok): Likewise.
20026 (rs6000_option_override_internal): Likewise.
20027 (rs6000_opt_masks): Likewise.
20028 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20029 options in terms of whether -mvsx or -mpower8-vector was used.
20030 (TARGET_UPPER_REGS_DI): Likewise.
20031 (TARGET_UPPER_REGS_SF): Likewise.
20032 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20033 -mupper-regs-* options.
20035 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20037 * passes.c (emergency_dump_function): Print some empty lines and a
20038 header before the RTL dump.
20040 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20042 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20044 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20047 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20048 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20050 2017-07-24 Carl Love <cel@us.ibm.com>
20052 * config/rs6000/rs6000-c.c: Add support for built-in functions
20053 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20054 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20055 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20056 vec_extract_fp_from_shortl): Add defines for the two builtins.
20057 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20058 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20060 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20061 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20062 * doc/extend.texi: Update the built-in documentation file for the
20063 new built-in function.
20065 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20068 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20070 * doc/generic.texi: Likewise.
20071 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20072 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20074 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20076 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20077 (aarch64_mls_elt_merge<mode>): Likewise.
20079 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20081 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20082 having __cxa_atexit.
20084 2017-07-23 Michael Collison <michael.collison@arm.com>
20086 * config/arm/arm.c (arm_option_override): Deprecate
20087 use of -mstructure-size-boundary.
20088 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20089 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20091 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20094 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20095 Reduce cost estimate for direct moves.
20097 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20100 * config/i386/i386.c (ix86_option_override_internal): Disable
20101 BMI, BMI2 and TBM instructions for -m16.
20103 2017-07-21 Carl Love <cel@us.ibm.com>
20105 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20106 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20107 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20108 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20109 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20110 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20111 VMULOSW): New enum "unspec" values.
20112 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20113 altivec_vmulosw): New patterns.
20114 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20115 VMULOSW): Add definitions.
20117 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20119 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20120 (qdf24xx): Likewise.
20121 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20122 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20123 (AARCH64_FL_V8_1): Renumber.
20124 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20125 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20126 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20127 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20128 rdma to feature modifiers list.
20130 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20132 PR middle-end/56727
20133 * ipa-visibility (function_and_variable_visibility): Convert
20134 recursive PLT call to direct call if appropriate.
20136 2017-07-21 Andrew Pinski <apinski@cavium.com>
20138 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20139 operand 1 to see if the types precision matches.
20140 * fold-const.c (operand_equal_p): Likewise.
20142 2017-07-21 Richard Biener <rguenther@suse.de>
20144 PR tree-optimization/81303
20145 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20146 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20147 (vect_peeling_hash_get_lowest_cost): Adjust.
20148 (vect_enhance_data_refs_alignment): Likewise. Use
20149 vect_get_peeling_costs_all_drs to compute the penalty for no
20150 peeling to match up costs.
20152 2017-07-21 Richard Biener <rguenther@suse.de>
20154 PR tree-optimization/81500
20155 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20156 we didn't identify a reduction path.
20158 2017-07-21 Tom de Vries <tom@codesourcery.com>
20159 Cesar Philippidis <cesar@codesourcery.com>
20161 PR gcov-profile/81442
20162 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20165 2017-07-21 Tom de Vries <tom@codesourcery.com>
20168 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20170 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20171 nvptx_override_options_after_change.
20173 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20175 * dwarf2out.c (output_file_names): Avoid double testing for
20176 dwarf_version >= 5.
20178 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20180 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20182 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20184 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20186 (try_crossjump_to_edge): Do not punt on partitioned functions.
20188 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20190 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20191 Put all BBs reachable only via paths crossing cold region to cold
20193 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20195 2016-07-21 Richard Biener <rguenther@suse.de>
20197 PR tree-optimization/81303
20198 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20199 into account prologue and epilogue iterations when raising
20200 min_profitable_iters to sth at least covering one vector iteration.
20202 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20204 * config/arm/arm.c (arm_test_cpu_arch_dat):
20207 2017-07-20 Nathan Sidwell <nathan@acm.org>
20209 Remove TYPE_METHODS.
20210 * tree.h (TYPE_METHODS): Delete.
20211 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20212 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20213 (dbxout_type_methods): Scan TYPE_FIELDS.
20214 (dbxout_type): Don't check TYPE_METHODS here.
20215 * function.c (use_register_for_decl): Always ignore register for
20216 class types when not optimizing.
20217 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20218 * tree.c (free_lang_data_in_type): Stitch out member functions and
20219 templates from TYPE_FIELDS.
20220 (build_distinct_type_copy, verify_type_variant,
20221 verify_type): Member fns are on TYPE_FIELDS.
20222 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20223 * tree-pretty-print.c (dump_generic_node): Likewise.
20225 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20228 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20229 V2TImode and V4TImode.
20230 (ix86_expand_vector_extract): Likewise.
20231 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20232 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20233 (ssescalarmode): Handle V4TImode and V2TImode.
20234 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20235 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20236 (VEXTRACTI128_MODE): New mode iterator.
20237 (splitter for *vec_extractv?ti first element): New.
20238 (VEC_INIT_MODE): New mode iterator.
20239 (vec_init<mode>): Consolidate 3 expanders into one using
20240 VEC_INIT_MODE mode iterator.
20242 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20244 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20245 non_spilled_static_chain_regno_p.
20247 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20249 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20251 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20253 * bb-reorder.c (connect_traces): Allow copying of blocks within
20256 2017-07-20 Richard Biener <rguenther@suse.de>
20258 * gimple.h (gimple_phi_result): Add gphi * overload.
20259 (gimple_phi_result_ptr): Likewise.
20260 (gimple_phi_arg): Likewise. Adjust index assert to only
20261 allow actual argument accesses rather than all slots available
20263 (gimple_phi_arg_def): Add gphi * overload.
20264 * tree-phinodes.c (make_phi_node): Initialize only actual
20266 (resize_phi_node): Clear memory not covered by old node,
20267 do not initialize excess argument slots.
20268 (reserve_phi_args_for_new_edge): Initialize new argument slot
20271 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20273 PR tree-optimization/81388
20275 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20277 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20278 by removing computation of may_be_zero.
20280 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20281 Tom de Vries <tom@codesourcery.com>
20283 PR middle-end/81030
20284 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20285 when gimple level profile disagrees with what RTL expander did.
20287 2017-07-20 Richard Biener <rguenther@suse.de>
20289 PR tree-optimization/61171
20290 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20291 (vect_analyze_stmt): Add slp instance parameter.
20292 (vectorizable_reduction): Likewise.
20293 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20294 (vect_is_simple_reduction): Deal with chains not detected
20295 as SLP reduction chain, specifically not properly associated
20296 chains containing a mix of plus/minus.
20297 (get_reduction_op): Remove.
20298 (get_initial_defs_for_reduction): Simplify, pass in whether
20299 this is a reduction chain, pass in the SLP node for the PHIs.
20300 (vect_create_epilog_for_reduction): Get the SLP instance as
20302 (vectorizable_reduction): Get the SLP instance as arg.
20303 During analysis remember the SLP node with the PHIs in the
20304 instance. Simplify getting at the vectorized reduction PHIs.
20305 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20306 through SLP instance.
20307 (vect_slp_analyze_operations): Likewise.
20308 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20309 (vect_transform_stmt): Likewise.
20311 2017-07-20 Tom de Vries <tom@codesourcery.com>
20313 PR tree-optimization/81489
20314 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20315 read of phi arg location to before loop that modifies phi.
20317 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20319 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20322 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20324 PR middle-end/81331
20325 * except.c (execute): Fix ordering issue.
20327 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20329 PR rtl-optimization/81423
20330 * combine.c (make_compound_operation_int): Don't try to optimize
20331 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20333 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20335 PR rtl-optimization/81423
20336 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20337 with a constant that is -1 in the truncated to mode.
20339 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20341 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20342 (determine_unlikely_bbs): ... here.
20343 * predict.h (propagate_unlikely_bbs_forward): Declare.
20344 * cfgexpand.c (pass_expand::execute): Use it.
20345 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20347 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20348 propagate_unlikely_bbs_forward.
20350 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20352 PR middle-end/81331
20353 * except.c (maybe_add_nop_after_section_switch): New function.
20356 2017-07-19 Tom de Vries <tom@codesourcery.com>
20358 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20360 2017-07-19 Tom de Vries <tom@codesourcery.com>
20362 * gimple.h (gimple_phi_arg): Make assert more strict.
20364 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20366 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20367 * config/rs6000/mmintrin.h: New file.
20368 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20370 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20372 PR tree-optimization/81346
20373 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20375 2017-07-19 Tom de Vries <tom@codesourcery.com>
20377 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20379 2017-07-19 Tom de Vries <tom@codesourcery.com>
20381 * config/nvptx/nvptx-modes.def: Add V2DImode.
20382 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20383 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20384 (nvptx_output_mov_insn): Handle lack of mov.b128.
20385 (nvptx_print_operand): Handle 'H' and 'L' codes.
20386 (nvptx_vector_mode_supported): Allow V2DImode.
20387 (nvptx_preferred_simd_mode): New function.
20388 (nvptx_data_alignment): New function.
20389 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20390 nvptx_preferred_simd_mode.
20391 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20393 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20395 2017-07-19 Tom de Vries <tom@codesourcery.com>
20397 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20398 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20399 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20400 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20401 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20402 (mov<VECIM>_insn): New define_insn.
20403 (define_expand "mov<VECIM>): New define_expand.
20405 2017-07-19 Tom de Vries <tom@codesourcery.com>
20407 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20409 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20411 PR tree-optimization/81346
20412 * fold-const.h (fold_div_compare, range_check_type): Declare.
20413 * fold-const.c (range_check_type): New function.
20414 (build_range_check): Use range_check_type.
20415 (fold_div_compare): No longer static, rewritten into
20416 a match.pd helper function.
20417 (fold_comparison): Don't call fold_div_compare here.
20418 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20419 as helper function.
20421 2017-07-19 Nathan Sidwell <nathan@acm.org>
20423 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20424 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20425 * tree.c (find_decls_types_r, verify_type): Use
20426 TYPE_{MIN,MAX}_VALUE_RAW.
20427 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20428 (hash_tree): Likewise.
20429 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20431 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20434 2017-07-18 Tom de Vries <tom@codesourcery.com>
20436 PR middle-end/81464
20437 * omp-expand.c (expand_omp_for_static_chunk): Handle
20438 equal-argument loop exit phi.
20440 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20443 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20444 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20445 operand 2 predicate.
20446 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20447 operand 2 predicate.
20448 (ror,rol -> rorx splitters): Use const_int_operand as
20449 operand 2 predicate.
20451 2017-06-18 Richard Biener <rguenther@suse.de>
20453 PR tree-optimization/81410
20454 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20455 the gap in the ! slp_perm SLP case after each group.
20457 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20459 PR middle-end/81463
20460 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20463 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20465 PR middle-end/81462
20466 * predict.c (set_even_probabilities): Cleanup; do not affect
20467 probabilities that are already known.
20468 (combine_predictions_for_bb): Call even when count is set.
20470 2017-07-18 Nathan Sidwell <nathan@acm.org>
20472 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20475 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20478 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20479 optimization for loop niter analysis.
20481 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20484 * config/avr/avr.c (avr_optimize_casesi): Don't use
20485 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20487 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20489 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20490 body_cost_vec from _vect_peel_extended_info.
20491 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20492 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20495 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20497 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20499 2017-07-18 Richard Biener <rguenther@suse.de>
20501 PR tree-optimization/80620
20502 PR tree-optimization/81403
20503 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20504 info when re-using a VN table entry.
20506 2017-07-18 Richard Biener <rguenther@suse.de>
20508 PR tree-optimization/81418
20509 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20510 vectype_in. Verify that with lane-reducing reduction operations
20511 we have a single def-use cycle.
20513 2017-07-17 Carl Love <cel@us.ibm.com>
20515 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20517 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20518 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20519 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20520 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20521 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20522 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20523 VMULOSW): New enum "unspec" values.
20524 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20525 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20526 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20527 altivec_vmulosw): New patterns.
20528 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20529 VMULOSW): Add definitions.
20531 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20533 * config/alpha/alpha.c: Include predict.h.
20535 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20537 * tree-vrp.c (compare_assert_loc): Fix comparison function
20538 to return predictable results.
20540 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20542 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20544 (subdi3): Likewise.
20545 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20546 * doc/invoke.texi (mexpand-adddi): Update text.
20548 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20550 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20551 that also clobbers the CC register. The old expand code is moved
20553 (*arc_clzsi2): ... here.
20554 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20555 the CC register. The old expand code is moved to ...
20556 (arc_ctzsi2): ... here.
20558 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20560 * config/arc/arc.opt (mindexed-loads): Use initial value
20561 TARGET_INDEXED_LOADS_DEFAULT.
20562 (mauto-modify-reg): Use initial value
20563 TARGET_AUTO_MODIFY_REG_DEFAULT.
20564 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20565 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20566 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20567 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20569 2017-07-17 Martin Liska <mliska@suse.cz>
20572 * opts.c (finish_options): Do not allow -fgnu-tm
20573 w/ -fsanitize={kernel-,}address. Say sorry.
20575 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20578 * tree-loop-distribution.c (classify_partition): Only assert on
20579 numer of iterations.
20580 (merge_dep_scc_partitions): Delete prameter. Update function call.
20581 (distribute_loop): Remove code handling loop with unknown niters.
20582 (pass_loop_distribution::execute): Skip loop with unknown niters.
20584 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20587 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20588 function sort_partitions_by_post_order.
20590 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20592 PR tree-optimization/81374
20593 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20594 the max index of basic blocks, rather than number of basic blocks.
20596 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20598 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20600 (arc_legitimate_pic_operand_p): Likewise.
20601 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20603 (arc_needs_pcl_p): Likewise.
20604 (arc_legitimate_pc_offset_p): Likewise.
20605 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20606 function is also used in constrains.md.
20607 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20608 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20609 PLUS. Only return true/false in known cases, otherwise assert.
20610 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20611 is already called in arc_legitimate_constant_p.
20612 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20614 (LEGITIMATE_PIC_OPERAND_P): Use
20615 arc_raw_symbolic_reference_mentioned_p function.
20616 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20621 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20622 Andrew Burgess <andrew.burgess@embecosm.com>
20624 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20625 (arc_return_address_register): New function.
20626 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20627 (arc_handle_fndecl_attribute): Add naked attribute.
20628 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20629 (TARGET_WARN_FUNC_RETURN): Likewise.
20630 (arc_allocate_stack_slots_for_args): New function.
20631 (arc_warn_func_return): Likewise.
20632 (machine_function): Change type fn_type.
20633 (arc_compute_function_type): Consider new naked function type,
20634 change function return type.
20635 (arc_must_save_register): Adapt to handle new
20636 arc_compute_function_type's return type.
20637 (arc_expand_prologue): Likewise.
20638 (arc_expand_epilogue): Likewise.
20639 (arc_return_address_regs): Delete.
20640 (arc_return_address_register): New function.
20641 (arc_epilogue_uses): Use above function.
20642 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20643 (arc_function_type): Change encoding, add naked type.
20644 (ARC_INTERRUPT_P): Change to handle the new encoding.
20645 (ARC_FAST_INTERRUPT_P): Likewise.
20646 (ARC_NORMAL_P): Define.
20647 (ARC_NAKED_P): Likewise.
20648 (arc_compute_function_type): Delete prototype.
20649 * config/arc/arc.md (in_ret_delay_slot): Use
20650 arc_return_address_register function.
20651 (simple_return): Likewise.
20652 (p_return_i): Likewise.
20654 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20656 PR tree-optimization/81428
20657 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20658 can't be built for those types.
20660 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20662 Remove stuff dead since r239246.
20664 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20665 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20666 (avr_inform_devices): Remove dead stuff.
20668 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20670 * config/arm/arm_neon.h: Fix softp typo.
20672 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20674 PR tree-optimization/81365
20675 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20676 aggregate moves onto bb predecessor edges, make sure there are no
20677 loads that could alias the lhs in between the start of bb and the
20680 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20683 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20684 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20685 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20687 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20689 PR tree-optimization/81396
20690 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20691 (init_symbolic_number): Initialize it to 1.
20692 (perform_symbolic_merge): Add n_ops from both operands into the new
20694 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20695 without base_addr as useless if they need more than one operation.
20696 (bswap_replace): Handle !bswap case for NULL base_addr.
20698 2017-07-17 Tom de Vries <tom@codesourcery.com>
20701 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20704 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20706 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20707 conditional builtin define __FIX_LEON3FT_B2BST.
20709 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20711 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20712 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20715 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20717 PR rtl-optimization/81424
20718 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20719 to remove potential trapping from operands if -fnon-call-exceptions.
20721 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20723 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20724 profile_proability for scalling.
20725 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20727 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20729 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20731 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20733 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20734 fixpoint arithmetics.
20736 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20738 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20739 fixpoint arithmetics.
20741 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20743 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20744 fixpoint arithmetics.
20746 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20748 * profile-count.h (profile_probability::from_reg_br_prob_note,
20749 profile_probability::to_reg_br_prob_note): New functions.
20750 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20751 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20752 * predict.c (probability_reliable_p): Update.
20753 (edge_probability_reliable_p): Update.
20754 (br_prob_note_reliable_p): Update.
20755 (invert_br_probabilities): Update.
20756 (add_reg_br_prob_note): New function.
20757 (combine_predictions_for_insn): Update.
20758 * asan.c (asan_clear_shadow): Update.
20759 * cfgbuild.c (compute_outgoing_frequencies): Update.
20760 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20761 (update_br_prob_note): Update.
20762 (rtl_verify_edges): Update.
20763 (purge_dead_edges): Update.
20764 (fixup_reorder_chain): Update.
20765 * emit-rtl.c (try_split): Update.
20766 * ifcvt.c (cond_exec_process_insns): Update.
20767 (cond_exec_process_if_block): Update.
20768 (dead_or_predicable): Update.
20769 * internal-fn.c (expand_addsub_overflow): Update.
20770 (expand_neg_overflow): Update.
20771 (expand_mul_overflow): Update.
20772 * loop-doloop.c (doloop_modify): Update.
20773 * loop-unroll.c (compare_and_jump_seq): Update.
20774 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20775 * predict.h: Update.
20776 * reorg.c (mostly_true_jump): Update.
20778 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20779 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20780 * config/arc/arc.c: (emit_unlikely_jump): Update.
20781 * config/arm/arm.c: (emit_unlikely_jump): Update.
20782 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20783 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20784 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20785 (ix86_print_operand): Update.
20786 (ix86_split_fp_branch): Update.
20787 (predict_jump): Update.
20788 * config/ia64/ia64.c (ia64_print_operand): Update.
20789 * config/mmix/mmix.c (mmix_print_operand): Update.
20790 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20791 (rs6000_expand_split_stack_prologue): Update.
20792 * config/rs6000/rs6000.c: Update.
20793 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20794 (s390_expand_vec_movstr): Update.
20795 (s390_expand_cs_tdsi): Update.
20796 (s390_expand_split_stack_prologue): Update.
20797 * config/sh/sh.c (sh_print_operand): Update.
20798 (expand_cbranchsi4): Update.
20799 (expand_cbranchdi4): Update.
20800 * config/sparc/sparc.c (output_v9branch): Update.
20801 * config/spu/spu.c (get_branch_target): Update.
20802 (ea_load_store_inline): Update.
20803 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20804 * config/tilepro/tilepro.c: Update.
20806 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20808 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20809 (gimplify_save_expr): Likewise.
20811 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20813 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20815 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20817 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20819 * timevar.def (TV_IPA_FNSUMMARY): Define.
20821 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20823 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20824 to back store errata sensitive sequence from being generated.
20825 (sqrtdf2_fix): Likewise.
20827 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20829 * tree-ssa-threadupdate.c (compute_path_counts,
20830 update_joiner_offpath_counts): Use profile_probability.
20832 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20835 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20837 * config/arm/arm-c.c (arm_cpu_builtins): Define
20838 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20840 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20842 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20843 array entries to represent __ieee128 versions of the
20844 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20845 scalar_extract_sig, and scalar_insert_exp built-in functions.
20846 (altivec_resolve_overloaded_builtin): Add special case handling
20847 for the __builtin_scalar_insert_exp function, as represented by
20848 the P9V_BUILTIN_VEC_VSIEDP constant.
20849 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20850 exponent support for __ieee128 argument.
20851 (VSESQP): Add scalar extract signature support for __ieee128
20853 (VSTDCNQP): Add scalar test negative support for __ieee128
20855 (VSIEQP): Add scalar insert exponent support for __int128 argument
20856 with __ieee128 result.
20857 (VSIEQPF): Add scalar insert exponent support for __ieee128
20858 argument with __ieee128 result.
20859 (VSTDCQP): Add scalar test data class support for __ieee128
20861 (VSTDCNQP): Add overload support for scalar test negative with
20862 __ieee128 argument.
20863 (VSTDCQP): Add overload support for scalar test data class
20864 __ieee128 argument.
20865 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20866 UNSPEC_VSX_SXSIGDP.
20867 (UNSPEC_VSX_SIEXPQP): New constant.
20868 (xsxexpqp): New insn for VSX scalar extract exponent quad
20870 (xsxsigqp): New insn for VSX scalar extract significand quad
20872 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20873 precision with floating point argument.
20874 (xststdcqp): New expand for VSX scalar test data class quad
20876 (xststdcnegqp): New expand for VSX scalar test negative quad
20878 (xststdcqp): New insn to match expansions for VSX scalar test data
20879 class quad precision and VSX scalar test negative quad precision.
20880 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20881 special case operand checking to enforce that second operand of
20882 VSX scalar test data class with quad precision argument is a 7-bit
20884 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20885 prototypes and descriptions of __ieee128 versions of
20886 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20887 scalar_test_data_class, and scalar_test_neg built-in functions.
20889 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20891 PR tree-optimization/81162
20892 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20893 replace a negate with an add.
20895 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20897 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20899 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20901 * config/arm/arm-c.c (arm_cpu_builtins): Define
20902 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20904 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20906 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20907 (armv8-r): Set ARM Cortex-R52 as default CPU.
20908 * config/arm/arm-tables.opt: Regenerate.
20909 * config/arm/arm-tune.md: Regenerate.
20910 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20912 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20913 extension for -mcpu=cortex-r52.
20915 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20917 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20918 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20919 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20920 (fp-armv8): Define it as FP_ARMv8 only.
20921 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20922 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20924 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20925 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20926 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20927 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20928 than TARGET_FPU_ARMV8.
20929 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20930 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20931 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20933 * config/arm/neon.md (neon_vrint): Likewise.
20934 (neon_vcvt): Likewise.
20935 (neon_<fmaxmin_op><mode>): Likewise.
20936 (<fmaxmin><mode>3): Likewise.
20937 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20938 * config/arm/predicates.md (arm_cond_move_operator): Check against
20939 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20941 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20943 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20944 to top of function.
20946 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20948 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20949 loop in comment with memset.
20951 2017-07-14 Martin Liska <mliska@suse.cz>
20953 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20954 * dwarf2out.c (is_java): Remove the function.
20955 (output_pubname): Remove usage of the function.
20956 (lower_bound_default): Remove usage of DW_LANG_Java.
20957 (gen_compile_unit_die): Likewise.
20958 * gcc.c: Remove compiler defaults for .java and .zip files.
20959 * gimple-expr.c (remove_suffix): Change as there's no longer
20960 extension than 4-letter one.
20961 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20962 (gimplify_save_expr): Likewise.
20963 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20964 as it's possible even for other languages than Java.
20965 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20966 * lto-opts.c (lto_write_options): Remove reference to Java.
20967 * opts.c (strip_off_ending): Update file extension handling.
20968 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20969 * tree-eh.c (lower_resx): Likewise.
20970 * tree.c (free_lang_data_in_type): Remove dead code.
20971 (find_decls_types_r): Likewise.
20972 (build_common_builtin_nodes): Remove Java from a comment.
20973 (verify_type): Remove dead code.
20974 * varasm.c (assemble_external): Remove Java from a comment.
20976 2017-07-14 Martin Liska <mliska@suse.cz>
20978 * opts.c (finish_options): Add quotes.
20979 (common_handle_option): Likewise.
20981 2017-07-14 Martin Liska <mliska@suse.cz>
20983 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20984 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20985 Remove N_SO_PASCAL.
20986 * dwarf2out.c (lower_bound_default): Do not handle
20988 (gen_compile_unit_die): Likewise.
20989 * gcc.c: Remove default extension binding for GNU Pascal.
20990 * stmt.c: Remove Pascal language from a comment.
20991 * xcoffout.c: Likewise.
20993 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20996 * diagnostic-show-locus.c (fixit_cmp): New function.
20997 (layout::layout): Sort m_fixit_hints.
20998 (column_range::column_range): Assert that the values are valid.
20999 (struct char_span): New struct.
21000 (correction::overwrite): New method.
21001 (struct source_line): New struct.
21002 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21003 calls in terms of classes source_line and char_span, and
21004 correction::overwrite.
21005 (selftest::test_overlapped_fixit_printing_2): New function.
21006 (selftest::diagnostic_show_locus_c_tests): Call it.
21008 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21010 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21011 early if there is no lhs.
21013 2017-07-13 Martin Liska <mliska@suse.cz>
21015 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21016 (gen_reference_type_die): Likewise.
21017 * stor-layout.c: Remove Pascal-related comment.
21019 2017-07-13 Martin Liska <mliska@suse.cz>
21021 * opts.c (finish_options): Add quotes to error messages.
21022 (parse_sanitizer_options): Likewise.
21024 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21026 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21028 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21030 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21032 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21034 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21036 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21038 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21041 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21042 provides the hardware capability bits, define the macro
21043 __BUILTIN_CPU_SUPPORTS__.
21044 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21045 if GLIBC does not provide the hardware capability bits. Add a
21046 gcc_unreachable call if the built-in cpu function is neither
21047 __builtin_cpu_is nor __builtin_cpu_supports.
21048 (rs6000_get_function_versions_dispatcher): Change the warning
21049 that an old GLIBC is used which does not export the capability
21050 bits to be an error.
21051 * doc/extend.texi (target_clones attribute): Document the
21052 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21053 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21054 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21055 the macros defined by GCC if the newer GLIBC is available.
21057 2017-07-12 Jeff Law <law@redhat.com>
21059 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21060 remaining includes slightly.
21061 * config/riscv/riscv-builtins.c: Include profile-count.h.
21063 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21066 * config/avr/avr.c (avr_set_current_function): In diagnostic
21067 messages: Quote keywords and (parts of) identifiers.
21068 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21071 2017-07-12 Carl Love <cel@us.ibm.com>
21073 * config/rs6000/rs6000-c.c: Add support for built-in functions
21074 vector bool char vec_revb (vector bool char);
21075 vector bool short vec_revb (vector short char);
21076 vector bool int vec_revb (vector bool int);
21077 vector bool long long vec_revb (vector bool long long);
21078 * doc/extend.texi: Update the built-in documentation file for the
21079 new built-in functions.
21081 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21083 * config/s390/s390.md: Remove movcc splitter.
21085 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21087 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21088 load/store on condition.
21090 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21093 * config/avr/avr.c (avr_encode_section_info)
21094 [progmem && !TREE_READONLY]: Error if progmem object needs
21097 2017-07-11 Michael Collison <michael.collison@arm.com>
21099 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21102 2017-07-11 Carl Love <cel@us.ibm.com>
21104 * config/rs6000/rs6000-c.c: Add support for builtins
21105 vector unsigned int vec_parity_lsbb (vector signed int);
21106 vector unsigned int vec_parity_lsbb (vector unsigned int);
21107 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21108 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21109 vector unsigned long long vec_parity_lsbb (vector signed long long);
21110 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21111 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21112 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21113 * doc/extend.texi: Update the built-in documentation file for the
21114 new built-in functions.
21116 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21118 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21119 (layout::m_primary_loc): New field.
21120 (layout::layout): Initialize new field. Move location filtering
21121 logic from here to...
21122 (layout::maybe_add_location_range): ...this new method. Add
21123 support for filtering to just the lines already specified by other
21125 (layout::will_show_line_p): New method.
21126 (gcc_rich_location::add_location_if_nearby): New method.
21127 (selftest::test_add_location_if_nearby): New test function.
21128 (selftest::diagnostic_show_locus_c_tests): Call it.
21129 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21132 2017-07-11 Tom de Vries <tom@codesourcery.com>
21134 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21135 (bb_first_real_insn): New function.
21136 (nvptx_single): Add extra initialization of broadcasted condition
21139 2017-07-11 Nathan Sidwell <nathan@acm.org>
21141 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21143 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21145 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21146 Binutils doc as TEXI will mess them up.
21147 * doc/invoke.texi (AVR Options): Same here.
21149 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21151 * config/sparc/sparc.opt (mfix-ut700): New option.
21152 (mfix-gr712rc): Likewise.
21153 (sparc_fix_b2bst): New variable.
21154 * doc/invoke.texi (SPARC options): Document them.
21155 (ARM options): Fix warnings.
21156 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21157 instructions to prevent sequences that can trigger the store-store
21158 errata for certain LEON3FT processors.
21159 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21160 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21161 * config/sparc/sparc.md (fix_b2bst): New attribute.
21162 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21164 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21167 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21169 (*rsqrtsf2_sse): Ditto.
21171 (div<mode>3): Macroize insn from divdf3 and divsf3
21172 using MODEF mode iterator.
21174 2017-07-10 Martin Sebor <msebor@redhat.com>
21176 PR tree-optimization/80397
21177 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21178 instead of testing for equality to INTEGER_TYPE.
21180 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21182 * config.gcc: Remove uclibc from arc target spec.
21184 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21186 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21188 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21191 * lto-wrapper.c (remove_option): New function.
21192 (merge_and_complain): Merge PIC/PIE options more realistically.
21194 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21196 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21200 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21201 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21202 * config.in: Regenerate.
21203 * configure: Regenerate.
21204 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21205 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21206 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21207 (TARGET_GASISR_PROLOGUES): ...target mask.
21208 * common/config/avr/avr-common.c
21209 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21210 Set -mgas-isr-prologues.
21211 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21212 INSERT_PASS_BEFORE for it.
21213 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21214 * config/avr/avr.c (avr_option_override)
21215 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21216 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21217 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21218 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21219 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21221 (make_avr_pass_pre_proep): New function.
21222 (emit_push_sfr) <treg>: Add argument to function and use it
21223 instead of TMP_REG.
21224 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21225 and set machine->gasisr.yes.
21226 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21227 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21228 __gcc_isr.n_pushed to .L__stack_usage.
21229 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21230 (avr_asm_final_postscan_insn): ...this new static function.
21231 * config/avr/avr.h (machine_function)
21232 <is_no_gccisr, use_L__stack_usage>: New fields.
21233 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21234 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21235 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21236 (gasisr, *gasisr): New expander and insn.
21237 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21238 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21239 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21241 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21243 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21246 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21248 Move jump-tables out of .text again.
21251 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21252 (ASM_OUTPUT_ADDR_VEC): New function.
21253 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21254 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21255 INSN_ADDRESSes as asm comment.
21256 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21257 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21258 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21259 * config/avr/avr.md (*tablejump): Adjust comment.
21260 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21261 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21263 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21264 (avr_output_addr_vec): New proto.
21265 (avr_log_t) <insn_addresses>: New field.
21267 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21270 * config/i386/i386.c (ix86_function_arg_advance): Set
21271 outgoing_args_on_stack to true if there are outgoing arguments
21273 (ix86_function_arg): Likewise.
21274 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21275 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21276 * config/i386/i386.h (machine_function): Add
21277 outgoing_args_on_stack.
21279 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21281 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21282 supporting pthreds.
21283 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21285 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21287 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21288 (REAL_H): Remove $(MACHMODE_H).
21289 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21291 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21292 $(MACHMODE_H) and double-int.h.
21293 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21295 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21298 2017-07-07 Andrew Pinski <apinski@cavium.com>
21300 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21301 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21303 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21305 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21306 Add warning if GCC was not configured to link against a GLIBC that
21307 exports the hardware capability bits.
21308 (make_resolver_func): Make resolver function private and not a
21309 COMDAT function. Create the name with clone_function_name instead
21310 of make_unique_name.
21313 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21314 correct operand in doing the split.
21316 2017-07-07 Carl Love <cel@us.ibm.com>
21318 * config/rs6000/rs6000-c: Add support for built-in function
21319 vector unsigned short vec_pack_to_short_fp32 (vector float,
21321 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21322 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21323 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21324 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21325 (convert_4f32_8i16): Add define_expand.
21326 * doc/extend.texi: Update the built-in documentation file for the
21327 new built-in function.
21329 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21331 * config/sparc/m8.md: New file.
21332 * config/sparc/sparc.md: Include m8.md.
21334 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21336 * config/sparc/sparc.opt: New option -mvis4b.
21337 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21338 (sparc_option_override): Handle VIS4B.
21339 (enum sparc_builtins): Define
21340 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21341 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21342 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21343 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21344 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21345 (check_constant_argument): New function.
21346 (sparc_vis_init_builtins): Define builtins
21347 __builtin_vis_dictunpack{8,16,32},
21348 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21349 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21350 __builtin_vis_fpcmpde{8,16,32}shl and
21351 __builtin_vis_fpcmpur{8,16,32}shl.
21352 (sparc_expand_builtin): Check that the constant operands to
21353 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21354 constant and in range.
21355 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21357 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21358 (SPARC_IMM5_P): Likewise.
21359 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21360 (enabled): Handle vis4b.
21361 (UNSPEC_DICTUNPACK): New unspec.
21362 (UNSPEC_FPCMPSHL): Likewise.
21363 (UNSPEC_FPUCMPSHL): Likewise.
21364 (UNSPEC_FPCMPDESHL): Likewise.
21365 (UNSPEC_FPCMPURSHL): Likewise.
21366 (cpu_feature): New CPU feature `vis4b'.
21367 (dictunpack{8,16,32}): New insns.
21368 (FPCSMODE): New mode iterator.
21369 (fpcscond): New code iterator.
21370 (fpcsucond): Likewise.
21371 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21372 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21373 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21374 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21375 * config/sparc/constraints.md: Define constraints `q' for unsigned
21376 2-bit integer constants and `t' for unsigned 5-bit integer
21378 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21380 (imm5_operand_dictunpack16): Likewise.
21381 (imm5_operand_dictunpack32): Likewise.
21382 (imm2_operand): Likewise.
21383 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21384 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21385 ditunpack* and fpcmp*shl builtins.
21387 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21389 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21390 * config.in: Add HAVE_AS_SPARC6 define.
21391 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21393 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21395 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21396 (CPP_CPU_SPEC): Handle m8.
21397 (ASM_CPU_SPEC): Likewise.
21398 * config/sparc/sparc-opts.h (enum processor_type): Add
21400 * config/sparc/sparc.c (m8_costs): New struct.
21401 (sparc_option_override): Handle TARGET_CPU_m8.
21402 (sparc32_initialize_trampoline): Likewise.
21403 (sparc64_initialize_trampoline): Likewise.
21404 (sparc_issue_rate): Likewise.
21405 (sparc_register_move_cost): Likewise.
21406 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21407 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21408 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21409 (CPP_CPU_SPEC): Handle M8.
21410 (ASM_CPU_SPEC): Likewise.
21411 (AS_M8_FLAG): Define.
21412 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21413 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21414 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21416 * configure: Regenerate.
21417 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21420 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21422 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21424 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21425 ("*movdi_insn_sp32"): Do not set v3pipe.
21426 ("*movsi_insn"): Likewise.
21427 ("*movdi_insn_sp64"): Likewise.
21428 ("*movsf_insn"): Likewise.
21429 ("*movdf_insn_sp32"): Likewise.
21430 ("*movdf_insn_sp64"): Likewise.
21431 ("*zero_extendsidi2_insn_sp64"): Likewise.
21432 ("*sign_extendsidi2_insn"): Likewise.
21433 ("*mov<VM32:mode>_insn"): Likewise.
21434 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21435 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21436 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21437 ("<vlop:code><VL:mode>3"): Likewise.
21438 ("*not_<vlop:code><VL:mode>3"): Likewise.
21439 ("*nand<VL:mode>_vis"): Likewise.
21440 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21441 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21442 ("one_cmpl<VL:mode>2"): Likewise.
21443 ("faligndata<VM64:mode>_vis"): Likewise.
21444 ("alignaddrsi_vis"): Likewise.
21445 ("alignaddrdi_vis"): Likweise.
21446 ("alignaddrlsi_vis"): Likewise.
21447 ("alignaddrldi_vis"): Likewise.
21448 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21449 ("bmaskdi_vis"): Likewise.
21450 ("bmasksi_vis"): Likewise.
21451 ("bshuffle<VM64:mode>_vis"): Likewise.
21452 ("cmask8<P:mode>_vis"): Likewise.
21453 ("cmask16<P:mode>_vis"): Likewise.
21454 ("cmask32<P:mode>_vis"): Likewise.
21455 ("pdistn<P:mode>_vis"): Likewise.
21456 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21458 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21460 * config/sparc/sparc.md ("subtype"): New insn attribute.
21461 ("*wrgsr_sp64"): Set insn subtype.
21462 ("*rdgsr_sp64"): Likewise.
21463 ("alignaddrsi_vis"): Likewise.
21464 ("alignaddrdi_vis"): Likewise.
21465 ("alignaddrlsi_vis"): Likewise.
21466 ("alignaddrldi_vis"): Likewise.
21467 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21468 ("fexpand_vis"): Likewise.
21469 ("fpmerge_vis"): Likewise.
21470 ("faligndata<VM64:mode>_vis"): Likewise.
21471 ("bshuffle<VM64:mode>_vis"): Likewise.
21472 ("cmask8<P:mode>_vis"): Likewise.
21473 ("cmask16<P:mode>_vis"): Likewise.
21474 ("cmask32<P:mode>_vis"): Likewise.
21475 ("fchksm16_vis"): Likewise.
21476 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21477 ("fmean16_vis"): Likewise.
21478 ("fp<plusminus_insn>64_vis"): Likewise.
21479 ("<plusminus_insn>v8qi3"): Likewise.
21480 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21481 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21482 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21483 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21484 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21485 ("*movqi_insn"): Likewise.
21486 ("*movhi_insn"): Likewise.
21487 ("*movsi_insn"): Likewise.
21488 ("movsi_pic_gotdata_op"): Likewise.
21489 ("*movdi_insn_sp32"): Likewise.
21490 ("*movdi_insn_sp64"): Likewise.
21491 ("movdi_pic_gotdata_op"): Likewise.
21492 ("*movsf_insn"): Likewise.
21493 ("*movdf_insn_sp32"): Likewise.
21494 ("*movdf_insn_sp64"): Likewise.
21495 ("*zero_extendhisi2_insn"): Likewise.
21496 ("*zero_extendqihi2_insn"): Likewise.
21497 ("*zero_extendqisi2_insn"): Likewise.
21498 ("*zero_extendqidi2_insn"): Likewise.
21499 ("*zero_extendhidi2_insn"): Likewise.
21500 ("*zero_extendsidi2_insn_sp64"): Likewise.
21501 ("ldfsr"): Likewise.
21502 ("prefetch_64"): Likewise.
21503 ("prefetch_32"): Likewise.
21504 ("tie_ld32"): Likewise.
21505 ("tie_ld64"): Likewise.
21506 ("*tldo_ldub_sp32"): Likewise.
21507 ("*tldo_ldub1_sp32"): Likewise.
21508 ("*tldo_ldub2_sp32"): Likewise.
21509 ("*tldo_ldub_sp64"): Likewise.
21510 ("*tldo_ldub1_sp64"): Likewise.
21511 ("*tldo_ldub2_sp64"): Likewise.
21512 ("*tldo_ldub3_sp64"): Likewise.
21513 ("*tldo_lduh_sp32"): Likewise.
21514 ("*tldo_lduh1_sp32"): Likewise.
21515 ("*tldo_lduh_sp64"): Likewise.
21516 ("*tldo_lduh1_sp64"): Likewise.
21517 ("*tldo_lduh2_sp64"): Likewise.
21518 ("*tldo_lduw_sp32"): Likewise.
21519 ("*tldo_lduw_sp64"): Likewise.
21520 ("*tldo_lduw1_sp64"): Likewise.
21521 ("*tldo_ldx_sp64"): Likewise.
21522 ("*mov<VM32:mode>_insn"): Likewise.
21523 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21524 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21526 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21528 * config/sparc/sparc.md ("type"): New insn type viscmp.
21529 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21531 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21532 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21533 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21534 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21536 ("n7_vis_logical_11cycle"): Likewise.
21537 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21538 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21539 * config/sparc/niagara.md ("niag_vis"): Likewise.
21540 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21541 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21543 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21545 * config/sparc/sparc.md: New instruction type `bmask'.
21546 (bmaskdi_vis): Use the `bmask' type.
21547 (bmasksi_vis): Likewise.
21548 * config/sparc/ultra3.md (us3_array): Likewise.
21549 * config/sparc/niagara7.md (n7_array): Likewise.
21550 * config/sparc/niagara4.md (n4_array): Likewise.
21551 * config/sparc/niagara2.md (niag2_vis): Likewise.
21552 (niag3_vis): Likewise.
21553 * config/sparc/niagara.md (niag_vis): Likewise.
21555 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21557 * ipa-comdats.c: Remove optimize check from gate.
21558 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21559 for functions not optimized.
21560 (ipa_fn_summary_read): Skip optimize check.
21561 (ipa_fn_summary_write): Likewise.
21562 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21564 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21566 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21567 (check_callers): Give up on caller which is not optimized.
21568 (inline_small_functions): Likewise.
21569 (ipa_inline): Do not give up when not optimizing.
21570 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21571 away unoptimizes cdtors.
21572 (whole_program_function_and_variable_visibility): Do
21573 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21574 * ipa.c (process_references): Do not check optimize.
21575 (symbol_table::remove_unreachable_nodes): Update optimize check.
21576 (set_writeonly_bit): Update optimize check.
21577 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21578 (pass_ipa_single_use::gate): Remove.
21580 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21582 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21583 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21584 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21585 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21586 permute_load, permute_store, adjust_extract, adjust_splat,
21587 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21588 replace_swap_with_copy, dump_swap_insn_table,
21589 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21590 recombine_lvx_pattern, recombine_stvx_pattern,
21591 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21592 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21593 to file rs6000-p8swap.c.
21594 * config/rs6000/rs6000-p8swap.c: New file.
21595 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21596 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21597 and rs6000*-*-* targets.
21599 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21601 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21603 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21605 * lto-wrapper.c (merge_and_complain): Do not merge
21606 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21607 fsigned_zeros, ftrapping_math, fwrapv.
21608 (append_compiler_options): Do not track these options.
21609 (append_linker_options): Likewie
21611 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21613 * cgraphunit.c (cgraph_node::finalize_function): When
21614 !flag_toplevel_reorde set no_reorder flag.
21615 (varpool_node::finalize_decl): Likewise.
21616 (symbol_table::compile): Drop no toplevel reorder path.
21618 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21620 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21621 edges; zero probability is not better than uninitialized.
21623 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21625 * asan.h (asan_sanitize_allocas_p): Declare.
21626 * asan.c (asan_sanitize_allocas_p): New function.
21627 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21628 (handle_builtin_alloca): Likewise.
21629 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21630 if !asan_sanitize_allocas_p.
21631 * params.def (asan-instrument-allocas): Add new option.
21632 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21633 * opts.c (common_handle_option): Disable allocas sanitization for
21636 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21638 * asan.c: Include gimple-fold.h.
21639 (get_last_alloca_addr): New function.
21640 (handle_builtin_stackrestore): Likewise.
21641 (handle_builtin_alloca): Likewise.
21642 (asan_emit_allocas_unpoison): Likewise.
21643 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21644 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21645 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21646 (instrument_builtin_call): Pass gimple iterator to
21647 get_mem_refs_of_builtin_call.
21648 (last_alloca_addr): New global.
21649 * asan.h (asan_emit_allocas_unpoison): Declare.
21650 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21651 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21652 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21653 if function calls alloca.
21654 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21655 * gimple-fold.h (replace_call_with_value): Declare.
21656 * internal-fn.c: Include asan.h.
21657 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21658 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21660 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21662 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21663 (C_SELFTEST_FLAGS): New.
21664 (CPP_SELFTEST_FLAGS): New.
21665 (SELFTEST_DEPS): New, from deps of s-selftest.
21666 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21667 (CPP_SELFTEST_DEPS): New.
21668 (selftest): Add dependency on s-selftest-c++.
21669 (s-selftest): Rename to...
21670 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21671 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21672 than SELFTEST_FLAGS.
21673 (selftest-gdb): Rename to...
21674 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21676 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21677 (selftest-valgrind): Rename to...
21678 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21680 (selftest-valgrind): Reintroduce as an alias for
21681 selftest-c-valgrind.
21682 (s-selftest-c++): New.
21683 (selftest-c++-gdb): New.
21684 (selftest-c++-valgrind): New.
21686 2017-07-06 Olivier Hainque <hainque@adacore.com>
21688 * gcc.c (process_command): When deciding if undefined variables
21689 should be ignored when processing specs, accept "gcc -v" as well.
21691 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21693 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21694 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21696 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21698 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21699 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21700 * config/arm/arm-tables.opt: Regenerate.
21701 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21703 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21705 2017-07-06 Carl Love <cel@us.ibm.com>
21707 * ChangeLog: Clean up from mid air collision
21709 2017-07-06 Carl Love <cel@us.ibm.com>
21711 * config/rs6000/rs6000-c.c: Add support for built-in functions
21712 vector signed int vec_subc (vector signed int, vector signed int);
21713 vector signed __int128 vec_subc (vector signed __int128,
21714 vector signed __int128);
21715 vector unsigned __int128 vec_subc (vector unsigned __int128,
21716 vector unsigned __int128);
21717 vector signed int vec_sube (vector signed int, vector signed int,
21718 vector signed int);
21719 vector unsigned int vec_sube (vector unsigned int,
21720 vector unsigned int,
21721 vector unsigned int);
21722 vector signed __int128 vec_sube (vector signed __int128,
21723 vector signed __int128,
21724 vector signed__int128);
21725 vector unsigned __int128 vec_sube (vector unsigned __int128,
21726 vector unsigned __int128,
21727 vector unsigned __int128);
21728 vector signed int vec_subec (vector signed int, vector signed int,
21729 vector signed int);
21730 vector unsigned int vec_subec (vector unsigned int,
21731 vector unsigned int,
21732 vector unsigned int);
21733 vector signed __int128 vec_subec (vector signed __int128,
21734 vector signed __int128,
21735 vector signed__int128);
21736 vector unsigned __int128 vec_subec (vector unsigned __int128,
21737 vector unsigned __int128,
21738 vector unsigned __int128);
21739 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21740 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21741 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21742 BU_ALTIVEC_OVERLOAD_X definitions.
21743 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21744 * doc/extend.texi: Update the built-in documentation file for the new
21745 built-in functions.
21747 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21750 * diagnostic-show-locus.c (layout::layout): Use start and finish
21751 spelling location for the start and finish of each range.
21752 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21753 Add unused aspect param.
21754 * input.c (expand_location_1): Add "aspect" param, and use it
21755 to access the correct part of the location.
21756 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21758 (expand_location_to_spelling_point): Likewise.
21759 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21760 param, and pass it to expand_location_1.
21762 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21764 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21765 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21766 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21767 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21768 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21769 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21770 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21771 _mm_maskz_getmant_ss): New intrinsics.
21772 (__builtin_ia32_getexpss128_mask): Changed to ...
21773 __builtin_ia32_getexpss128_round ... this.
21774 (__builtin_ia32_getexpsd128_mask): Changed to ...
21775 __builtin_ia32_getexpsd128_round ... this.
21776 * config/i386/i386-builtin-types.def
21777 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21778 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21779 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21780 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21781 __builtin_ia32_getmantss_mask_round): New builtins.
21782 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21783 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21784 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21785 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21786 * config/i386/sse.md
21787 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21788 avx512f_sgetexp<mode><mask_scalar_name>
21789 <round_saeonly_scalar_name> ... this.
21790 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21791 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21792 vgetexp<ssescalarmodesuffix>
21793 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21794 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21795 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21796 avx512f_vgetmant<mode><mask_scalar_name>
21797 <round_saeonly_scalar_name> ... this.
21798 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21799 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21800 vgetmant<ssescalarmodesuffix>
21801 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21802 %0<mask_scalar_operand4>, %1, %2
21803 <round_saeonly_scalar_mask_op4>, %3} ... this.
21804 * config/i386/subst.md (mask_scalar_operand4,
21805 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21806 round_saeonly_scalar_nimm_predicate): New subst attributes.
21808 2017-07-06 Julia Koval <julia.koval@intel.com>
21810 * config/i386/i386.c (ix86_erase_embedded_rounding):
21811 Remove code for old rounding pattern.
21813 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21815 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21817 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21819 * doc/sourcebuild.texi (Test Directives, Variants of
21820 dg-require-support): Add documentation for dg-require-stack-check.
21822 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21824 * config/i386/subst.md (mask_scalar, round_scalar,
21825 round_saeonly_scalar): New meta-templates.
21826 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21827 round_scalar_mask_operand3, round_scalar_mask_op3,
21828 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21829 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21830 round_saeonly_scalar_constraint,
21831 round_saeonly_scalar_prefix): New subst attribute.
21832 * config/i386/sse.md
21833 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21834 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21835 <round_scalar_name> ... this.
21836 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21837 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21838 <round_scalar_name> ... this.
21839 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21840 <sse>_vm<code><mode>3<mask_scalar_name>
21841 <round_saeonly_scalar_name> ... this.
21842 (v<plusminus_mnemonic><ssescalarmodesuffix>
21843 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21844 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21845 v<plusminus_mnemonic><ssescalarmodesuffix>
21846 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21847 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21848 (v<multdiv_mnemonic><ssescalarmodesuffix>
21849 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21850 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21851 v<multdiv_mnemonic><ssescalarmodesuffix>
21852 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21853 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21854 (v<maxmin_float><ssescalarmodesuffix>
21855 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21856 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21857 v<maxmin_float><ssescalarmodesuffix>
21858 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21859 %0<mask_scalar_operand3>, %1, %<iptr>2
21860 <round_saeonly_scalar_mask_op3>} ... this.
21862 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21864 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21865 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21867 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21868 Alan Hayward <alan.hayward@arm.com>
21869 David Sherwood <david.sherwood@arm.com>
21871 * combine.c (simplify_if_then_else): Remove "enum" before
21873 * compare-elim.c (can_eliminate_compare): Likewise.
21874 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21876 (aarch64_lookup_simd_builtin_type): Likewise.
21877 (aarch64_simd_builtin_type): Likewise.
21878 (aarch64_init_simd_builtin_types): Likewise.
21879 (aarch64_simd_expand_args): Likewise.
21880 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21882 (aarch64_reverse_mask): Likewise.
21883 (aarch64_simd_emit_reg_reg_move): Likewise.
21884 (aarch64_gen_adjusted_ldpstp): Likewise.
21885 (aarch64_ccmp_mode_to_code): Likewise.
21886 (aarch64_operands_ok_for_ldpstp): Likewise.
21887 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21888 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21890 (aarch64_min_divisions_for_recip_mul): Likewise.
21891 (aarch64_reassociation_width): Likewise.
21892 (aarch64_get_condition_code_1): Likewise.
21893 (aarch64_simd_emit_reg_reg_move): Likewise.
21894 (aarch64_simd_attr_length_rglist): Likewise.
21895 (aarch64_reverse_mask): Likewise.
21896 (aarch64_operands_ok_for_ldpstp): Likewise.
21897 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21898 (aarch64_gen_adjusted_ldpstp): Likewise.
21899 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21901 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21902 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21903 (arm_lookup_simd_builtin_type): Likewise.
21904 (arm_simd_builtin_type): Likewise.
21905 (arm_init_simd_builtin_types): Likewise.
21906 (arm_expand_builtin_args): Likewise.
21907 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21908 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21909 (ft32_setup_incoming_varargs): Likewise.
21910 (ft32_function_arg): Likewise.
21911 (ft32_function_arg_advance): Likewise.
21912 (ft32_pass_by_reference): Likewise.
21913 (ft32_arg_partial_bytes): Likewise.
21914 (ft32_valid_pointer_mode): Likewise.
21915 (ft32_addr_space_pointer_mode): Likewise.
21916 (ft32_addr_space_legitimate_address_p): Likewise.
21917 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21919 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21920 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21921 (iamcu_alignment): Likewise.
21922 (canonicalize_vector_int_perm): Likewise.
21923 (ix86_noce_conversion_profitable_p): Likewise.
21924 (ix86_mpx_bound_mode): Likewise.
21925 (ix86_operands_ok_for_move_multiple): Likewise.
21926 * config/microblaze/microblaze-protos.h
21927 (microblaze_expand_conditional_branch_reg): Likewise.
21928 * config/microblaze/microblaze.c
21929 (microblaze_expand_conditional_branch_reg): Likewise.
21930 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21932 (rs6000_reassociation_width): Likewise.
21933 (rs6000_invalid_binary_op): Likewise.
21934 (fusion_p9_p): Likewise.
21935 (emit_fusion_p9_load): Likewise.
21936 (emit_fusion_p9_store): Likewise.
21937 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21939 (riscv_hard_regno_mode_ok_p): Likewise.
21940 (riscv_address_insns): Likewise.
21941 (riscv_split_symbol): Likewise.
21942 (riscv_legitimize_move): Likewise.
21943 (riscv_function_value): Likewise.
21944 (riscv_hard_regno_nregs): Likewise.
21945 (riscv_expand_builtin): Likewise.
21946 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21947 (riscv_build_integer): Likewise.
21948 (riscv_split_integer): Likewise.
21949 (riscv_legitimate_constant_p): Likewise.
21950 (riscv_cannot_force_const_mem): Likewise.
21951 (riscv_regno_mode_ok_for_base_p): Likewise.
21952 (riscv_valid_base_register_p): Likewise.
21953 (riscv_valid_offset_p): Likewise.
21954 (riscv_valid_lo_sum_p): Likewise.
21955 (riscv_classify_address): Likewise.
21956 (riscv_legitimate_address_p): Likewise.
21957 (riscv_address_insns): Likewise.
21958 (riscv_load_store_insns): Likewise.
21959 (riscv_force_binary): Likewise.
21960 (riscv_split_symbol): Likewise.
21961 (riscv_force_address): Likewise.
21962 (riscv_legitimize_address): Likewise.
21963 (riscv_move_integer): Likewise.
21964 (riscv_legitimize_const_move): Likewise.
21965 (riscv_legitimize_move): Likewise.
21966 (riscv_address_cost): Likewise.
21967 (riscv_subword): Likewise.
21968 (riscv_output_move): Likewise.
21969 (riscv_canonicalize_int_order_test): Likewise.
21970 (riscv_emit_int_order_test): Likewise.
21971 (riscv_function_arg_boundary): Likewise.
21972 (riscv_pass_mode_in_fpr_p): Likewise.
21973 (riscv_pass_fpr_single): Likewise.
21974 (riscv_pass_fpr_pair): Likewise.
21975 (riscv_get_arg_info): Likewise.
21976 (riscv_function_arg): Likewise.
21977 (riscv_function_arg_advance): Likewise.
21978 (riscv_arg_partial_bytes): Likewise.
21979 (riscv_function_value): Likewise.
21980 (riscv_pass_by_reference): Likewise.
21981 (riscv_setup_incoming_varargs): Likewise.
21982 (riscv_print_operand): Likewise.
21983 (riscv_elf_select_rtx_section): Likewise.
21984 (riscv_save_restore_reg): Likewise.
21985 (riscv_for_each_saved_reg): Likewise.
21986 (riscv_register_move_cost): Likewise.
21987 (riscv_hard_regno_mode_ok_p): Likewise.
21988 (riscv_hard_regno_nregs): Likewise.
21989 (riscv_class_max_nregs): Likewise.
21990 (riscv_memory_move_cost): Likewise.
21991 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21992 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21993 (rl78_addr_space_address_mode): Likewise.
21994 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21996 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21997 (rs6000_reassociation_width): Likewise.
21998 (rs6000_invalid_binary_op): Likewise.
21999 (fusion_p9_p): Likewise.
22000 (emit_fusion_p9_load): Likewise.
22001 (emit_fusion_p9_store): Likewise.
22002 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22003 (ok_for_simple_move_operands): Likewise.
22004 (ok_for_simple_move_strict_operands): Likewise.
22005 (ok_for_simple_arith_logic_operands): Likewise.
22006 (visium_legitimize_reload_address): Likewise.
22007 (visium_select_cc_mode): Likewise.
22008 (output_cbranch): Likewise.
22009 (visium_split_double_move): Likewise.
22010 (visium_expand_copysign): Likewise.
22011 (visium_expand_int_cstore): Likewise.
22012 (visium_expand_fp_cstore): Likewise.
22013 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22014 (visium_function_arg): Likewise.
22015 (visium_function_arg_advance): Likewise.
22016 (visium_libcall_value): Likewise.
22017 (visium_setup_incoming_varargs): Likewise.
22018 (visium_legitimate_constant_p): Likewise.
22019 (visium_legitimate_address_p): Likewise.
22020 (visium_legitimize_address): Likewise.
22021 (visium_secondary_reload): Likewise.
22022 (visium_register_move_cost): Likewise.
22023 (visium_memory_move_cost): Likewise.
22024 (prepare_move_operands): Likewise.
22025 (ok_for_simple_move_operands): Likewise.
22026 (ok_for_simple_move_strict_operands): Likewise.
22027 (ok_for_simple_arith_logic_operands): Likewise.
22028 (visium_function_value_1): Likewise.
22029 (rtx_ok_for_offset_p): Likewise.
22030 (visium_legitimize_reload_address): Likewise.
22031 (visium_split_double_move): Likewise.
22032 (visium_expand_copysign): Likewise.
22033 (visium_expand_int_cstore): Likewise.
22034 (visium_expand_fp_cstore): Likewise.
22035 (visium_split_cstore): Likewise.
22036 (visium_select_cc_mode): Likewise.
22037 (visium_split_cbranch): Likewise.
22038 (output_cbranch): Likewise.
22039 (visium_print_operand_address): Likewise.
22040 * expmed.c (flip_storage_order): Likewise.
22041 * expmed.h (emit_cstore): Likewise.
22042 (flip_storage_order): Likewise.
22043 * genrecog.c (validate_pattern): Likewise.
22044 * hsa-gen.c (gen_hsa_addr): Likewise.
22045 * internal-fn.c (expand_arith_overflow): Likewise.
22046 * ira-color.c (allocno_copy_cost_saving): Likewise.
22047 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22048 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22049 (process_invariant_for_inheritance): Likewise.
22050 * lra-eliminations.c (move_plus_up): Likewise.
22051 * omp-low.c (lower_oacc_reductions): Likewise.
22052 * simplify-rtx.c (simplify_subreg): Likewise.
22053 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22054 (TARGET_CHKP_BOUND_MODE): Likewise..
22055 * targhooks.c (default_chkp_bound_mode): Likewise.
22056 (default_setup_incoming_vararg_bounds): Likewise.
22057 * targhooks.h (default_chkp_bound_mode): Likewise.
22058 (default_setup_incoming_vararg_bounds): Likewise.
22059 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22060 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22061 (have_whole_vector_shift): Likewise.
22062 * tree-vect-stmts.c (vectorizable_load): Likewise.
22063 * doc/tm.texi: Regenerate.
22065 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22067 Graceful degrade if Binutils PR21472 is not available.
22070 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22071 .rodata in flash test fails.
22072 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22073 * confgure: Regenerate.
22074 * config.in: Regenerate.
22075 * config/avr/avr.c (avr_asm_named_section)
22076 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22077 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22078 (avr_asm_init_sections): Same.
22080 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22082 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22083 (fma<VH:mode>4_intrinsic): Likewise.
22084 (*fmsub<VCVTF:mode>4): Likewise.
22085 (*fmsub<VH:mode>4_intrinsic): Likewise.
22087 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22090 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22091 Don't depend on "optimize > 0".
22092 (out_movhi_r_mr, out_movqi_mr_r): Same.
22093 (out_movhi_mr_r, out_movqi_r_mr): Same.
22094 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22095 io_address_operand on "optimize > 0".
22097 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22099 * tree-loop-distribution.c: Add general explanantion on the pass.
22100 (generate_loops_for_partition): Mark distributed loop.
22101 (pg_add_dependence_edges): New parameter. Handle alias data
22102 dependence specially and record it in the parameter if asked.
22103 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22104 (init_partition_graph_vertices, add_partition_graph_edge): New.
22105 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22106 (free_partition_graph_vdata, build_partition_graph): New.
22107 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22108 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22109 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22110 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22111 (version_for_distribution_p, finalize_partitions): New.
22112 (distribute_loop): Handle alias data dependence specially. Factor
22113 out loop fusion code as functions and call these functions.
22115 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22117 * tree-loop-distribution.c (classify_partition): New parameter and
22118 better handle reduction statement.
22119 (rdg_build_partitions): Revise comment.
22120 (distribute_loop): Compute statements in all partitions and pass it
22121 to classify_partition.
22123 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22125 * tree-loop-distribution.c (enum partition_type): New.
22126 (struct partition): New field type.
22127 (partition_merge_into): Add parameter. Update partition type.
22128 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22129 (build_rdg_partition_for_vertex): Compute partition type.
22130 (rdg_build_partitions): Dump partition type.
22131 (distribute_loop): Update calls to partition_merge_into.
22133 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22135 * tree-loop-distribution.c (struct ddr_hasher): New.
22136 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22138 (classify_partition): Call get_data_dependence.
22139 (pg_add_dependence_edges): Ditto.
22140 (distribute_loop): Release data dependence hash table.
22142 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22144 * tree-loop-distribution.c (ref_base_address): Delete.
22145 (similar_memory_accesses): Rename ...
22146 (share_memory_accesses): ... to this. Check if partitions access
22147 the same memory reference.
22148 (distribute_loop): Call share_memory_accesses.
22150 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22152 * tree-loop-distribution.c (struct partition): New field recording
22153 its data reference.
22154 (partition_alloc, partition_free): Init and release data refs.
22155 (partition_merge_into): Merge data refs.
22156 (build_rdg_partition_for_vertex): Collect data refs for partition.
22157 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22159 (distribute_loop): Remve data refs from vertice data of partition
22162 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22164 * tree-loop-distribution.c (params.h): Include header file.
22165 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22166 (datarefs_vec): New global var.
22167 (create_rdg_vertices): Use datarefs_vec directly.
22168 (free_rdg): Don't free data references.
22169 (build_rdg): Update use. Don't free data references.
22170 (distribute_loop): Compute global variable for data references.
22171 Bail out if there are too many data references.
22173 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22175 * tree-loop-distribution.c (loop_nest): New global var.
22176 (build_rdg): Use loop directly, rather than loop nest.
22177 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22179 (distribute_loop): Compute global variable loop nest. Update use.
22181 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22183 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22184 (partition_merge_into): New parameter. Dump reason for fusion.
22185 (distribute_loop): Update use of partition_merge_into.
22187 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22189 * tree-loop-distribution.c (bb_top_order_index): New.
22190 (bb_top_order_index_size, bb_top_order_cmp): New.
22191 (stmts_from_loop): Use topological order.
22192 (pass_loop_distribution::execute): Compute and release topological
22193 order for basic blocks.
22195 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22197 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22200 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22202 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22203 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22204 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22205 * internal-fn.def (LOOP_DIST_ALIAS): New.
22206 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22207 (fold_loop_internal_call): ... this.
22208 (vect_loop_dist_alias_call): New function.
22209 (set_uid_loop_bbs): Call fold_loop_internal_call.
22210 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22213 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22216 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22217 Require dead FLAGS_REG at the beginning of a peephole.
22219 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22222 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22223 arguments in the call to __builtin_ia32_sbb_u32.
22224 (_subborrow_u64): Swap _X and _Y arguments in the call to
22225 __builtin_ia32_sbb_u64.
22227 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22230 * tree-vrp.c (compare_assert_loc): Turn into a function template
22231 with stable template parameter. Only test if a->e is NULL,
22232 !a->e == !b->e has been verified already. Use e == NULL or
22233 e != NULL instead of e or ! e tests. If stable is true, don't use
22234 iterative_hash_expr, on the other side allow a or b or both NULL
22235 and sort the NULLs last.
22236 (process_assert_insertions): Sort using compare_assert_loc<false>
22237 instead of compare_assert_loc, later sort using
22238 compare_assert_loc<true> before calling process_assert_insertions_for
22239 in a loop. Use break instead of continue once seen NULL pointer.
22241 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22243 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22244 Cortex-R7 and Cortex-R8 processors.
22246 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22248 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22249 uninitialized while src is not.
22251 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22253 * common/config/arm/arm-common.c: Adjust include path for
22255 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22256 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22257 (arm-cpu-data.h): Likewise.
22258 (arm-cpu-cdata.h): Likewise.
22259 * config/arm/arm-cpu.h: Delete.
22260 * config/arm/arm-cpu-cdata.h: Delete.
22261 * config/arm/arm-cpu-data.h: Delete.
22263 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22265 * config/arm/arm-cpus.in (cortex-a55): New.
22266 (cortex-a75): Likewise.
22267 (cortex-a75.cortex-a55): Likewise.
22268 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22270 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22271 * config/arm/arm-cpu-cdata.h: Regenerate.
22272 * config/arm/arm-cpu-data.h: Regenerate.
22273 * config/arm/arm-cpu.h: Regenerate.
22274 * config/arm/arm-tables.opt: Regenerate.
22275 * config/arm/arm-tune.md: Regenerate.
22277 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22279 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22281 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22283 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22286 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22288 PR tree-optimization/81292
22289 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22290 full_string_p, also call adjust_related_strinfos if the adjustment
22291 is simple, otherwise invalidate related strinfos.
22293 2017-07-04 Martin Liska <mliska@suse.cz>
22296 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22297 newly created variable as DECL_IGNORED_P.
22299 2017-07-04 Martin Liska <mliska@suse.cz>
22302 * ipa-inline.c (inline_small_functions):
22303 Use xstrdup_for_dump.
22305 2017-07-04 Tom de Vries <tom@codesourcery.com>
22307 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22309 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22312 * config/darwin.c (darwin_function_switched_text_sections):
22315 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22317 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22319 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22321 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22323 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22325 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22326 min_profitable_iters, and th as inclusive lower bounds.
22327 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22328 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22329 for min_profitable_iters and min_profitable_estimate.
22330 (vect_transform_loop): Treat th as an inclusive lower bound.
22331 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22333 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22336 * config/darwin.c (darwin_function_switched_text_sections):
22337 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22338 in two pieces, and suppress the use of buf.
22340 2017-07-03 Nathan Sidwell <nathan@acm.org>
22342 * hash-table.h (hash_table_mod1): Fix indentation.
22344 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22346 PR middle-end/81290
22347 * predict.c (force_edge_cold): Be more careful about propagation
22349 * profile-count.h (profile_probability::guessed,
22350 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22352 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22354 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22356 * doc/invoke.texi (rcpc architecture extension): Document it.
22358 2017-07-03 Richard Biener <rguenther@suse.de>
22360 PR tree-optimization/60510
22361 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22362 the scalar reduction PHI and use it.
22363 (vectorizable_reduction): Properly guard the single_defuse_cycle
22364 path for non-SLP reduction chains where we cannot use it.
22365 Rework reduc_def/index and vector type deduction. Rework
22366 vector operand gathering during reduction op code-gen.
22367 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22368 chains dissolve the chain and leave it to non-SLP reduction
22371 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22373 * tree-data-ref.h (dr_alignment): Declare.
22374 * tree-data-ref.c (dr_alignment): New function.
22375 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22376 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22378 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22380 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22382 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22383 and base_misalignment fields.
22384 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22385 * tree-data-ref.c: Include builtins.h.
22386 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22387 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22388 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22389 * tree-vect-data-refs.c: Include tree-cfg.h.
22390 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22391 fields instead of calculating an alignment here.
22392 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22393 innermost_loop_behavior fields.
22395 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22397 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22399 (DR_STEP_ALIGNMENT): New macro.
22400 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22401 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22402 (create_data_ref): Print it.
22403 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22404 to tell whether the step preserves vector (mis)alignment.
22405 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22406 Move the check for an integer step and generalise to all INTEGER_CST.
22407 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22408 Print the outer step alignment.
22410 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22412 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22413 with offset_alignment.
22414 (DR_ALIGNED_TO): Delete.
22415 (DR_OFFSET_ALIGNMENT): New macro.
22416 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22417 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22418 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22419 (create_data_ref): Likewise.
22420 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22421 (vect_analyze_data_refs): Likewise.
22422 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22423 creating dummy innermost behavior.
22425 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22427 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22428 with a "innermost_loop_behavior *" and refeence tree.
22429 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22430 (create_data_ref): Update call accordingly.
22431 * tree-predcom.c (find_looparound_phi): Likewise.
22433 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22435 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22436 fields with dr_wrt_vec_loop.
22437 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22438 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22439 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22440 (vect_dr_behavior): New function.
22441 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22442 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22443 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22444 track whether the step preserves the misalignment.
22445 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22446 Use vect_dr_behavior.
22447 (vect_setup_realignment): Update call accordingly.
22448 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22449 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22450 call to vect_create_addr_base_for_vector_ref.
22451 (vect_create_cond_for_align_checks): Likewise.
22452 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22453 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22454 (vect_recog_mask_conversion_pattern): Likewise.
22455 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22456 (new_stmt_vec_info): Remove redundant zeroing.
22458 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22460 * common/config/arm/arm-common.c (arm_be8_option): New function.
22461 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22462 (ISA_ARMv6): Add isa_bit_be8.
22463 * config/arm/arm.h (arm_be8_option): Add prototype.
22464 (BE8_SPEC_FUNCTION): New define.
22465 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22466 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22467 (mlittle-endian): Similarly.
22468 (mbe8, mbe32): New options.
22469 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22470 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22472 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22474 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22476 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22478 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22479 (cleanup_tree_cfg_bb): Use it.
22480 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22482 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22484 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22487 * loop-doloop.c (add_test): Update profile.
22489 2017-07-03 Martin Liska <mliska@suse.cz>
22492 * sanopt.c (rewrite_usage_of_param): New function.
22493 (sanitize_rewrite_addressable_params): Likewise.
22494 (pass_sanopt::execute): Call rewrite_usage_of_param.
22496 2017-07-03 Richard Biener <rguenther@suse.de>
22498 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22499 back to using VIEW_CONVERT_EXPR.
22501 2017-07-03 Martin Liska <mliska@suse.cz>
22504 * doc/extend.texi: Document when a resolver function is
22505 generated for target_clones.
22507 2017-07-03 Martin Liska <mliska@suse.cz>
22509 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22510 and shadow memory of auto variables which are subject of
22511 use-after-scope sanitization.
22512 (asan_expand_mark_ifn): Add do set only when is_poison.
22514 2016-07-03 Richard Biener <rguenther@suse.de>
22516 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22518 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22519 (vectorizable_reduction): Perform reduction PHI creation when
22520 visiting a reduction PHI and adjust and simplify code generation
22521 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22522 (vect_transform_loop): Visit reduction PHIs.
22523 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22524 defs into the SLP tree.
22525 (vect_build_slp_tree): Reduction defs terminate the recursion.
22526 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22528 (vect_get_vec_defs_for_stmt_copy): Export.
22529 (vect_get_vec_defs): Likewise.
22530 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22532 (vect_get_vec_defs_for_stmt_copy): Declare.
22533 (vect_get_vec_defs): Likewise.
22535 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22537 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22538 parameter with a "loop" parameter and use it instead of the
22539 loop containing DR_STMT. Don't check simple_iv when doing
22540 BB analysis. Describe the two analysis modes in the comment.
22542 2017-07-03 Tom de Vries <tom@codesourcery.com>
22544 PR tree-optimization/69468
22545 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22546 (find_same_succ_bb): Handle ignore_edge_flags.
22548 2017-07-03 Tom de Vries <tom@codesourcery.com>
22550 PR tree-optimization/81192
22551 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22553 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22555 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22557 2017-07-03 Tom de Vries <tom@codesourcery.com>
22559 PR tree-optimization/81192
22560 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22561 BB_SAME_SUCC (bb) == NULL.
22563 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22565 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22568 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22570 * dumpfile.c: Include profile-count.h
22571 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22573 (insert_cond_bb): Update profile.
22574 * tree-cfg.h (insert_cond_bb): Update prototype.
22575 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22576 * tree-dump.c: Do not include tree-cfg.
22578 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22580 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22582 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22584 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22587 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22589 * tree-complex.c (expand_complex_div_wide): update profile.
22591 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22592 Alan Hayward <alan.hayward@arm.com>
22593 David Sherwood <david.sherwood@arm.com>
22595 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22596 (CORETYPES_H): New define.
22597 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22598 (insn-modes-inline.h, s-modes-inline-h): New rules.
22599 (generated_files): Add insn-modes-inline.h.
22600 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22601 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22602 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22603 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22604 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22605 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22606 (build/gencodes.o, build/genconditions.o): Likewise.
22607 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22608 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22609 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22610 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22611 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22612 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22613 * coretypes.h: Include everything up to real.h for generators.
22614 Include insn-modes.h first. Include wide-int-print.h after
22615 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22616 * machmode.h: Don't include insn-modes.h here.
22617 * function-tests.c: Remove includes of signop.h, machmode.h,
22618 double-int.h and wide-int.h.
22620 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22622 * optc-save-gen.awk: Likewise.
22623 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22624 * godump.c: Remove include of wide-int-print.h.
22625 * pretty-print.h: Likewise.
22626 * wide-int-print.cc: Likewise.
22627 * wide-int.cc: Likewise.
22628 * hash-map-tests.c: Remove include of signop.h.
22629 * hash-set-tests.c: Likewise.
22630 * rtl-tests.c: Likewise.
22631 * mkconfig.sh: Remove include of machmode.h.
22632 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22634 (emit_insn_modes_inline_h): ...this new function. Emit the code
22635 into an insn-modes-inline.h header file, adding appropriate
22636 include guards and end comments.
22637 (emit_insn_modes_c_header): Remove include of machmode.h.
22638 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22640 (main): Handle -i flag and call emit_insn_modes_inline_h when
22643 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22645 * tree-ssa-strlen.c (strinfo): Rename the length field to
22646 nonzero_chars. Add a full_string_p field.
22647 (compare_nonzero_chars, zero_length_string_p): New functions.
22648 (get_addr_stridx): Add an offset_out parameter.
22649 Use compare_nonzero_chars.
22650 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22651 (new_strinfo): Update after above changes to strinfo.
22652 (set_endptr_and_length): Set full_string_p.
22653 (get_string_length): Update after above changes to strinfo.
22654 (unshare_strinfo): Update call to new_strinfo.
22655 (maybe_invalidate): Likewise.
22656 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22657 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22658 as a uhwi instead of an shwi. Update after above changes to
22659 strinfo and new_strinfo.
22660 (zero_length_string): Assert that chainsi contains full strings.
22661 Use zero_length_string_p. Update call to new_strinfo.
22662 (adjust_related_strinfos): Update after above changes to strinfo.
22663 Copy full_string_p from origsi.
22664 (adjust_last_stmt): Use zero_length_string_p.
22665 (handle_builtin_strlen): Update after above changes to strinfo and
22666 new_strinfo. Install the lhs as the string length if the previous
22667 entry didn't describe a full string.
22668 (handle_builtin_strchr): Update after above changes to strinfo
22670 (handle_builtin_strcpy): Likewise.
22671 (handle_builtin_strcat): Likewise.
22672 (handle_builtin_malloc): Likewise.
22673 (handle_pointer_plus): Likewise.
22674 (handle_builtin_memcpy): Likewise. Track nonzero characters
22675 that aren't necessarily followed by a nul terminator.
22676 (handle_char_store): Likewise.
22678 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22680 PR tree-optimization/80769
22681 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22682 for malloc and calloc. Document the new invariant that all related
22683 strinfos have delayed lengths or none do.
22684 (verify_related_strinfos): Move earlier in file.
22685 (set_endptr_and_length): New function, split out from...
22686 (get_string_length): ...here. Also set the lengths of related
22688 (zero_length_string): Assert that chainsi has known (rather than
22690 (adjust_related_strinfos): Likewise.
22692 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22694 PR tree-optimization/81136
22695 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22696 assert that two references with the same misalignment have the same
22697 compile-time misalignment if those compile-time misalignments
22700 2017-07-01 Andi Kleen <ak@linux.intel.com>
22702 * print-tree.c (print_node): Print all attributes.
22704 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22706 * cfg.c (scale_bbs_frequencies): New function.
22707 * cfg.h (scale_bbs_frequencies): Declare it.
22708 * cfgloopanal.c (single_likely_exit): Cleanup.
22709 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22711 (scale_loop_profile): Likewise.
22712 (loop_version): Likewise.
22713 (create_empty_loop_on_edge): Update.
22714 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22715 scale_loop_frequencies, scale_loop_profile, loopify,
22716 loop_version): Update prototypes.
22717 * modulo-sched.c (sms_schedule): Update.
22718 * predict.c (unlikely_executed_edge_p): Also check probability.
22719 (probably_never_executed_edge_p): Fix typo.
22720 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22721 * tree-parloops.c (gen_parallel_loop): Update.
22722 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22723 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22724 * tree-ssa-loop-split.c (split_loop): Update.
22725 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22726 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22727 (vect_loop_versioning): Update.
22728 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22730 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22732 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22734 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22736 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22737 to keep profile consistent.
22739 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22741 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22742 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22743 * profile-count.h (max_safe_multiplier): Make unsigned.
22744 (profile_count::guessed_zero): New.
22746 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22748 * bb-reorder.c (fix_up_crossing_landing_pad,
22749 fix_crossing_conditional_branches): Use make_single_succ_edge
22750 to keep profile consistent.
22752 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22754 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22757 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22760 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22761 the right scopes, make sure cond_jump isn't preserved between multiple
22762 iterations. Search for fallthru edge whenever there are 3+ edges and
22763 use find_fallthru_edge for it.
22765 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22767 Patch by Alexander Monakov <amonakov@ispras.ru>
22768 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22769 probabilities consistently.
22771 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22773 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22774 emit_cmp_and_jump_insns.
22776 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22779 * tree-inline.c (expand_call_inline): Combine profile statuses.
22781 2017-06-30 Andrew Pinski <apinski@cavium.com>
22783 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22784 fold_stmt returned true.
22786 2017-06-30 Nathan Sidwell <nathan@acm.org>
22788 * ggc.h (empty_string): Delete.
22789 * cfgexpand.c (expand_asm_stmt): Use plain "".
22790 * optabs.c (expand_asm_memory_barrier): Likewise.
22791 * stringpool.c (empty_string): Delete.
22792 (digit_vector, digit_string): Delete.
22793 (ggc_alloc_string): Use plain "", don't optimize single digit
22794 strings. Use ggc_alloc_atomic.
22796 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22798 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22799 comparison set and one other set, use the cost of the non-comparison
22802 2017-06-30 Nathan Sidwell <nathan@acm.org>
22804 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22807 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22809 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22810 loops. Remove now unneeded calls to gimple_switch_set_label() that
22811 just set removed labels to NULL_TREE.
22813 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22815 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22816 (set_range_info): ...here. Only call set_range_info_raw if domain
22818 (set_nonzero_bits): Call set_range_info_raw.
22819 * tree-ssanames.h (set_range_info_raw): New.
22821 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22824 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22825 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22826 of nonimmediate_operand and <store_mask_constraint> instead of m
22827 for the input operand. For V8FI iterator, always split if input
22828 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22829 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22830 <store_mask_predicate> instead of register_operand and
22831 <store_mask_constraint> instead of v for the input operand. Make
22832 sure both operands aren't MEMs for if not <mask_applied>.
22834 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22836 * lto-wrapper.c (copy_file) Close both file descriptors before
22839 2017-06-30 Martin Liska <mliska@suse.cz>
22842 * multiple_target.c (create_dispatcher_calls): Make ifunc
22843 also for function that don't have calls or are not referenced.
22845 2017-06-30 Richard Biener <rguenther@suse.de>
22847 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22848 analyze the first scalar stmt. Move vector type computation
22849 for the BB case here from ...
22850 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22851 live operation processing in the SLP case properly.
22853 2017-06-30 Richard Biener <rguenther@suse.de>
22855 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22857 2017-06-30 Martin Liska <mliska@suse.cz>
22860 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22861 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22863 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22865 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22868 2017-06-30 Richard Biener <rguenther@suse.de>
22870 PR tree-optimization/81249
22871 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22872 condition reduction result to original scalar type.
22874 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22876 * profile-count.h (enum profile_quality): Fix typos and whitespace
22879 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22881 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22882 type for branch probabilities.
22884 2017-06-29 Julian Brown <julian@codesourcery.com>
22885 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22887 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22888 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22889 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22890 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22892 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22894 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22895 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22896 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22897 CC usage from generic code to here.
22898 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22899 CC usage into the target macros.
22901 2017-06-29 Maya Rashish <coypu@sdf.org>
22903 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22906 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22908 * arm/arm-builtins.c: Include profile-count.h
22909 * except.c (sjlj_emit_function_enter): Use
22910 profile_probability::unlikely.
22912 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22914 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22915 and tocrel_offset be pointer args rather than implicitly using
22917 (legitimate_constant_pool_address_p, rs6000_emit_move,
22918 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22919 tocrel_offset and use in toc_relative_expr_p call.
22920 (print_operand, print_operand_address): Use static tocrel_base_oac
22921 and tocrel_offset_oac.
22922 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22925 2017-06-29 Maya Rashish <coypu@sdf.org>
22927 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22929 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22931 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22932 objects, take into account only the alignment of 'op0' and 'mode1' if
22935 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22937 * ccmp.c (ccmp_tree_comparison_p): New function.
22938 (ccmp_candidate_p): Update to use above function.
22939 (get_compare_parts): New function.
22940 (expand_ccmp_next): Update to use new functions.
22941 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22943 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22944 take mode as argument.
22945 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22946 * expr.c (expand_expr_real_1): Pass mode as argument.
22948 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22950 * combine.c (combine_instructions): Print insns to dump_file, together
22953 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22955 * asan.c (asan_emit_stack_protection): Update.
22956 (create_cond_insert_point): Update.
22957 * auto-profile.c (afdo_propagate_circuit): Update.
22958 * basic-block.h (struct edge_def): Turn probability to
22959 profile_probability.
22960 (EDGE_FREQUENCY): Update.
22961 * bb-reorder.c (find_traces_1_round): Update.
22962 (better_edge_p): Update.
22963 (sanitize_hot_paths): Update.
22964 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22965 (make_single_succ_edge): Update.
22966 (check_bb_profile): Update.
22967 (dump_edge_info): Update.
22968 (update_bb_profile_for_threading): Update.
22969 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22970 probabilitycount to 0.
22971 * cfgbuild.c (compute_outgoing_frequencies): Update.
22972 * cfgcleanup.c (try_forward_edges): Update.
22973 (outgoing_edges_match): Update.
22974 (try_crossjump_to_edge): Update.
22975 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22976 (expand_gimple_tailcall): Update.
22977 (construct_init_block): Use make_single_succ_edge.
22978 (construct_exit_block): Use make_single_succ_edge.
22979 * cfghooks.c (verify_flow_info): Update.
22980 (redirect_edge_succ_nodup): Update.
22981 (split_edge): Update.
22982 (account_profile_record): Update.
22983 * cfgloopanal.c (single_likely_exit): Update.
22984 * cfgloopmanip.c (scale_loop_profile): Update.
22985 (set_zero_probability): Remove.
22986 (duplicate_loop_to_header_edge): Update.
22987 * cfgloopmanip.h (loop_version): Update prototype.
22988 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22989 (force_nonfallthru_and_redirect): Update.
22990 (update_br_prob_note): Update.
22991 (rtl_verify_edges): Update.
22992 (purge_dead_edges): Update.
22993 (rtl_lv_add_condition_to_bb): Update.
22994 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22995 * cgraphunit.c (init_lowered_empty_function): Update.
22996 (cgraph_node::expand_thunk): Update.
22997 * cilk-common.c: Include profile-count.h
22998 * dojump.c (inv): Remove.
22999 (jumpifnot): Update.
23000 (jumpifnot_1): Update.
23001 (do_jump_1): Update.
23003 (do_jump_by_parts_greater_rtx): Update.
23004 (do_compare_rtx_and_jump): Update.
23005 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23006 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23007 * dwarf2cfi.c: Include profile-count.h
23008 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23009 (sjlj_emit_dispatch_table): Likewise.
23010 * explow.c: Include profile-count.h
23011 * expmed.c (emit_store_flag_force): Update.
23012 (do_cmp_and_jump): Update.
23013 * expr.c (compare_by_pieces_d::generate): Update.
23014 (compare_by_pieces_d::finish_mode): Update.
23015 (emit_block_move_via_loop): Update.
23016 (store_expr_with_bounds): Update.
23017 (store_constructor): Update.
23018 (expand_expr_real_2): Update.
23019 (expand_expr_real_1): Update.
23020 * expr.h (try_casesi, try_tablejump): Update prototypes.
23021 * gimple-pretty-print.c (dump_probability): Update.
23022 (dump_profile): New.
23023 (dump_gimple_label): Update.
23024 (dump_gimple_bb_header): Update.
23025 * graph.c (draw_cfg_node_succ_edges): Update.
23026 * hsa-gen.c (convert_switch_statements): Update.
23027 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23028 (find_if_case_1): Update.
23029 (find_if_case_2): Update.
23030 * internal-fn.c (expand_arith_overflow_result_store): Update.
23031 (expand_addsub_overflow): Update.
23032 (expand_neg_overflow): Update.
23033 (expand_mul_overflow): Update.
23034 (expand_vector_ubsan_overflow): Update.
23035 * ipa-cp.c (good_cloning_opportunity_p): Update.
23036 * ipa-split.c (split_function): Use make_single_succ_edge.
23037 * ipa-utils.c (ipa_merge_profiles): Update.
23038 * loop-doloop.c (add_test): Update.
23039 (doloop_modify): Update.
23040 * loop-unroll.c (compare_and_jump_seq): Update.
23041 (unroll_loop_runtime_iterations): Update.
23042 * lra-constraints.c (lra_inheritance): Update.
23043 * lto-streamer-in.c (input_cfg): Update.
23044 * lto-streamer-out.c (output_cfg): Update.
23045 * mcf.c (adjust_cfg_counts): Update.
23046 * modulo-sched.c (sms_schedule): Update.
23047 * omp-expand.c (expand_omp_for_init_counts): Update.
23048 (extract_omp_for_update_vars): Update.
23049 (expand_omp_ordered_sink): Update.
23050 (expand_omp_for_ordered_loops): Update.
23051 (expand_omp_for_generic): Update.
23052 (expand_omp_for_static_nochunk): Update.
23053 (expand_omp_for_static_chunk): Update.
23054 (expand_cilk_for): Update.
23055 (expand_omp_simd): Update.
23056 (expand_omp_taskloop_for_outer): Update.
23057 (expand_omp_taskloop_for_inner): Update.
23058 * omp-simd-clone.c (simd_clone_adjust): Update.
23059 * optabs.c (expand_doubleword_shift): Update.
23060 (expand_abs): Update.
23061 (emit_cmp_and_jump_insn_1): Update.
23062 (expand_compare_and_swap_loop): Update.
23063 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23064 * predict.c (predictable_edge_p): Update.
23065 (edge_probability_reliable_p): Update.
23066 (set_even_probabilities): Update.
23067 (combine_predictions_for_insn): Update.
23068 (combine_predictions_for_bb): Update.
23069 (propagate_freq): Update.
23070 (estimate_bb_frequencies): Update.
23071 (force_edge_cold): Update.
23072 * profile-count.c (profile_count::dump): Add missing space into dump.
23073 (profile_count::debug): Add newline.
23074 (profile_count::differs_from_p): Explicitly convert to unsigned.
23075 (profile_count::stream_in): Update.
23076 (profile_probability::dump): New member function.
23077 (profile_probability::debug): New member function.
23078 (profile_probability::differs_from_p): New member function.
23079 (profile_probability::differs_lot_from_p): New member function.
23080 (profile_probability::stream_in): New member function.
23081 (profile_probability::stream_out): New member function.
23082 * profile-count.h (profile_count_quality): Rename to ...
23083 (profile_quality): ... this one.
23084 (profile_probability): New.
23085 (profile_count): Update.
23086 * profile.c (compute_branch_probabilities): Update.
23087 * recog.c (peep2_attempt): Update.
23088 * sched-ebb.c (schedule_ebbs): Update.
23089 * sched-rgn.c (find_single_block_region): Update.
23090 (compute_dom_prob_ps): Update.
23091 (schedule_region): Update.
23092 * sel-sched-ir.c (compute_succs_info): Update.
23093 * stmt.c (struct case_node): Update.
23094 (do_jump_if_equal): Update.
23095 (get_outgoing_edge_probs): Update.
23096 (conditional_probability): Update.
23097 (emit_case_dispatch_table): Update.
23098 (expand_case): Update.
23099 (expand_sjlj_dispatch_table): Update.
23100 (emit_case_nodes): Update.
23101 * targhooks.c: Update.
23102 * tracer.c (better_p): Update.
23103 (find_best_successor): Update.
23104 * trans-mem.c (expand_transaction): Update.
23105 * tree-call-cdce.c: Update.
23106 * tree-cfg.c (gimple_split_edge): Upate.
23107 (move_sese_region_to_fn): Upate.
23108 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23109 * tree-eh.c (lower_resx): Upate.
23110 (cleanup_empty_eh_move_lp): Upate.
23111 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23112 * tree-inline.c (copy_edges_for_bb): Update.
23113 (copy_cfg_body): Update.
23114 * tree-parloops.c (gen_parallel_loop): Update.
23115 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23116 (gimple_gen_time_profiler): Update.
23117 * tree-ssa-dce.c (remove_dead_stmt): Update.
23118 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23119 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23120 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23121 (unloop_loops): Update.
23122 (try_peel_loop): Update.
23123 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23124 * tree-ssa-loop-split.c (connect_loops): Update.
23125 (split_loop): Update.
23126 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23127 (hoist_guard): Update.
23128 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23129 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23130 (value_replacement): Update.
23131 * tree-ssa-reassoc.c (branch_fixup): Update.
23132 * tree-ssa-tail-merge.c (replace_block_by): Update.
23133 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23134 (create_edge_and_update_destination_phis): Update.
23135 (compute_path_counts): Update.
23136 (recompute_probabilities): Update.
23137 (update_joiner_offpath_counts): Update.
23138 (freqs_to_counts_path): Update.
23139 (duplicate_thread_path): Update.
23140 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23141 (struct switch_conv_info): Update.
23142 (gen_inbound_check): Update.
23143 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23144 (vect_do_peeling): Update.
23145 (vect_loop_versioning): Update.
23146 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23147 (optimize_mask_stores): Update.
23148 * ubsan.c (ubsan_expand_null_ifn): Update.
23149 * value-prof.c (gimple_divmod_fixed_value): Update.
23150 (gimple_divmod_fixed_value_transform): Update.
23151 (gimple_mod_pow2): Update.
23152 (gimple_mod_pow2_value_transform): Update.
23153 (gimple_mod_subtract): Update.
23154 (gimple_mod_subtract_transform): Update.
23155 (gimple_ic): Update.
23156 (gimple_stringop_fixed_value): Update.
23157 (gimple_stringops_transform): Update.
23158 * value-prof.h: Update.
23160 2017-06-29 Carl Love <cel@us.ibm.com>
23162 * config/rs6000/rs6000-c.c: Add support for built-in functions
23163 vector signed int vec_signed (vector float);
23164 vector signed long long vec_signed (vector double);
23165 vector signed int vec_signed2 (vector double, vector double);
23166 vector signed int vec_signede (vector double);
23167 vector signed int vec_signedo (vector double);
23168 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23169 instruction generator.
23170 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23171 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23172 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23174 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23175 vunsignede_v2df): Add define_expands.
23176 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23177 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23178 VEC_UNSIGNEDO): Add definitions.
23179 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23180 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23181 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23182 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23183 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23184 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23185 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23186 vec_unsignede, vec_unsignedo): Add builtin defines.
23187 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23189 * doc/extend.texi: Update the built-in documentation file for the
23190 new built-in functions.
23192 2017-06-29 Richard Biener <rguenther@suse.de>
23194 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23195 reduction chains to LOOP_VINFO_REDUCTIONS.
23196 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23197 SLP reductions after processing reduction chains.
23199 2017-06-29 Nathan Sidwell <nathan@acm.org>
23201 * builtins.c (fold_builtin_FUNCTION): Use
23202 lang_hooks.decl_printable_name.
23204 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23206 PR middle-end/81194
23207 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23208 with only one label.
23209 * stmt.c (expand_case): Assert NCASES is greater than one.
23211 2017-06-29 Richard Biener <rguenther@suse.de>
23213 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23215 (group_case_labels): Likewise.
23216 (find_taken_edge): Push sanity checking on val to workers...
23217 (find_taken_edge_cond_expr): ... here
23218 (find_taken_edge_switch_expr): ... and here, handle cases
23219 with just a default label.
23220 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23221 (group_case_labels): Likewise.
23222 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23223 group_case_labels does anything cleanup the CFG again.
23225 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23227 PR tree-optimization/81196
23228 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23229 exit condition comparing two IVs.
23231 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23233 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23234 profile to the dummy entry at the end of the list of architectures.
23235 * config/arm/arm-cpu-cdata.h: Regenerated.
23237 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23238 Michael Collison <michael.collison@arm.com>
23241 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23243 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23244 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23245 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23246 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23247 of shift when the shift amount is masked with constant equal to
23248 the size of the mode.
23249 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23252 2017-06-29 Martin Liska <mliska@suse.cz>
23254 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23256 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23259 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23262 (edge_growth_cache_entry::edge_growth_cache_entry): New
23264 (reset_edge_growth_cache): Update to use constructor.
23266 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23268 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23269 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23270 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23272 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23274 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23275 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23277 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23279 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23280 (*-linux-uclibc*): Add t-uclibc tmake_file.
23281 * config/t-musl: New.
23282 * config/t-uclibc: New.
23284 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23286 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23288 (gen_comm_data): Emit architectural setting of arch_prof.
23289 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23291 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23292 (armv8-m.base, armv8-m.main): Likewise.
23293 * arm-protos.h (arm_build_target): Add profile field.
23294 (arch_option): Likewise.
23295 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23297 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23298 arm_active_target.profile.
23300 2017-06-28 Richard Biener <rguenther@suse.de>
23302 PR middle-end/81227
23303 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23304 TYPE_OVERFLOW_WRAPS.
23305 * match.pd (negate_expr_p): Likewise.
23306 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23307 fold_build2, not fold_binary.
23309 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23311 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23312 Convert memory address to Pmode.
23313 (aarch64_print_operand): Assert MEM operands are always Pmode.
23315 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23318 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23319 Remove redundant if.
23320 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23321 * config/arm/aarch-common-protos.h
23322 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23323 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23325 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23328 * multiple_target.c (create_dispatcher_calls): Set the default
23329 clone to be static, not public.
23331 2017-06-28 Richard Biener <rguenther@suse.de>
23333 * tree-vect-loop.c (vectorizable_reduction): Move special
23334 cond reduction IV var creation ...
23335 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23336 parameter. Use STMT_VINFO_VECTYPE.
23337 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23340 2017-06-28 Martin Liska <mliska@suse.cz>
23343 * ipa-visibility.c (non_local_p): Handle visibility.
23345 2017-06-28 Martin Liska <mliska@suse.cz>
23348 * common.opt: Add IntegerRange to various options.
23349 * opt-functions.awk (integer_range_info): New function.
23350 * optc-gen.awk: Add integer_range_info to cl_options struct.
23351 * opts-common.c (decode_cmdline_option): Handle
23352 CL_ERR_INT_RANGE_ARG.
23353 (cmdline_handle_error): Likewise.
23354 * opts.c (print_filtered_help): Show valid interval in
23355 when --help is provided.
23356 * opts.h (struct cl_option): Add range_min and range_max fields.
23357 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23359 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23361 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23362 (x * C EQ/NE y * C): New transformation.
23364 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23366 * genmultilib (combination_space): Accept '+' in option names.
23368 2017-06-28 Martin Liska <mliska@suse.cz>
23371 * asan.c (instrument_derefs): Bail out inner references
23372 that are hard register variables.
23374 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23377 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23378 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23380 2017-06-28 Richard Biener <rguenther@suse.de>
23382 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23383 (vect_get_slp_defs): Adjust.
23384 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23386 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23388 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23389 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23390 (vectorizable_reduction): Adjust.
23391 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23393 (vect_get_slp_defs): Likewise.
23394 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23395 (vectorizable_bswap): Adjust.
23396 (vectorizable_call): Likewise.
23397 (vectorizable_conversion): Likewise.
23398 (vectorizable_assignment): Likewise.
23399 (vectorizable_shift): Likewise.
23400 (vectorizable_operation): Likewise.
23401 (vectorizable_store): Likewise.
23402 (vectorizable_condition): Likewise.
23403 (vectorizable_comparison): Likewise.
23405 2017-06-28 Michael Collison <michael.collison@arm.com>
23408 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23410 (arm_gen_movmemqi): Removed unused variable 'i'.
23411 Convert 'for' loop into 'while' loop.
23412 (arm_expand_prologue): Remove last unnecessary set of insn.
23413 (thumb_pop): Remove unused variable 'pushed_words'.
23414 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23416 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23418 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23419 * config/s390/s390-protos.h: Add prototype of
23420 s390_rel_address_ok_p.
23421 * config/s390/s390.c (s390_got_symbol): New function.
23422 (s390_rel_address_ok_p): New function.
23423 (legitimize_pic_address): Use s390_rel_address_ok_p.
23424 (s390_load_got): Use s390_got_symbol.
23425 (s390_option_override): Issue error if
23426 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23427 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23429 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23431 2017-06-27 Andrew Pinski <apinski@cavium.com>
23433 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23434 (X * copysign (1.0, X)): New pattern.
23435 (X * copysign (1.0, -X)): New pattern.
23436 (copysign (-1.0, CST)): New pattern.
23438 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23440 * genmultilib (combination_space): Remove variable.
23441 Validate reuse rules against regular expression for any sequence
23442 of multilib options in any order.
23444 2017-06-27 Michael Collison <michael.collison@arm.com>
23446 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23447 call aarch64_split_simd_combine.
23448 * (aarch64_combine_internal<mode>): Delete pattern.
23449 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23450 Allow register and subreg operands.
23452 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23454 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23455 specific need, just fallback on defaults.
23456 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23458 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23459 Olivier Hainque <hainque@adacore.com>
23461 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23463 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23464 targets. Pick a default if no particular attempt applied.
23465 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23468 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23470 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23471 (x86_64-wrs-vxworks7): Likewise.
23473 2017-06-27 Marek Polacek <polacek@redhat.com>
23476 * ubsan.c (instrument_null): Check get_base_address's result for null.
23478 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23480 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23482 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23484 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23485 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23486 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23487 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23488 New function types.
23489 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23490 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23491 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23492 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23493 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23494 BUILT_IN_FEUPDATEENV): New builtins.
23495 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23496 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23497 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23498 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23500 (builtin_structptr_types): Adjust size.
23501 * tree.c (builtin_structptr_types): Add four entries.
23503 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23504 Olivier Hainque <hainque@adacore.com>
23506 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23507 (TLS_SYM): New local macro, forcing reference to __tls__ on
23508 link command lines for VxWorks 7 RTPs, triggering initialization
23510 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23511 OS features TLS support, true for RTPs on VxWorks 7.
23512 * config/vxworks.c (vxworks_override_options): Setup emutls
23515 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23517 * predict.c (test_prediction_value_range): Use -1U instead of -1
23518 to avoid narrowing conversion warning.
23519 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23520 to avoid narrowing conversion warning.
23521 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23523 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23525 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23527 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23528 64bit configurations.
23529 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23530 (SIZE_TYPE): Likewise.
23531 * config/vxworks.c (vxworks_emutls_var_fields): Use
23532 long_unsigned_type_node instead of unsigned_type_node as the offset
23533 field type, which is "pointer" mode in emutls.c.
23535 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23538 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23540 PR middle-end/81207
23541 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23542 gimple_vuse copying separately from gimple_vdef copying.
23544 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23546 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23549 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23550 Olivier Hainque <hainque@adacore.com>
23552 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23553 for all vxworks7 targets.
23554 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23555 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23556 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23557 variations for VX6/VX7 and 32/64bits later on in ...
23558 (VXWORKS_LIB_SPEC): Leverage new macros.
23559 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23560 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23562 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23564 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23565 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23567 2017-06-26 Carl Love <cel@us.ibm.com>
23569 * config/rs6000/rs6000-c.c: Add support for built-in functions
23570 vector bool char vec_reve (vector bool char);
23571 vector signed char vec_reve (vector signed char);
23572 vector unsigned char vec_reve (vector unsigned char);
23573 vector bool int vec_reve (vector bool int);
23574 vector signed int vec_reve (vector signed int);
23575 vector unsigned int vec_reve (vector unsigned int);
23576 vector bool long long vec_reve (vector bool long long);
23577 vector signed long long vec_reve (vector signed long long);
23578 vector unsigned long long vec_reve (vector unsigned long long);
23579 vector bool short vec_reve (vector bool short);
23580 vector signed short vec_reve (vector signed short);
23581 vector double vec_reve (vector double);
23582 vector float vec_reve (vector float);
23583 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23584 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23585 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23586 (altivec_vreve): New pattern.
23587 * config/rs6000/altivec.h (vec_reve): New define.
23588 * doc/extend.texi (vec_rev): Update the built-in documentation file
23589 for the new built-in functions.
23591 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23593 PR tree-optimization/71815
23594 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23596 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23598 (slsr_process_phi): Likewise.
23599 (replace_uncond_cands_and_profitable_phis): Don't replace a
23600 multiply candidate with a stride of 1 (copy or cast).
23601 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23603 (lowest_cost_path): Likewise.
23604 (total_savings): Likewise.
23606 2017-06-26 Richard Biener <rguenther@suse.de>
23609 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23610 Use def_builtin_pure for all gather builtins.
23612 2017-06-26 Richard Biener <rguenther@suse.de>
23614 PR tree-optimization/81203
23615 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23616 non-dominating BBs.
23618 2017-06-26 Marek Polacek <polacek@redhat.com>
23621 * doc/invoke.texi: Document -Wmultistatement-macros.
23623 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23625 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23626 arm_neon_ok_no_float_abi effective target.
23628 2017-06-26 Richard Biener <rguenther@suse.de>
23630 PR tree-optimization/80928
23631 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23632 (copy_bbs): Set BB_DUPLICATED flag early.
23633 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23635 (execute_on_shrinking_pred): Likewise.
23636 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23637 BB_DUPLICATED blocks.
23638 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23639 iterate over all PHIs considering removal of *gsi.
23641 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23643 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23646 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23648 * config/rs6000/rs6000-string.c: (expand_block_clear,
23649 do_load_for_compare, select_block_compare_mode,
23650 compute_current_alignment, expand_block_compare,
23651 expand_strncmp_align_check, expand_strn_compare,
23652 expand_block_move, rs6000_output_load_multiple)
23653 Move functions related to string/block move/compare
23654 to a separate file.
23655 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23656 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23657 for this function which is now used in two files.
23658 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23659 * config.gcc: Add rs6000-string.o to extra_objs for
23660 targets powerpc*-*-* and rs6000*-*-*.
23662 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23665 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23666 32-bit, since indexed is not valid for DImode.
23667 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23668 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23669 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23670 (define_peephole2 for Altivec d-form store): Likewise.
23673 * multiple_target.c (create_dispatcher_calls): Only create the
23674 dispatcher call if the function is the default clone of a
23675 versioned function.
23677 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23679 PR middle-end/80902
23680 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23681 a call, force the call to not be a tail call.
23683 2017-06-23 Jeff Law <law@redhat.com>
23685 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23688 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23690 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23691 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23692 handling for early expansion of vector shifts (sl,sr,sra,rl).
23693 (builtin_function_type): Add vector shift right instructions
23694 to the unsigned argument list.
23696 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23698 rtl-optimizatoin/79286
23699 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23700 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23701 trap. PIC register plus a const unspec without offset can never trap.
23703 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23705 * tree.h (builtin_structptr_type): New type.
23706 (builtin_structptr_types): Declare new array.
23707 * tree.c (builtin_structptr_types): New array.
23708 (free_lang_data, build_common_tree_nodes): Use it.
23710 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23713 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23716 2017-06-22 Matt Turner <mattst88@gmail.com>
23718 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23719 Lake models to skylake case. Assume skylake for unknown
23720 models with clflushopt.
23722 2017-06-22 Jeff Law <law@redhat.com>
23724 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23725 frame sizes that do not satisfy aarch64_uimm12_shift.
23727 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23729 * profile-count.h (apply_probability,
23730 apply_scale, probability_in): Fix checks for zero.
23732 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23734 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23735 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23737 2016-06-22 Richard Biener <rguenther@suse.de>
23739 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23740 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23741 REDUC_MAX_EXPR support.
23742 (vectorizable_reduction): Likewise.
23743 (vect_create_epilog_for_reduction): Likewise.
23745 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23747 * match.pd (A / (1 << B) -> A >> B): New.
23748 * generic-match-head.c: Include optabs-tree.h.
23749 * gimple-match-head.c: Likewise.
23750 * optabs-tree.h (target_supports_op_p): New.
23751 * optabs-tree.c (target_supports_op_p): New.
23753 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23755 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23756 $gcc_cv_ld --help output.
23757 (gcc_cv_ld_demangle): Likewise.
23758 (gcc_cv_ld_eh_frame_hdr): Likewise.
23759 (gcc_cv_ld_pie): Likewise.
23760 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23761 (gcc_cv_ld_buildid): Likewise.
23762 (gcc_cv_ld_sysroot): Likewise.
23763 (ld_bndplt_support): Likewise.
23764 (ld_pushpopstate_support): Likewise.
23765 * configure: Regenerate.
23766 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23768 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23771 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23772 operands indexes to avoid gap between operands and match_dups.
23774 2017-06-21 Andrew Pinski <apinski@cavium.com>
23776 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23777 Increment Arith_shift and Arith_shift_reg by 1.
23778 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23780 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23781 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23782 (aarch64_strip_extend): Add new argument and test for it.
23783 (aarch64_cheap_mult_shift_p): New function.
23784 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23785 add a cost if it is true.
23786 Update calls to aarch64_strip_extend.
23787 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23789 2017-06-21 Andrew Pinski <apinski@cavium.com>
23791 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23793 (thunderxt88): Likewise.
23794 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23795 (thunderx_prefetch_tune): New variable.
23796 (thunderx2t99_prefetch_tune): Update for the correct values.
23797 (thunderxt88_tunings): New variable.
23798 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23799 generic_prefetch_tune.
23800 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23802 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23804 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23805 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23806 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23807 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23808 (aarch64_atomic_cas<mode>, GPI): Likewise.
23810 2017-06-21 Martin Liska <mliska@suse.cz>
23812 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23813 statements on cold and hot labels.
23814 * predict.c (tree_estimate_probability_bb): Remove the
23815 prediction from this place.
23817 2017-06-21 Martin Liska <mliska@suse.cz>
23819 PR tree-optimization/79489
23820 * gimplify.c (maybe_add_early_return_predict_stmt): New
23822 (gimplify_return_expr): Call the function.
23823 * predict.c (tree_estimate_probability_bb): Remove handling
23825 * predict.def: Update comment about early return predictor.
23826 * gimple-predict.h (is_gimple_predict): New function.
23827 * predict.def: Change default value of early return to 66.
23828 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23830 * passes.def: Put pass_strip_predict_hints to the beginning of
23833 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23835 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23836 FUNCTION_DECL declarations.
23837 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23839 (dwaf2out_decl): Likewise.
23840 * godump.c (go_early_global_decl): Skip call to the real debug hook
23841 for FUNCTION_DECL declarations.
23842 * passes.c (rest_of_decl_compilation): Skip call to the
23843 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23844 -fdump-go-spec is passed.
23846 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23848 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23849 Reorder for compactness.
23850 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23851 (def_builtin_pure, def_builtin_pure2): New functions.
23852 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23854 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23856 * match.pd (nop_convert): New predicate.
23857 ((A +- CST1) +- CST2): Allow some NOP conversions.
23859 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23862 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23863 with ctors/dtors if GOVD_SHARED is set.
23865 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23867 * config/aarch64/aarch64.md (movti_aarch64):
23868 Emit mov rather than orr.
23869 (movtf_aarch64): Likewise.
23870 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23871 Emit mov rather than orr.
23873 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23875 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23876 Swap alternatives, make integer dup more expensive.
23878 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23880 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23881 Return true for non-tls symbols.
23883 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23885 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23886 (cortex-a75): Likewise.
23887 (cortex-a75.cortex-a55): Likewise.
23888 * config/aarch64/aarch64-tune.md: Regenerate.
23889 * doc/invoke.texi (-mtune): Document new values for -mtune.
23891 2017-06-21 Tom de Vries <tom@codesourcery.com>
23893 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23894 stack_size feature.
23895 (Effective-Target Keywords, Other attributes): Suggest using
23896 dg-add-options stack_size feature to get stack limit in stack_size
23897 effective target documentation.
23899 2017-06-21 Julian Brown <julian@codesourcery.com>
23900 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23902 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23903 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23904 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23906 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23907 attribute type list for neon_multiply.
23908 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23909 attribute type list for neon_multiply.
23910 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23911 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23912 attribute type list for neon_multiply.
23913 * config/arm/types.md (crypto_pmull): Add.
23914 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23915 attribute type list.
23917 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23919 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23922 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23924 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23925 to make sure not to dereference a NULL cost_classes_ptr pointer.
23927 2017-06-20 Carl Love <cel@us.ibm.com>
23929 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23930 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23931 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23932 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23933 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23934 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23935 VMULOSW): New enum "unspec" values.
23936 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23937 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23938 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23939 altivec_vmulosw): New patterns.
23940 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23941 VMULOSW): Add definitions.
23943 2017-06-20 Julia Koval <julia.koval@intel.com>
23945 * config/i386/i386.c: Fix rounding expand for new pattern.
23946 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23948 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23950 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23951 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23953 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23955 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23958 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23960 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23961 architecture, then by implementer ID.
23962 * config/aarch64/aarch64-tune.md: Regenerate.
23964 2017-06-20 Richard Biener <rguenther@suse.de>
23966 PR middle-end/81097
23967 * fold-const.c (split_tree): Fold to type before negating.
23969 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23971 * diagnostic-show-locus.c
23972 (selftest::test_fixit_deletion_affecting_newline): New function.
23973 (selftest::diagnostic_show_locus_c_tests): Call it.
23975 2017-06-20 Andreas Schwab <schwab@suse.de>
23978 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23981 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23983 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23984 __ARM_FEATURE_COPROC according to support.
23986 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23988 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23989 Rewritten to avoid overflow for > 32-bit pointers.
23992 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23993 by removing enum keyword.
23994 (ubsan_type_descriptor): Likewise. Formatting fix.
23997 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23998 splitter): Require TARGET_SSE2 in the condition.
24000 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24003 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24004 for doing vector set of SFmode on ISA 3.0.
24005 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24006 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24008 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24009 SFmode value into a V4SF variable that was extracted from another
24010 V4SF variable without converting the element to double precision
24011 and back to single precision vector format.
24012 (vsx_insert_extract_v4sf_p9_2): Likewise.
24014 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24016 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24017 in UWHI to avoid undefined overflow.
24020 * ubsan.h (enum ubsan_encode_value_phase): New.
24021 (ubsan_encode_value): Change second argument to
24022 enum ubsan_encode_value_phase with default value of
24023 UBSAN_ENCODE_VALUE_GENERIC.
24024 * ubsan.c (ubsan_encode_value): Change second argument to
24025 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24026 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24027 create_tmp_var_raw instead of create_tmp_var and use a
24029 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24030 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24031 ubsan_encode_value callers.
24034 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24035 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24036 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24038 2017-06-19 Richard Biener <rguenther@suse.de>
24040 PR middle-end/81118
24041 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24042 estimates if we changed anything.
24044 2017-06-19 Richard Biener <rguenther@suse.de>
24046 PR tree-optimization/80887
24047 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24048 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24049 simplified lookups, then reset mprts_hook.
24050 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24052 (try_to_simplify): Likewise.
24054 2017-06-19 Martin Liska <mliska@suse.cz>
24057 * gimplify.c (gimplify_switch_expr):
24058 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24060 2017-06-19 Martin Liska <mliska@suse.cz>
24062 * doc/install.texi: Document that PGO runs in 4 stages.
24064 2017-06-19 Martin Liska <mliska@suse.cz>
24067 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24068 to dispatcher function name.
24069 * multiple_target.c (replace_function_decl): New function.
24070 (create_dispatcher_calls): Redirect both edges and references.
24072 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24074 * profile-count.c (profile_count::dump): Dump quality.
24075 (profile_count::differs_from_p): Update for unsigned val.
24076 * profile-count.h (profile_count_quality): New enum.
24077 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24079 2017-06-19 Richard Biener <rguenther@suse.de>
24081 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24082 struct function as arg.
24083 (estimate_numbers_of_iterations): Export overload with loop arg.
24084 (free_numbers_of_iterations_estimates_loop): Use an overload of
24085 free_numbers_of_iterations_estimates instead.
24086 * tree-cfg.c (remove_bb): Adjust.
24087 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24088 * tree-parloops.c (gen_parallel_loop): Likewise.
24089 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24091 (tree_unroll_loops_completely): Likewise.
24092 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24093 Use an overload instead and export.
24094 (estimated_loop_iterations): Adjust.
24095 (max_loop_iterations): Likewise.
24096 (likely_max_loop_iterations): Likewise.
24097 (estimate_numbers_of_iterations): Take struct function as arg
24099 (loop_exits_before_overflow): Adjust.
24100 (free_numbers_of_iterations_estimates_loop): Use an overload.
24101 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24102 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24104 2017-06-19 Richard Biener <rguenther@suse.de>
24107 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24108 RANGE_EXPR conservatively.
24110 2017-06-16 Carl Love <cel@us.ibm.com>
24112 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24113 definitions for vec_float, vec_float2, vec_floato,
24114 vec_floate built-ins.
24115 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24116 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24118 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24119 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24120 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24121 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24122 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24123 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24124 vec_floato): Add builtin defines.
24125 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24126 Update the built-in documentation file for the new built-in
24129 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24131 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24132 (mthumb): Mark as the negative of -marm.
24134 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24136 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24138 (ARM Options, -mtune): Document that this accepts the same
24139 extension options as -mcpu.
24140 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24144 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24145 permitted extensions.
24147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24149 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24150 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24151 (armv8-m.main): Add option +nodsp.
24152 * config/arm/arm-cpu-cdata.h: Regenerated.
24154 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24156 * config/arm/t-fuchsia: New file.
24157 * config.gcc (arm*-*-fuchsia*): Use it.
24159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24161 * config/arm/t-symbian: Rewrite for new option infrastructure.
24163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24165 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24166 (MULTILIB_REQUIRED): Likewise.
24168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24170 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24171 (MULTILIB_RESUE): Likewise.
24172 (MULTILIB_MATCHES): Likewise.
24173 (MULTLIB_REQUIRED): Likewise.
24175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24177 * config/arm/t-rtems: Rewrite for new option framework.
24179 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24181 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24182 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24183 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24184 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24185 * config/arm/t-multilib: ... here.
24186 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24187 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24188 armv7-a and armv8*-a when A-profile libraries have not been built.
24189 * config/arm/t-rmprofile: Rewrite.
24191 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24193 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24194 with a backslash. Remove the backslash after substituting unescaped
24196 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24198 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24200 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24201 the driver to pass through the default setting of -mfloat-abi.
24202 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24204 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24205 (all_feat_combs): New rule.
24206 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24208 * config/arm/t-aprofile: Rewrite.
24210 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24212 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24213 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24214 fpu is not specified by the user/command-line.
24215 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24216 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24217 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24218 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24219 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24220 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24222 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24224 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24225 * config/arm/t-arm-elf: Rewritten.
24227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24229 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24230 have some floating-point instructions.
24231 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24232 (TARGET_MAYBE_HARD_FLOAT): New macro.
24233 * config/arm/arm-builtins.c (arm_init_builtins): Use
24234 TARGET_MAYBE_HARD_FLOAT.
24235 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24237 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24239 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24240 (configargs.h): Include it.
24241 (arm_print_hint_for_fpu_option): New function.
24242 (arm_parse_fpu_option): New function.
24243 (candidate_extension): New class.
24244 (arm_canon_for_multilib): New function.
24245 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24246 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24247 (ARCH_CANONICAL_SPECS): New macro.
24248 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24250 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24252 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24253 are set after handling multilib fragments. Set target_cpu_default2
24256 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24258 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24260 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24261 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24263 (check_arch): Likewise.
24264 * config/arm/arm.c (arm_configure_build_target): Handle
24265 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24266 options in the default.
24268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24270 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24271 when an option is an alias of another.
24272 * config/arm/parsecpu.awk (optalias): New parser token.
24273 (gen_comm_data): Mark non-alias options as such. Emit entries
24274 for extension aliases.
24275 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24276 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24277 (armv6kz, armv6zk, armv6t2): Likewise.
24278 (armv7): Make vfpv3-d16 an alias.
24279 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24281 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24282 Sort in canonical order.
24283 (armv8-a): Sort in canonical order.
24284 (armv8.1-a, armv8.2-a): Likewise.
24285 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24287 (cortex-a9): Sort in canonical order.
24288 * config/arm/arm.c (selftests.h): Include it.
24289 (arm_test_cpu_arch_data): New function.
24290 (arm_run_self_tests): New function.
24291 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24292 (targetm): Move declaration to the end of the file.
24293 * arm-cpu-cdata.h: Regenerated.
24295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24297 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24298 call to target_mode_check describing the type of option passed.
24299 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24300 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24301 arm_parse_cpu_option_name to match parameters against list of
24303 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24304 arm_arch_core_flags data structure.
24305 * config/arm/arm-cpu_cdata.h: Regenerated.
24307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24309 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24311 (arm_print_hint_for_cpu_option): Likewise.
24312 (arm_print_hint_for_arch_option): Likewise.
24313 (arm_parse_cpu_option_name): Likewise.
24314 (arm_parse_arch_option_name): Likewise.
24315 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24316 of entries in the all_fpus list.
24317 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24318 (arm_parse_cpu_option_name): Declare.
24319 (arm_parse_arch_option_name): Declare.
24320 (arm_parse_option_features): Declare.
24321 (arm_intialize_isa): Declare.
24322 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24324 (gen_comm_data): ... here. Make definitions non-static.
24325 * config/arm/arm-cpu-data.h: Regenerated.
24326 * config/arm/arm-cpu-cdata.h: Regenerated.
24328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24330 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24331 (cpu_arch_extension): New structure.
24332 (cpu_arch_option, arch_option, cpu_option): New structures.
24333 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24334 architecture types.
24335 (gen_data): Generate new format data tables.
24336 * config/arm/arm.c (cpu_tune): New structure.
24337 (cpu_option, processors): Delete.
24338 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24339 (arm_print_hint_for_cpu_option): ... this and ...
24340 (arm_print_hint_for_arch_option): ... this.
24341 (arm_parse_arch_cpu_name): Delete. Replace with ...
24342 (arm_parse_cpu_option_name): ... this and ...
24343 (arm_parse_arch_option_name): ... this.
24344 (arm_unrecognized_feature): Change type of target parameter to
24346 (arm_parse_arch_cpu_features): Delete. Replace with ...
24347 (arm_parse_option_features): ... this.
24348 (arm_configure_build_target): Rework to use new configuration data
24350 (arm_print_tune_info): Rework for new configuration data tables.
24351 * config/arm/arm-cpu-data.h: Regenerated.
24352 * config/arm/arm-cpu.h: Regenerated.
24354 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24356 * Makefile.in (OBJS): Move sbitmap.o from here ...
24357 (OBJS-libcommon): ... to here.
24359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24361 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24362 (ISA_ALL_CRYPTO): New macro.
24363 (ISA_ALL_SIMD): New macro
24364 (ISA_ALL_FP): New macro.
24365 * config/arm/arm.c (fpu_bitlist): Update initializer.
24366 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24368 (arm9e): Add fpu. Add option for nofp
24369 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24370 (arm926ej-s, arm1026ej-s): Likewise.
24371 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24372 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24373 neon-fp16, neon-vfpv4, nofp and nosimd.
24374 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24375 (cortex-a8): Add fpu. Add option for nofp.
24376 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24377 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24378 (cortex-r4f): Add fpu.
24379 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24380 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24382 (cortex-r8): Likewise.
24383 (cortex-m4): Add fpu. Add option for nofp.
24384 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24385 (cortex-a17.cortex-a7): Likewise.
24386 (cortex-a32): Add fpu. Add options for crypto and nofp.
24387 (cortex-a35, cortex-a53): Likewise.
24388 (cortex-a57): Add fpu. Add option for crypto.
24389 (cortex-a72, cortex-a73): Likewise.
24390 (exynos-m1): Likewise.
24391 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24392 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24393 (cortex-m33): Add fpu. Add option for nofp.
24394 * config/arm/arm-cpu-cdata.h: Regenerated
24395 * config/arm/arm-cpu-data.h: Regenerated.
24397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24399 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24400 (armv5te, armv5tej): Likewise.
24401 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24402 (armv7): Add options fp and vfpv3-d16.
24403 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24404 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24406 (armv7ve): Likewise.
24407 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24408 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24409 (armv8-a): Add nocrypto option.
24410 (armv8.1-a, armv8.2-a): Likewise.
24411 (armv8-m.main): add options fp, fp.dp and nofp.
24413 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24415 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24417 (armv8-a+crc): Delete.
24418 (armv8.1-a): Add options simd, crypto and nofp.
24419 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24420 (armv8.2-a+fp16): Delete.
24421 (armv8-m.main): Add option dsp.
24422 (armv8-m.main+dsp): Delete.
24423 (cortex-a8): Add fpu. Add nofp option.
24424 (cortex-a9): Add fpu. Add nofp and nosimd options.
24425 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24426 link to main cpu and architecture data structures.
24427 (gen_comm_data): Only put isa attributes from the main architecture
24429 (option): New statement for architecture and CPU entries.
24430 * arm.c (struct cpu_option): New structure.
24431 (struct processors): Add entry for options.
24432 (arm_unrecognized_feature): New function.
24433 (arm_parse_arch_cpu_name): Ignore any characters after the first
24435 (arm_parse_arch_cpu_feature): New function.
24436 (arm_configure_build_target): Separate out any CPU and architecture
24437 features and parse separately. Don't error out if -mfpu=auto is
24438 used with only an architecture string.
24439 (arm_print_asm_arch_directives): New function.
24440 (arm_file_start): Call it.
24441 * config/arm/arm-cpu-cdata.h: Regenerated.
24442 * config/arm/arm-cpu-data.h: Likewise.
24443 * config/arm/arm-tables.opt: Likewise.
24445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24447 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24448 assembler when it is not -mfpu=auto.
24450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24452 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24453 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24454 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24455 (ASM_CPU_SPEC): Rewrite.
24456 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24457 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24458 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24459 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24460 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24461 copied string is NUL-terminated. Also strip any characters prefixed
24463 (arm_rewrite_selected_arch): New function.
24464 (arm_rewrite_march): New function.
24466 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24468 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24469 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24470 (march, mcpu, mtune): Convert to string-based options.
24471 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24472 (arm_parse_arch_cpu_name): New function.
24473 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24474 identify selected architecture or CPU.
24475 (arm_option_save): New function.
24476 (TARGET_OPTION_SAVE): Redefine.
24477 (arm_option_restore): Restore string options.
24478 (arm_option_print): Print string options.
24480 2017-06-16 Martin Sebor <msebor@redhat.com>
24482 PR tree-optimization/80933
24483 PR tree-optimization/80934
24484 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24485 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24486 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24487 (gimple_fold_builtin): Call them.
24489 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24491 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24492 as unlikely; update profile.
24494 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24496 * predict.c (force_edge_cold): Handle declaring edges impossible
24499 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24501 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24503 (try_unroll_loop_completely): Fix reporting.
24505 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24507 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24509 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24512 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24513 if given a non-constant argument for an intrinsic which requires a
24516 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24518 * profile.c (compare_freqs): New function.
24519 (branch_prob): Sort edge list.
24520 (find_spanning_tree): Assume that the list is priority sorted.
24522 2017-06-16 Richard Biener <rguenther@suse.de>
24524 PR tree-optimization/81090
24525 * passes.def (pass_record_bounds): Remove.
24526 * tree-pass.h (make_pass_record_bounds): Likewise.
24527 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24528 make_pass_record_bounds): Likewise.
24529 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24530 not free niter estimates at the beginning but at the end.
24531 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24533 2017-06-16 Richard Biener <rguenther@suse.de>
24535 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24536 initializer to workaround ICE in host GCC 4.8.
24538 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24540 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24542 (clone_inlined_nodes): Update.
24544 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24546 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24547 prefetch settings, and enable prefetching by default at -O3.
24549 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24551 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24552 Set flag_prefetch_loop_arrays according to tuning data.
24554 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24556 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24557 New tune structure.
24558 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24559 [Unrelated to main purpose of the patch] Place the pointer field last
24560 to enable type checking errors when tune structure are wrongly merged.
24561 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24562 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24563 (thunderx2t99_prefetch_tune): New tune constants.
24564 (tune_params *_tunings): Update all tunings (no functional change).
24565 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24566 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24567 from tunings structures.
24569 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24572 * ubsan.c (instrument_null): Add T argument, use it instead
24573 of computing it based on IS_LHS.
24574 (instrument_object_size): Likewise.
24575 (pass_ubsan::execute): Adjust instrument_null and
24576 instrument_object_size callers to pass gimple_get_lhs or
24577 gimple_assign_rhs1 result to it. Use instrument_null instead of
24578 calling get_base_address and instrument_mem_ref. Handle
24579 aggregate call arguments for object-size sanitization.
24581 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24583 PR tree-optimization/81089
24584 * tree-vrp.c (is_masked_range_test): Validate operands of
24587 2017-06-15 Martin Sebor <msebor@redhat.com>
24590 * dumpfile.c (dump_register): Avoid calling memset to initialize
24591 a class with a default ctor.
24592 * gcc.c (struct compiler): Remove const qualification.
24593 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24594 * hash-table.h: Ditto.
24595 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24597 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24598 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24600 * params.h (struct param_info): Make struct members non-const.
24601 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24602 with default initialization.
24603 * vec.h (vec_copy_construct, vec_default_construct): New helper
24605 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24606 with vec_copy_construct.
24607 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24608 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24609 * doc/invoke.texi (-Wclass-memaccess): Document.
24611 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24613 * emit-rtl.h (is_leaf): Update comment about local
24614 register allocator.
24616 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24619 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24620 for a variable to have a section before checking if the section has a
24622 Set section to.persistent if persistent attribute is set.
24623 Warn if .persistent attribute is used on an automatic variable.
24625 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24627 PR rtl-optimization/80474
24628 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24630 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24632 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24635 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24637 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24638 (casesi): Emit bounds checking as RTL.
24639 (casesi_internal_mips16_<mode>): Remove bounds checking.
24641 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24643 * config/xtensa/xtensa.c (xtensa_option_override): Append
24644 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24645 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24646 xtensa_doloop_hooks): Define unconditionally.
24647 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24649 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24650 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24651 for it in xtensa_option_override.
24652 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24653 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24655 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24657 * doc/cppopts.texi: Document '-' special value to -MF.
24659 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24661 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24662 (cortex_a53_fconst): Likewise.
24663 (cortex_a53_fpmul): Likewise.
24664 (cortex_a53_f_load_64): Likewise.
24665 (cortex_a53_f_load_many): Likewise.
24666 (cortex_a53_advsimd_alu): Likewise.
24667 (cortex_a53_advsimd_alu_q): Likewise.
24668 (cortex_a53_advsimd_mul): Likewise.
24669 (cortex_a53_advsimd_mul_q): Likewise.
24670 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24671 Add missing fmul, r2f_cvt and fconst cases.
24673 2017-06-14 Richard Biener <rguenther@suse.de>
24675 PR middle-end/81088
24676 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24678 (fold_binary_loc): When associating do not treat pre-existing
24679 TREE_OVERFLOW on literal constants as a reason to allow
24680 TREE_OVERFLOW on associated literal constants.
24682 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24684 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24685 (MASK_FEATURES): New macro.
24686 * config/sparc/sparc.c (sparc_option_override): Remove the special
24687 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24689 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24691 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24692 a division of 0 if non-call exceptions are enabled.
24694 2017-06-14 Andrew Pinski <apinski@cavium.com>
24695 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24698 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24699 Improve vector initialization code gen for only variable case.
24701 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24703 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24705 2017-06-14 Richard Biener <rguenther@suse.de>
24707 PR tree-optimization/81083
24708 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24711 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24713 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24714 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24715 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24716 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24717 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24718 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24720 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24722 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24723 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24725 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24727 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24729 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24731 * config/rs6000/t-rtems: Don't handle SPE.
24733 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24735 * config/rs6000/t-linux: Don't handle SPE.
24737 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24739 * config/rs6000/eabispe.h: Delete file.
24741 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24743 * config/rs6000/t-spe: Delete file.
24745 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24747 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24748 (rs6000_legitimate_offset_address_p): Return false for anything in
24749 V2SImode or V2SFmode.
24751 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24753 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24754 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24755 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24757 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24758 (rs6000_legitimate_offset_address_p): Ditto.
24759 (rs6000_emit_move): Ditto.
24760 (rs6000_init_builtins): Remove V4HI_type_node.
24762 2017-06-13 Martin Liska <mliska@suse.cz>
24765 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24766 (gate_asan): Likewise.
24767 * asan.h (asan_no_sanitize_address_p): Remove the function.
24768 (sanitize_flags_p): New function.
24769 * builtins.def: Fix coding style.
24770 * common.opt: Use renamed enum value.
24771 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24772 * doc/extend.texi: Document no_sanitize attribute.
24773 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24774 to SANITIZE_UNDEFINED_NONDEFAULT.
24775 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24776 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24777 Use sanitize_flags_p.
24778 * gimplify.c (gimplify_function_tree): Likewise.
24779 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24780 * opts.c (parse_no_sanitize_attribute): New function.
24781 (common_handle_option): Use renamed enum value.
24782 * opts.h (parse_no_sanitize_attribute): Declare.
24783 * tree.c (sanitize_flags_p): New function.
24784 * tree.h: Declared here.
24785 * tsan.c: Use sanitize_flags_p.
24786 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24787 (instrument_mem_ref): Likewise.
24788 (instrument_bool_enum_load): Likewise.
24789 (do_ubsan_in_current_function): Remove the function.
24790 (pass_ubsan::execute): Use sanitize_flags_p.
24791 * ubsan.h: Remove do_ubsan_in_current_function
24792 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24793 (dump_function_to_file): Use it here.
24795 2017-06-13 Martin Jambor <mjambor@suse.cz>
24797 PR tree-optimization/80803
24798 PR tree-optimization/81063
24799 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24800 (propagate_subaccesses_across_link): Enqueue subtree whenever
24801 necessary instead of relying on the caller.
24803 2017-06-13 Martin Jambor <mjambor@suse.cz>
24805 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24806 that have a first_link.
24807 (sort_and_splice_var_accesses): Do not check first_link before
24809 (subtree_mark_written_and_enqueue): Likewise.
24810 (propagate_all_subaccesses): Likewise and do not stop at first
24811 parent with a first_link.
24813 2017-06-13 Martin Jambor <mjambor@suse.cz>
24815 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24818 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24820 * match.pd: New pattern.
24822 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24824 * tree-vrp.c (is_masked_range_test): New function.
24825 (register_edge_assert_for): Determine ranges for
24828 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24830 PR tree-optimization/67328
24831 * fold-const.c (maskable_range_p): New function.
24832 (build_range_check): Generate bittests if possible.
24834 2017-06-13 Martin Liska <mliska@suse.cz>
24836 * gimple-pretty-print.c (dump_probability): Add new argument.
24837 (dump_edge_probability): Dump both probability and count.
24838 (dump_gimple_label): Likewise.
24839 (dump_gimple_bb_header): Likewise.
24841 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24844 * config/avr/avr-devices.c: Fix indentation.
24845 * config/avr/gen-avr-mmcu-specs.c: Dito.
24847 2017-06-13 Richard Biener <rguenther@suse.de>
24849 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24850 instead get vector type from stmt_info.
24851 (vectorizable_reduction): Adjust. Remove dead code.
24853 2017-06-13 Richard Biener <rguenther@suse.de>
24855 PR middle-end/81065
24856 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24857 case of C * (x * C2 + C3).
24858 (fold_addr_of_array_ref_difference): Properly fold index difference.
24860 2017-06-12 David S. Miller <davem@davemloft.net>
24863 * config/sparc/sparc.md (return expander): Emit frame blockage if
24864 function uses alloca.
24866 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24868 * combine.c (make_field_assignment): Check len rather than the mode
24869 precision when calling force_to_mode.
24871 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24873 Support multilibs and devices that see flash in RAM address range.
24876 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24877 (avr_mcu_t) <flash_pm_offset>: New field.
24878 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24879 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24880 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24881 (AVR_TINY_PM_OFFSET): Remove macro.
24882 * config/avr/avr.opt (-mshort-calls): New option.
24883 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24884 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24885 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24886 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24887 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24888 instead of avr_arch->have_jmp_call.
24889 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24890 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24891 avr_arch->flash_pm_offset to define.
24892 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24893 new field flash_pm_offset. Add entry for avrxmega3.
24894 (avr_texinfo): Add entry for avrxmega3.
24895 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24896 attiny212, attiny214,
24897 attiny412, attiny414, attiny416, attiny417,
24898 attiny814, attiny816, attiny817,
24899 attiny1614, attiny1616, attiny1617,
24900 attiny3214, attiny3216, attiny3217.
24901 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24902 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24903 (avr_print_operand_address) [AVR_TINY]: Same.
24904 (avr_asm_init_sections) <readonly_data_section>: Only patch
24905 callback if avr_arch->flash_pm_offset = 0.
24906 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24907 for rodata if avr_arch->flash_pm_offset != 0.
24908 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24909 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24910 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24911 (m_options): Append opt_rcall.
24912 (m_dirnames): Append dir_rcall.
24913 * config/avr/t-multilib: Regenerate.
24915 * configure.ac [target=avr]: Check whether avrxmega3 default
24916 linker description file works as needed.
24917 * configure: Regenerate.
24918 * doc/avr-mmcu.texi: Regenerate.
24919 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24920 <__AVR_ARCH__>: Document avrxmega3 and 103.
24921 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24922 <__AVR_SHORT_CALLS__>: Document it.
24923 <__AVR_PM_BASE_ADDRESS__>: Document it.
24924 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24925 (AVR Variable Attributes) <progmem>: Document this is
24926 not needed for avrxmega3.
24927 (AVR Named Address Spaces) <__flash>: Dito.
24929 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24931 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24933 2017-06-12 Doug Rupp <rupp@adacore.com>
24935 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24936 Append vxworks-stdint.h to the tm_file list.
24937 * config/vxworks-stdint.h: New file.
24939 2017-06-12 Martin Liska <mliska@suse.cz>
24941 PR tree-optimization/81041
24942 * tree-profile.c (gimple_gen_ic_func_profiler):
24943 Create an extra BB in profile-generate
24944 (gimple_gen_time_profiler): Likewise.
24946 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24948 PR tree-optimization/81003
24949 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24950 (update_range_test): Use it instead of force_gimple_operand_gsi.
24952 2017-06-12 Richard Biener <rguenther@suse.de>
24954 PR tree-optimization/81053
24955 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24956 with backedge value not defined in loop. Simplify def stmt
24959 2017-06-11 Tom de Vries <tom@codesourcery.com>
24962 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24964 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24965 nvptx_cannot_force_const_mem.
24967 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24969 * opts.c (finish_options): Move test for flag_split_stack after
24970 it has been initialized.
24972 2017-06-11 Jason Merrill <jason@redhat.com>
24974 * tree.h (id_equal): New.
24975 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24976 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24977 instead of strcmp of IDENTIFIER_POINTER.
24979 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24981 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24982 (mark_all_inlined_calls_cdtor): Fix formating.
24983 (inline_transform): Rescale profile before inlining.
24985 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24987 * cgraph.h (cgraph_edge::clone): Update prototype.
24988 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24989 (cgraph_node::create_clone): Update.
24990 (cgraph_node::create_version_clone): Update.
24991 * tree-inline.c (copy_bb): Update.
24992 (expand_call_inline): Update.
24994 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24996 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24997 factored out from ...
24998 (rs6000_emit_prologue): ... here.
25000 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25002 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25003 factored out from ...
25004 (rs6000_emit_prologue): ... here.
25006 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25008 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25010 (handle_missing_profiles): Fix computation of tp_first_run.
25011 (counts_to_freqs): Do not touch freqs when count is 0.
25013 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25015 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25018 2017-06-10 Tom de Vries <tom@codesourcery.com>
25020 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25021 attributes): Document signal effective target.
25023 2017-06-10 Tom de Vries <tom@codesourcery.com>
25025 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25026 Document effective target stack_size.
25028 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25030 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25031 to the edit_context if they can be auto-applied.
25033 2017-06-9 Ian Lance Taylor <iant@golang.org>
25035 * opts.c (finish_options): If -fsplit-stack, disable implicit
25036 -forder-blocks-and-partition.
25037 * doc/invoke.texi (Optimize Options): Document that when using
25038 -fsplit-stack -forder-blocks-and-partition is not implicitly
25041 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25043 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25044 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25045 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25046 * builtins.def (abort, trap, unreachable): Declare cold.
25047 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25048 * tree-core.h (ECF_COLD): New.
25049 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25050 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25052 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25054 * predict.c (unlikely_executed_stmt_p): Cleanup.
25056 2017-06-09 Richard Biener <rguenther@suse.de>
25058 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25059 model if the ref is always written to.
25061 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25063 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25065 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25067 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25070 2017-06-09 Tom de Vries <tom@codesourcery.com>
25073 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25074 "target cannot support label values" when encountering LABEL_REF.
25076 2017-06-09 Martin Liska <mliska@suse.cz>
25078 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25079 (gimple_gen_ic_func_profiler): Emit direct comparison
25080 of __gcov_indirect_call_callee with NULL.
25081 (gimple_gen_time_profiler): Change probability from
25082 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25084 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25086 * profile.c (edge_gcov_counts): Turn to pointer.
25087 (compute_branch_probabilities, compute_branch_probabilities): Update.
25088 (branch_prob): Do not clear edge_gcov_count.
25089 * profile.h (edge_gcov_counts): Turn to pointer.
25090 (edge_gcov_count): Update.
25092 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25094 * gimple.h (gimple_check_failed): Mark cold.
25096 2017-06-09 Richard Biener <rguenther@suse.de>
25098 PR tree-optimization/66623
25099 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25100 refactor check_reduction into two parts, properly computing
25101 whether we have to check reduction validity for outer loop
25104 2017-06-09 Richard Biener <rguenther@suse.de>
25106 PR tree-optimization/79483
25107 * graphite-scop-detection.c (order): New global.
25108 (get_order): Compute bb to order mapping that satisfies code
25109 generation constraints.
25110 (cmp_pbbs): New helper.
25111 (build_scops): Start domwalk at entry block, sort generated
25114 2017-06-09 Richard Biener <rguenther@suse.de>
25116 PR middle-end/81007
25117 * ipa-polymorphic-call.c
25118 (ipa_polymorphic_call_context::restrict_to_inner_class):
25119 Skip FIELD_DECLs with error_mark_node type.
25120 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25123 2017-06-09 Martin Liska <mliska@suse.cz>
25125 * predict.c (struct branch_predictor): New struct.
25126 (test_prediction_value_range): New test.
25127 (predict_c_tests): New function.
25128 * selftest-run-tests.c (selftest::run_tests): Run the function.
25129 * selftest.h: Declare new tests.
25131 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25134 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25135 gen_add3_insn did not fail.
25136 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25137 r0, construct that number in a temporary reg and add that reg to r0.
25138 If asked to put the result in r0 as well, fail.
25140 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25142 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25143 for early expansion of vec_eqv.
25145 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25147 PR middle-end/81005
25148 * ubsan.c (instrument_null): Avoid pointless code temporary.
25149 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25151 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25155 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25158 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25161 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25163 * predict.c (unlikely_executed_edge_p): Move ahead.
25164 (probably_never_executed_edge_p): Use it.
25166 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25168 PR middle-end/79988
25169 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25170 gimple_call_builtin_p call.
25172 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25174 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25175 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25176 rtl_check_failed_type2, rtl_check_failed_code1,
25177 rtl_check_failed_code2, rtl_check_failed_code_mode,
25178 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25179 rtvec_check_failed_bounds, rtl_check_failed_flag,
25180 _fatal_insn_not_found, _fatal_insn): Likewise.
25181 * tree.h (tree_contains_struct_check_failed,
25182 tree_check_failed, tree_not_check_failed,
25183 tree_class_check_failed, tree_range_check_failed,
25184 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25185 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25186 tree_operand_check_failed, omp_clause_check_failed,
25187 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25190 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25192 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25193 flag_branch_probabilities.
25194 * ipa-inline.c (edge_badness): Likewise.
25195 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25196 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25197 * predict.c (maybe_hot_frequency_p): Likewise.
25198 (probably_never_executed): Likewise.
25199 * sched-ebb.c (schedule_ebbs): Likewise.
25200 * sched-rgn.c (find_single_block_region): Likewise.
25201 * tracer.c (tail_duplicate): Likewise.
25203 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25205 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25206 longer requires x_flag_profile_use.
25208 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25210 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25211 instead of flag_reorder_blocks_and_partition.
25212 * dbxout.c (dbxout_function_end): Likewise.
25213 * dwarf2out.c (gen_subprogram_die): Likewise.
25214 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25215 * hw-doloop.c (reorg_loops): Likewise.
25216 * varasm.c (assemble_start_function,
25217 assemble_end_function): Likewise.
25218 (decide_function_section): Do not check for
25219 flag_reorder_blocks_and_partition.
25221 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25223 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25225 (chkp_get_hard_register_fake_addr_expr): Ditto.
25226 (chkp_build_addr_expr): Add check for hard reg case.
25227 (chkp_parse_array_and_component_ref): Ditto.
25228 (chkp_find_bounds_1): Ditto.
25229 (chkp_process_stmt): Don't generate bounds store for
25232 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25234 * predict.c (maybe_hot_bb_p): Do not check profile status.
25235 (maybe_hot_edge_p): Likewise.
25236 (probably_never_executed): Check for zero counts even if profile
25238 (unlikely_executed_edge_p): New function.
25239 (unlikely_executed_stmt_p): New function.
25240 (unlikely_executed_bb_p): New function.
25241 (set_even_probabilities): Use unlikely predicates.
25242 (combine_predictions_for_bb): Likewise.
25243 (predict_paths_for_bb): Likewise.
25244 (predict_paths_leading_to_edge): Likewise.
25245 (determine_unlikely_bbs): New function.
25246 (estimate_bb_frequencies): Use it.
25247 (compute_function_frequency): Use zero counts even if profile is
25249 * profile-count.h: Fix typo.
25251 2017-08-08 Julia Koval <julia.koval@intel.com>
25253 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25254 _mm512_mask_cvtsepi16_storeu_epi8,
25255 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25256 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25257 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25258 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25259 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25260 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25261 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25262 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25263 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25264 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25265 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25266 __builtin_ia32_pmovuswb256mem_mask,
25267 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25268 __builtin_ia32_pmovwb512mem_mask): New builtins.
25270 2017-08-08 Julia Koval <julia.koval@intel.com>
25272 PR target/73350,80862
25273 * config/i386/subst.md (round): Fix round pattern.
25274 * config/i386/i386.c (ix86_erase_embedded_rounding):
25275 Fix erasing rounding for the fixed pattern.
25277 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25279 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25281 2017-06-08 Martin Liska <mliska@suse.cz>
25283 PR gcov-profile/80911
25284 * gcov.c (block_info::block_info): New constructor.
25286 2017-06-07 Carl Love <cel@us.ibm.com>
25288 * config/rs6000/rs6000-c: The return type of the following
25289 built-in functions was implemented as int not long long. Fix sign
25290 of return value for the unsigned version of vec_mulo and vec_mule.
25291 vector unsigned long long vec_bperm (vector unsigned long long,
25292 vector unsigned char)
25293 vector signed long long vec_mule (vector signed int,
25295 vector unsigned long long vec_mule (vector unsigned int,
25296 vector unsigned int)
25297 vector signed long long vec_mulo (vector signed int,
25299 vector unsigned long long vec_mulo (vector unsigned int,
25300 vector unsigned int)
25301 * doc/extend.texi: Fix the documentation for the built-in
25304 2017-06-07 Carl Love <cel@us.ibm.com>
25307 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25310 2017-06-07 Carl Love <cel@us.ibm.com>
25312 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25313 support, Generate doublehv for signed int/float for BE case only.
25315 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25317 * doc/invoke.texi (mcx16): Rewrite.
25319 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25321 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25322 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25323 *mov<mode>_softfloat, and an anonymous splitter): Use
25324 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25326 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25328 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25330 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25331 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25332 (rs6000_debug_reg_global): Adjust.
25333 (rs6000_init_hard_regno_mode_ok): Adjust.
25334 (rs6000_dbx_register_number): Adjust.
25335 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25336 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25337 Remove SPE_ACC and SPEFSCR.
25338 (REG_ALLOC_ORDER): Ditto.
25339 (FRAME_POINTER_REGNUM): Change to 111.
25340 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25341 (REG_CLASS_NAMES): Ditto.
25342 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25343 (REGISTER_NAMES): Ditto.
25344 (ADDITIONAL_REG_NAMES): Ditto.
25345 (rs6000_reg_names): Ditto.
25346 * config/rs6000/rs6000.md: Renumber some register number
25349 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25351 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25353 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25354 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25356 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25357 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25358 Delete the SPE high registers.
25359 (REG_ALLOC_ORDER): Ditto.
25360 (enum reg_class): Remove SPE_HIGH_REGS.
25361 (REG_CLASS_NAMES): Ditto.
25362 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25363 (REGISTER_NAMES): Ditto.
25364 (rs6000_reg_names): Ditto.
25365 * doc/tm.texi.in: Remove SPE as example.
25366 * doc/tm.texi: Regenerate.
25368 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25370 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25371 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25372 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25373 * config/rs6000/rs6000.md (type): Remove "brinc".
25375 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25377 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25378 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25379 * config/rs6000/linuxspe.h: Delete file.
25380 * config/rs6000/rs6000.md: Don't include spe.md.
25381 * config/rs6000/spe.h: Delete file.
25382 * config/rs6000/spe.md: Delete file.
25383 * config/rs6000/t-rs6000: Remove spe.md.
25385 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25387 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25388 (reg_or_none500mem_operand): Delete.
25389 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25390 instead of reg_or_none500mem_operand.
25392 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25394 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25395 handling of SPE flags.
25396 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25398 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25400 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25402 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25403 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25404 paired_divv2sf3): Similar.
25405 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25406 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25407 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25409 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25410 Rename the paired_* instruction patterns.
25411 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25413 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25414 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25415 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25416 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25417 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25418 PAIRED_VECTOR_MODE.
25419 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25420 (spe_func_has_64bit_regs_p): Delete.
25421 (spe_expand_predicate_builtin): Delete.
25422 (spe_expand_evsel_builtin): Delete.
25423 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25424 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25425 (invalid_e500_subreg): Delete.
25426 (rs6000_legitimize_address): Always force_reg op2 as well, for
25427 paired single memory accesses.
25428 (rs6000_member_type_forces_blk): Delete.
25429 (rs6000_spe_function_arg): Delete.
25430 (rs6000_expand_unop_builtin): Delete SPE handling.
25431 (rs6000_expand_binop_builtin): Ditto.
25432 (spe_expand_stv_builtin): Delete.
25433 (bdesc_2arg_spe): Delete.
25434 (spe_expand_builtin): Delete.
25435 (spe_expand_predicate_builtin): Delete.
25436 (spe_expand_evsel_builtin): Delete.
25437 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25438 (spe_init_builtins): Delete.
25439 (spe_func_has_64bit_regs_p): Delete.
25440 (savres_routine_name): Delete "info" parameter. Adjust callers.
25441 (rs6000_emit_stack_reset): Ditto.
25442 (rs6000_dwarf_register_span): Delete.
25443 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25444 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25445 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25447 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25449 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25450 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25451 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25452 mulv2sf3, divv2sf3): Delete expanders.
25454 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25456 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25458 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25460 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25461 * config/rs6000/rs6000.c: Ditto.
25463 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25465 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25466 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25467 comparison_operator.
25469 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25471 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25472 * config/rs6000/rs6000.opt: Ditto.
25473 * config/rs6000/t-rtems: Ditto.
25475 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25477 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25478 TARGET_E500_SINGLE by 0, simplify.
25479 * config/rs6000/rs6000.c: Ditto.
25480 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25481 (spe_build_register_parallel): Delete.
25482 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25483 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25484 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25485 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25486 (E500_CONVERT): Delete.
25487 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25489 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25491 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25492 * config/rs6000/dfp.md: Ditto.
25493 (negdd2, *negdd2_fpr): Merge.
25494 (absdd2, *absdd2_fpr): Merge.
25495 (negtd2, *negtd2_fpr): Merge.
25496 (abstd2, *abstd2_fpr): Merge.
25497 * config/rs6000/e500.h: Delete file.
25498 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25499 TARGET_FPRS by 1 and simplify.
25500 * config/rs6000/rs6000-c.c: Ditto.
25501 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25502 TARGET_DF_SPE by 0.
25503 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25505 * config/rs6000/rs6000.md: Ditto.
25506 (floatdidf2, *floatdidf2_fpr): Merge.
25507 (move_from_CR_gt_bit): Delete.
25508 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25509 (E500_CR_IOR_COMPARE): Delete.
25510 (All patterns that require !TARGET_FPRS): Delete.
25511 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25513 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25515 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25517 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25519 * graphds.c (add_edge): Intitialize edge's attached data.
25520 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25521 pointer parameter. Call pointed function on each edge during
25522 graph traversing. Skip traversing the edge when the function
25524 (graphds_dfs, graphds_scc): Ditto.
25525 (for_each_edge): New parameter. Pass the new parameter to callback
25527 * graphds.h (skip_edge_callback): New function pointer type.
25528 (graphds_dfs, graphds_scc): New function pointer parameter.
25529 (graphds_edge_callback, for_each_edge): New parameter.
25531 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25533 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25534 out code checking if runtime alias check is possible to below ...
25535 Call the new function.
25536 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25537 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25539 2017-06-07 Marek Polacek <polacek@redhat.com>
25542 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25543 TYPE_OVERFLOW_WRAPS check.
25545 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25547 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25548 if versioning is required.
25549 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25550 peeling with the check for versioning.
25552 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25554 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25555 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25556 Set true to new parameter if new ssa variable is defined.
25557 (vect_gen_vector_loop_niters): Refactor. Set range information
25558 for the new vector loop bound variable.
25559 (vect_do_peeling): Ditto.
25561 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25563 * tree-affine.c (ssa.h): Include header file.
25564 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25565 has wrapping overflow behavior.
25567 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25569 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25571 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25573 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25574 (tree_to_aff_combination): ... here.
25576 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25578 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25579 reg_pressure model function.
25580 (ivopts_global_cost_for_size): Delete.
25581 (determine_set_costs, iv_ca_recount_cost): Call new model function
25582 ivopts_estimate_reg_pressure.
25584 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25586 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25587 expensive than udiv. Remove floating point cases from mod.
25589 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25591 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25592 Increase idiv cost.
25594 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25596 * config/aarch64/aarch64.md
25597 (copysignsf3): Fix mask generation.
25599 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25601 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25603 (class dump_manager): Add register_dumps method.
25604 * dumpfile.c: Include langhooks.h.
25605 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25606 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25607 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25608 (dump_manager::dump_register): Start with 512 entries instead of 32.
25609 (dump_manager::register_dumps): New method.
25610 * toplev.c (general_init): Instead of invoking register_dumps
25611 langhook, invoke register_dumps method on the dump manager.
25612 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25615 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25617 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25618 Say that # requires an associated define_split to exist, and that
25619 the define_split must be suitable for use after register allocation.
25621 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25623 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25624 (compute_outgoing_frequencies): Also initialize zero counts.
25625 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25626 around loops; preserve more of profile when nothing changes.
25628 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25630 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25632 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25633 * config/arm/arm-cpu-cdata.h: Regenerate.
25634 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25635 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25636 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25637 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25638 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25640 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25642 * config/arm/t-rmprofile: Likewise.
25643 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25645 2017-06-06 David S. Miller <davem@davemloft.net>
25648 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25649 blockage if function uses alloca.
25651 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25653 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25654 New "uid" fields to hold pretty-print IDs of group and ref.
25655 Memory references are now identified as <group_id>:<ref_id>
25656 instead of using [random] addresses.
25657 (dump_mem_details): Simplify, no functional change.
25658 (dump_mem_ref): Simplify and make output more concise.
25659 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25660 (find_or_create_group): Initialize group uid.
25661 (record_ref): Initialize ref uid. Improve debug output.
25662 (prune_group_by_reuse, should_issue_prefetch_p,)
25663 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25664 (mark_nontemporal_store, determine_loop_nest_reuse):
25665 Improve debug output.
25667 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25669 * dbgcnt.def (prefetch): New debug counter.
25670 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25671 (schedule_prefetches): Stop issueing prefetches if debug counter
25674 2017-06-06 Tom de Vries <tom@codesourcery.com>
25676 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25677 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25679 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25681 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25682 Use aarch64_reg_or_zero predicate for operand 4.
25683 (aarch64_compare_and_swap<mode> define_insn_and_split):
25684 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25685 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25687 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25689 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25690 (arm_compute_save_core_reg_mask): This.
25691 (thumb1_compute_save_reg_mask): Rename into ...
25692 (thumb1_compute_save_core_reg_mask): This.
25693 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25694 (arm_compute_frame_layout): Likewise.
25696 2017-06-06 Richard Biener <rguenther@suse.de>
25698 PR tree-optimization/80974
25699 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25700 keep or clear leaders SSA info.
25702 2017-06-06 Tom de Vries <tom@codesourcery.com>
25704 * config/nvptx/nvptx.c (split_mode_p): New function.
25705 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25707 2017-06-06 Tom de Vries <tom@codesourcery.com>
25709 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25711 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25714 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25717 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25719 * shrink-wrap.c (handle_simple_exit): Update profile.
25720 (try_shrink_wrapping): Upate profile.
25722 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25724 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25725 (tree_guess_outgoing_edge_probabilities): New.
25726 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25727 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25729 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25731 * ipa-split.c (split_function): Initialize return bb profile.
25733 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25735 * profile.c (compute_branch_probabilities): Also initialize
25736 EXIT_BLOCK profile.
25738 2017-06-06 Richard Biener <rguenther@suse.de>
25740 PR tree-optimization/80928
25741 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25742 (vect_analyze_loop_operations): Properly guard analysis for
25744 (vect_transform_loop): Likewise.
25745 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25746 (vect_model_induction_cost): Do not cost for pure SLP.
25747 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25748 of induction in inner loop vectorization.
25749 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25750 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25751 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25753 (vect_analyze_slp_cost_1): Cost induction.
25754 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25755 (vect_get_slp_vect_defs): Likewise.
25756 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25757 (vect_transform_stmt): Handle SLP reductions.
25758 * tree-vectorizer.h (vectorizable_induction): Adjust.
25760 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25762 * config/rs6000/rs6000.c (make_resolver_func): Update
25763 init_lowered_empty_function call.
25765 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25767 * doc/invoke.texi: Document the -fprofile-abs-path option.
25768 * common.opt (fprofile-abs-path): New option.
25769 * gcov-io.h (gcov_write_filename): Declare.
25770 * gcov-io.c (gcov_write_filename): New function.
25771 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25772 * profile.c (output_location): Likewise.
25774 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25776 * shring-wrap.c: Revert accidental commit.
25778 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25780 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25782 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25784 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25786 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25787 profile in callgraph edge.
25788 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25789 (apply_scale): Likewise.
25790 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25791 Also scale profile when inlining function with zero profile.
25792 (initialize_cfun): Update exit block profile even when it is zero.
25793 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25794 when profile is read.
25796 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25798 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25799 (CLONE_*): New constants to define the processors we can generate
25800 code for with the target_clone attribute.
25801 (rs6000_clone_map): New array to identify which clone processors
25802 the current program is running on.
25803 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25804 target_clone attribute.
25805 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25806 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25807 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25808 (cpu_expand_builtin): Add support for target_clone attribute.
25809 (rs6000_valid_attribute_p): Allow "default" attribute.
25810 (get_decl_name): New debug function to simplify printing the
25811 current function name in debugging statements.
25812 (rs6000_clone_priority): New functions to support the target_clone
25813 attribute, and be able to generate code to switch between ISA 2.05
25814 through ISA 3.0 (power6 through power9).
25815 (rs6000_compare_version_priority): Likewise.
25816 (rs6000_get_function_versions_dispatcher): Likewise.
25817 (make_resolver_func): Likewise.
25818 (add_condition_to_bb): Likewise.
25819 (dispatch_function_versions): Likewise.
25820 (rs6000_generate_version_dispatcher_body): Likewise.
25821 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25822 (fusion_gpr_load_p): Fix a spacing issue.
25823 * doc/extend.texi (Common Function Attributes): Document that the
25824 PowerPC supports the target_clone attribute.
25826 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25828 * config/arm/arm.h: explain F symbol found in description of ARM
25829 register allocation in its legend.
25831 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25833 * config/mips/frame-header-opt.c: Include profile-count.h.
25834 * config/riscv/riscv.c: Include profile-count.h
25836 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25838 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25840 (sm_set_flag_if_changed): Add bbs field.
25841 (execute_sm_if_changed_flag_set): Pass BBS.
25842 (execute_sm): Update.
25844 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25846 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25849 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25851 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25852 (peephole2): New peephole2 to emit the above.
25853 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25855 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25857 * config/aarch64/aarch64.c (define_peephole2 above
25858 *sub_<shift>_<mode>): New peephole.
25860 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25862 * config/i386/i386.c (make_resolver_func): Update.
25863 * Makefile.in: Add profile-count.h and profile-count.o
25864 * auto-profile.c (afdo_indirect_call): Update to new API.
25865 (afdo_set_bb_count): Update.
25866 (afdo_propagate_edge): Update.
25867 (afdo_propagate_circuit): Update.
25868 (afdo_calculate_branch_prob): Update.
25869 (afdo_annotate_cfg): Update.
25870 * basic-block.h: Include profile-count.h
25871 (struct edge_def): Turn count to profile_count.
25872 (struct basic_block_def): Likewie.
25873 (REG_BR_PROB_BASE): Move to profile-count.h
25874 (RDIV): Move to profile-count.h
25875 * bb-reorder.c (max_entry_count): Turn to profile_count.
25876 (find_traces): Update.
25877 (rotate_loop):Update.
25878 (connect_traces):Update.
25879 (sanitize_hot_paths):Update.
25880 * bt-load.c (migrate_btr_defs): Update.
25881 * cfg.c (RDIV): Remove.
25882 (init_flow): Use alloc_block.
25883 (alloc_block): Uninitialize count.
25884 (unchecked_make_edge): Uninitialize count.
25885 (check_bb_profile): Update.
25886 (dump_edge_info): Update.
25887 (dump_bb_info): Update.
25888 (update_bb_profile_for_threading): Update.
25889 (scale_bbs_frequencies_int): Update.
25890 (scale_bbs_frequencies_gcov_type): Update.
25891 (scale_bbs_frequencies_profile_count): New.
25892 * cfg.h (update_bb_profile_for_threading): Update.
25893 (scale_bbs_frequencies_profile_count): Declare.
25894 * cfgbuild.c (compute_outgoing_frequencies): Update.
25895 (find_many_sub_basic_blocks): Update.
25896 * cfgcleanup.c (try_forward_edges): Update.
25897 (try_crossjump_to_edge): Update.
25898 * cfgexpand.c (expand_gimple_tailcall): Update.
25899 (construct_exit_block): Update.
25900 * cfghooks.c (verify_flow_info): Update.
25901 (dump_bb_for_graph): Update.
25902 (split_edge): Update.
25903 (make_forwarder_block): Update.
25904 (duplicate_block): Update.
25905 (account_profile_record): Update.
25906 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25907 (get_estimated_loop_iterations): Update.
25908 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25909 (single_likely_exit): Update.
25910 * cfgloopmanip.c (scale_loop_profile): Update.
25912 (set_zero_probability): Update.
25913 (lv_adjust_loop_entry_edge): Update.
25914 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25915 (purge_dead_edges): Update.
25916 (rtl_account_profile_record): Update.
25917 * cgraph.c (cgraph_node::create): Uninitialize count.
25918 (symbol_table::create_edge): Uninitialize count.
25919 (cgraph_update_edges_for_call_stmt_node): Update.
25920 (cgraph_edge::dump_edge_flags): Update.
25921 (cgraph_node::dump): Update.
25922 (cgraph_edge::maybe_hot_p): Update.
25923 * cgraph.h: Include profile-count.h
25924 (create_clone), create_edge, create_indirect_edge): Update.
25925 (cgraph_node): Turn count to profile_count.
25926 (cgraph_edge0: Likewise.
25927 (make_speculative, clone): Update.
25928 (create_edge): Update.
25929 (init_lowered_empty_function): Update.
25930 * cgraphclones.c (cgraph_edge::clone): Update.
25931 (duplicate_thunk_for_node): Update.
25932 (cgraph_node::create_clone): Update.
25933 * cgraphunit.c (cgraph_node::analyze): Update.
25934 (cgraph_node::expand_thunk): Update.
25935 * final.c (dump_basic_block_info): Update.
25936 * gimple-streamer-in.c (input_bb): Update.
25937 * gimple-streamer-out.c (output_bb): Update.
25938 * graphite.c (print_global_statistics): Update.
25939 (print_graphite_scop_statistics): Update.
25940 * hsa-brig.c: Include basic-block.h.
25941 * hsa-dump.c: Include basic-block.h.
25942 * hsa-gen.c (T sum_slice): Update.
25943 (convert_switch_statements):Update.
25944 * hsa-regalloc.c: Include basic-block.h.
25945 * ipa-chkp.c (chkp_produce_thunks): Update.
25946 * ipa-cp.c (struct caller_statistics): Update.
25947 (init_caller_stats): Update.
25948 (gather_caller_stats): Update.
25949 (ipcp_cloning_candidate_p): Update.
25950 (good_cloning_opportunity_p): Update.
25951 (get_info_about_necessary_edges): Update.
25952 (dump_profile_updates): Update.
25953 (update_profiling_info): Update.
25954 (update_specialized_profile): Update.
25955 (perhaps_add_new_callers): Update.
25956 (decide_about_value): Update.
25957 (ipa_cp_c_finalize): Update.
25958 * ipa-devirt.c (struct odr_type_warn_count): Update.
25959 (struct decl_warn_count): Update.
25960 (struct final_warning_record): Update.
25961 (possible_polymorphic_call_targets): Update.
25962 (ipa_devirt): Update.
25963 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25964 * ipa-icf.c (sem_function::merge): Update.
25965 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25966 * ipa-inline.c (compute_uninlined_call_time): Update.
25967 (compute_inlined_call_time): Update.
25968 (want_inline_small_function_p): Update.
25969 (want_inline_self_recursive_call_p): Update.
25970 (edge_badness): Update.
25971 (lookup_recursive_calls): Update.
25972 (recursive_inlining): Update.
25973 (inline_small_functions): Update.
25974 (dump_overall_stats): Update.
25975 (dump_inline_stats): Update.
25976 * ipa-profile.c (ipa_profile_generate_summary): Update.
25977 (ipa_propagate_frequency): Update.
25978 (ipa_profile): Update.
25979 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25980 * ipa-utils.c (ipa_merge_profiles): Update.
25981 * loop-doloop.c (doloop_modify): Update.
25982 * loop-unroll.c (report_unroll): Update.
25983 (unroll_loop_runtime_iterations): Update.
25984 * lto-cgraph.c (lto_output_edge): Update.
25985 (lto_output_node): Update.
25986 (input_node): Update.
25987 (input_edge): Update.
25988 (merge_profile_summaries): Update.
25989 * lto-streamer-in.c (input_cfg): Update.
25990 * lto-streamer-out.c (output_cfg): Update.
25991 * mcf.c (create_fixup_graph): Update.
25992 (adjust_cfg_counts): Update.
25993 (sum_edge_counts): Update.
25994 * modulo-sched.c (sms_schedule): Update.
25995 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25996 * predict.c (maybe_hot_count_p): Update.
25997 (probably_never_executed): Update.
25998 (dump_prediction): Update.
25999 (combine_predictions_for_bb): Update.
26000 (propagate_freq): Update.
26001 (handle_missing_profiles): Update.
26002 (counts_to_freqs): Update.
26003 (rebuild_frequencies): Update.
26004 (force_edge_cold): Update.
26005 * predict.h: Include profile-count.h
26006 (maybe_hot_count_p, counts_to_freqs): UPdate.
26007 * print-rtl-function.c: Do not include cfg.h
26008 * print-rtl.c: Include basic-block.h
26009 * profile-count.c: New file.
26010 * profile-count.h: New file.
26011 * profile.c (is_edge_inconsistent): Update.
26012 (correct_negative_edge_counts): Update.
26013 (is_inconsistent): Update.
26014 (set_bb_counts): Update.
26015 (read_profile_edge_counts): Update.
26016 (compute_frequency_overlap): Update.
26017 (compute_branch_probabilities): Update; Initialize and deinitialize
26019 (branch_prob): Update.
26020 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26021 (edge_gcov_count): New.
26022 (bb_gcov_count): New.
26023 * shrink-wrap.c (try_shrink_wrapping): Update.
26024 * tracer.c (better_p): Update.
26025 * trans-mem.c (expand_transaction): Update.
26026 (ipa_tm_insert_irr_call): Update.
26027 (ipa_tm_insert_gettmclone_call): Update.
26028 * tree-call-cdce.c: Update.
26029 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26030 (gimple_duplicate_sese_tail): Update.
26031 (gimple_account_profile_record): Update.
26032 (execute_fixup_cfg): Update.
26033 * tree-inline.c (copy_bb): Update.
26034 (copy_edges_for_bb): Update.
26035 (initialize_cfun): Update.
26036 (freqs_to_counts): Update.
26037 (copy_cfg_body): Update.
26038 (expand_call_inline): Update.
26039 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26040 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26041 (try_unroll_loop_completely): Update.
26042 (try_peel_loop): Update.
26043 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26044 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26045 * tree-ssa-loop-split.c (connect_loops): Update.
26046 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26047 * tree-ssa-reassoc.c (branch_fixup): Update.
26048 * tree-ssa-tail-merge.c (replace_block_by): Update.
26049 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26050 (compute_path_counts): Update.
26051 (update_profile): Update.
26052 (recompute_probabilities): Update.
26053 (update_joiner_offpath_counts): Update.
26054 (estimated_freqs_path): Update.
26055 (freqs_to_counts_path): Update.
26056 (clear_counts_path): Update.
26057 (ssa_fix_duplicate_block_edges): Update.
26058 (duplicate_thread_path): Update.
26059 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26060 (struct switch_conv_info): Update.
26061 * tree-tailcall.c (decrease_profile): Update.
26062 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26063 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26064 * value-prof.c (check_counter): Update.
26065 (gimple_divmod_fixed_value): Update.
26066 (gimple_mod_pow2): Update.
26067 (gimple_mod_subtract): Update.
26068 (gimple_ic_transform): Update.
26069 (gimple_stringop_fixed_value): Update.
26070 * value-prof.h (gimple_ic): Update.
26072 2017-06-02 Carl Love <cel@us.ibm.com>
26074 * config/rs6000/rs6000-c: Add support for built-in functions
26075 vector double vec_doublee (vector signed int);
26076 vector double vec_doublee (vector unsigned int);
26077 vector double vec_doublee (vector float);
26078 vector double vec_doubleh (vector signed int);
26079 vector double vec_doubleh (vector unsigned int);
26080 vector double vec_doubleh (vector float);
26081 vector double vec_doublel (vector signed int);
26082 vector double vec_doublel (vector unsigned int);
26083 vector double vec_doublel (vector float);
26084 vector double vec_doubleo (vector signed int);
26085 vector double vec_doubleo (vector unsigned int);
26086 vector double vec_doubleo (vector float);.
26087 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26088 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26090 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26091 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26092 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26094 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26095 vec_doublel, vec_doubleh.
26096 * doc/extend.texi: Update the built-in documentation file for the
26097 new built-in functions.
26099 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26102 * ipa-inline-analysis.c (free_growth_caches): Set
26103 edge_removal_hook_holder to NULL after removing it.
26105 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26107 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26108 comparision with zero.
26110 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26111 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26112 for early expansion of vec_min and vec_max builtins.
26113 (builtin_function_type): Add min/max unsigned variants to those
26114 identified as having unsigned arguments.
26116 2017-06-02 Olivier Hainque <hainque@adacore.com>
26118 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26120 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26122 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26123 Use VALL_F16 iterator rather than VALL.
26125 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26127 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26128 Emit CBNZ inside loop when doing a strong exchange and comparing
26129 against zero. Generate the CC flags after the loop.
26131 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26133 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26134 (dl_section_ref): New.
26135 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26136 On AIX, append an expression to subtract the size of the
26137 section length to dl_section_ref.
26139 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26141 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26142 for early expansion of vector absolute builtins.
26144 2017-06-02 Richard Biener <rguenther@suse.de>
26146 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26147 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26149 2017-06-02 Richard Biener <rguenther@suse.de>
26151 PR tree-optimization/80948
26152 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26153 stmt order as well.
26155 2017-06-02 Richard Biener <rguenther@suse.de>
26157 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26159 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26160 for inductions as relevant.
26162 2017-06-02 Richard Biener <rguenther@suse.de>
26164 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26165 (vectorizable_induction): ... this. Remove dead code.
26167 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26169 * builtins. (expand_builtin_alloca): Remove second parameter and
26170 infer its value from the first parameter instead.
26171 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26173 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26175 PR rtl-optimization/80903
26176 * loop-doloop.c (add_test): Unshare sequence.
26178 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26180 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26182 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26184 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26186 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26187 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26188 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26189 (xlogue_layout::get_stub_rtx): Make static.
26190 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26191 (xlogue_layout::compute_stub_managed_regs): Rename to...
26192 (xlogue_layout::count_stub_managed_regs): ...this.
26193 (xlogue_layout::is_stub_managed_reg): New function.
26194 (xlogue_layout::m_stub_names): Rename to...
26195 (xlogue_layout::s_stub_names): ...this, make static.
26196 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26197 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26198 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26199 xlogue_layout::s_stub_names): Instantiate statics.
26200 (stub_managed_regs): Remove.
26201 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26202 (disable_call_ms2sysv_xlogues): Rename to...
26203 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26204 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26206 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26207 change after reload_completed.
26208 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26210 (ix86_expand_prologue): Likewise.
26211 (ix86_expand_epilogue): Likewise.
26212 (ix86_expand_split_stack_prologue): Likewise.
26213 (ix86_compute_frame_layout): Remove frame parameter ...
26214 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26215 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26217 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26218 (ix86_frame): Move from here ...
26219 * config/i386/i386.h (ix86_frame): ... to here.
26220 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26221 complete ix86_frame data structure instead. Remove some_ld_name.
26223 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26225 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26226 symbols that hold a DECL_VALUE_EXPR.
26228 2017-06-01 Martin Jambor <mjambor@suse.cz>
26230 PR tree-optimization/80898
26231 * tree-sra.c (process_subtree_disqualification): Removed.
26232 (disqualify_candidate): Do not acll
26233 process_subtree_disqualification.
26234 (subtree_mark_written_and_enqueue): New function.
26235 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26236 RHS has been disqualified and re-queue LHS if necessary. Apart
26237 from that, ignore disqualified RHS.
26239 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26241 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26242 address fetch for z10 or later.
26244 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26246 * config/arc/arc.md (tst_movb): Add guard when splitting.
26248 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26250 * config/arc/arc.c (arc_can_eliminate): Test against
26251 arc_frame_pointer_needed.
26253 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26255 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26256 to prevent store reordering.
26257 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26258 (type): Add block type.
26259 (stack_tie): Define special instruction to be used in
26262 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26264 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26265 constraint. It is not valid for the pattern.
26266 (noncommutative_binary_comparison): Likewise.
26268 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26270 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26273 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26275 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26276 be used by the reg-alloc.
26278 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26280 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26281 reg-alloc when having mul64 or mul32x16 instructions.
26282 (mulsidi3): Likewise.
26283 (umulsidi3): Likewise.
26284 (mulsi32x16): New pattern.
26285 (mulsi64): Likewise.
26286 (mulsidi64): Likewise.
26287 (umulsidi64): Likewise.
26288 (MUL32x16_REG): Define.
26289 (mul64_600): Use MUL32x16_REG.
26290 (mac64_600): Likewise.
26291 (umul64_600): Likewise.
26292 (umac64_600): Likewise.
26294 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26296 * config/arc/arc.md (mulsi3_700): Make it commutative.
26298 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26300 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26302 (*sign_extendsidi2_insn): Likewise for movstosw.
26304 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26306 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26307 the type of the input discriminant value. Convert the
26308 discriminant value of signedness vary.
26310 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26312 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26313 Add to -Wall section.
26315 2017-06-01 Richard Biener <rguenther@suse.de>
26317 PR middle-end/66313
26318 * fold-const.c (fold_plusminus_mult_expr): If the factored
26319 factor may be zero use a wrapping type for the inner operation.
26320 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26321 and handle moved defs.
26322 (process_assignment): Properly guard the unary op case. Return a
26323 tri-state indicating that moving the stmt before the call may allow
26324 to continue. Pass through to_move.
26325 (find_tail_calls): Handle moving unrelated defs before
26328 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26331 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26332 splitter result in the canonical way.
26334 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26336 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26337 also for 32bit target. Update insn attributes.
26338 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26340 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26342 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26343 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26344 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26345 (_mm_maskz_min_ss): New intrinsics.
26347 2017-05-31 Martin Liska <mliska@suse.cz>
26349 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26350 Change comment style to one we normally use.
26351 (vectorizable_reduction): Likewise.
26352 (vectorizable_induction): Likewise.
26353 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26354 (vectorizable_call): Likewise.
26355 (vectorizable_simd_clone_call): Likewise.
26356 (vectorizable_conversion): Likewise.
26357 (vectorizable_assignment): Likewise.
26358 (vectorizable_shift): Likewise.
26359 (vectorizable_operation): Likewise.
26360 (vectorizable_store): Likewise.
26361 (vectorizable_load): Likewise.
26362 * tree-vectorizer.h: Likewise.
26364 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26366 * passes.c (emergency_dump_function): New.
26367 * tree-pass.h (emergency_dump_function): Declare.
26368 * plugin.c (plugins_internal_error_function): Remove.
26369 * plugin.h (plugins_internal_error_function): Remove declaration.
26370 * toplev.c (internal_error_function): New static function. Use it...
26371 (general_init): ...here.
26373 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26375 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26376 (arc_rtx_costs): Add costs for new patterns.
26377 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26378 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26380 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26382 * tree-ssa-strlen.c (get_next_strinfo): New function.
26383 (get_stridx_plus_constant): Use it.
26384 (zero_length_string): Likewise.
26385 (adjust_related_strinfos): Likewise.
26386 (adjust_last_stmt): Likewise.
26388 2017-05-31 Richard Biener <rguenther@suse.de>
26391 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26392 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26394 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26396 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26397 loop_vinfo argument and use of dependence distance vectors.
26398 Check instead whether the two references differ only in their
26399 initial value and assume that they have the same alignment if the
26400 difference is a multiple of the vector alignment.
26401 (vect_analyze_data_refs_alignment): Update call accordingly.
26403 2017-05-31 Martin Liska <mliska@suse.cz>
26406 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26408 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26410 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26411 (create_intersect_range_checks): Move from ...
26412 * tree-data-ref.c (create_intersect_range_checks_index)
26413 (create_intersect_range_checks): ... to here.
26414 (create_runtime_alias_checks): New function factored from ...
26415 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26416 here. Call above function.
26417 * tree-data-ref.h (create_runtime_alias_checks): New function.
26419 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26421 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26422 segment length for dr_b and compute it in wide_int.
26424 2017-05-31 Richard Biener <rguenther@suse.de>
26426 PR tree-optimization/80906
26427 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26428 and pass through iv_map.
26429 (copy_bb_and_scalar_dependences): Adjust.
26430 (translate_pending_phi_nodes): Likewise.
26431 (copy_loop_close_phi_args): Handle code-generating IVs instead
26434 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26436 * diagnostic-color.c (color_dict): Add "type-diff".
26437 (parse_gcc_colors): Update comment.
26438 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26439 -fdiagnostics-show-template-tree and -fno-elide-type.
26440 (GCC_COLORS): Add type-diff to example.
26442 (-fdiagnostics-show-template-tree): New.
26443 (-fno-elide-type): New.
26444 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26445 the pp_format_decoder callback. Call any m_format_postprocessor's
26447 (pretty_printer::pretty_printer): Initialize
26448 m_format_postprocessor.
26449 (pretty_printer::~pretty_printer): Delete any
26450 m_format_postprocessor.
26451 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26452 (class format_postprocessor): New class.
26453 (struct pretty_printer::format_decoder): Document the new parameters.
26454 (struct pretty_printer::m_format_postprocessor): New field.
26455 * tree-diagnostic.c (default_tree_printer): Update for new
26456 bool and const char ** params.
26457 * tree-diagnostic.h (default_tree_printer): Likewise.
26459 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26461 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26462 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26463 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26464 rs6000_gen_cell_microcode code.
26465 (rs6000_final_prescan_insn): Delete.
26466 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26467 "warn-cell-microcode" entries.
26468 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26469 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26470 throughout. Change cc_reg_not_micro_cr0_operand to
26471 cc_reg_not_cr0_operand throughout.
26472 (*extendhi<mode>2_noload): Delete.
26473 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26474 (mwarn-cell-microcode): Delete.
26475 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26476 -mgen-cell-microcode and -mwarn-cell-microcode.
26478 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26481 * config/i386/constraints.md (Yd): New constraint.
26483 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26484 and (?Yd, r) alternatives. Update insn attributes.
26485 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26486 and (?*Yd, r) alternatives. Update insn attributes.
26487 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26489 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26491 * gimplify.c (gimplify_modify_expr): Don't create a
26492 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26495 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26497 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26499 2017-05-30 Richard Biener <rguenther@suse.de>
26501 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26502 and reduc_def fields.
26503 (STMT_VINFO_REDUC_TYPE): New define.
26504 (STMT_VINFO_REDUC_DEF): Likewise.
26505 (vect_force_simple_reduction): Adjust prototype.
26506 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26507 (vect_is_simple_reduction): Remove check_reduction argument.
26508 (vect_force_simple_reduction): Adjust and set
26509 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26510 (vectorizable_reduction): Do not re-do reduction analysis
26511 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26512 * tree-parloops.c (gather_scalar_reductions): Adjust.
26514 2017-05-30 Richard Biener <rguenther@suse.de>
26516 PR middle-end/80901
26517 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26520 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26522 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26523 Introduce unknown_misalignment parameter and remove vf.
26524 (vect_peeling_hash_get_lowest_cost):
26525 Pass unknown_misalignment parameter.
26526 (vect_enhance_data_refs_alignment):
26527 Fix unsupportable data ref treatment.
26529 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26531 * tree-vect-data-refs.c (vect_get_data_access_cost):
26532 Workaround for SLP handling.
26533 (vect_enhance_data_refs_alignment):
26534 Compute costs for doing no peeling at all, compare to the best
26535 peeling costs so far and avoid peeling if cheaper.
26537 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26539 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26540 Return peeling info and set costs to zero for unlimited cost
26542 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26543 with unknown misalignment. Compute and costs for unknown
26544 misalignment, compare them to the costs for known misalignment
26545 and choose the cheapest for peeling.
26547 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26549 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26550 (vect_get_peeling_costs_all_drs): Create function.
26551 (vect_peeling_hash_get_lowest_cost):
26552 Use vect_get_peeling_costs_all_drs.
26553 (vect_peeling_supportable): Create function.
26554 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26556 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26558 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26559 DR_HAS_NEGATIVE_STEP.
26560 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26561 (vect_enhance_data_refs_alignment): Use.
26562 (vect_duplicate_ssa_name_ptr_info): Use.
26563 * tree-vectorizer.h (dr_misalignment): Use.
26564 (known_alignment_for_access_p): Use.
26566 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26569 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26571 (has_section_name): New function.
26573 2017-05-30 Martin Liska <mliska@suse.cz>
26576 * auto-profile.c (get_function_decl_from_block): Fix
26579 2017-05-30 Richard Biener <rguenther@suse.de>
26581 PR middle-end/80876
26582 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26584 2017-05-30 Martin Liska <mliska@suse.cz>
26586 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26587 * dumpfile.h (struct dump_file_info): Remove ctors.
26589 2017-05-30 Martin Liska <mliska@suse.cz>
26591 * predict.def: Fix GNU coding style.
26593 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26595 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26596 Mark 'to' argument with ATTRIBUTE_UNUSED.
26598 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26600 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26601 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26602 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26605 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26607 * doc/install.texi (Options specification): Restore entry of
26608 --enable-sjlj-exceptions.
26610 2017-05-27 Michael Eager <eager@eagercon.com>
26613 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26615 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26617 * config/microblaze/microblaze.h
26618 (FIXED_REGISTERS): Update in macro.
26619 (CALL_USED_REGISTERS): Update in macro.
26621 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26623 * doc/install.texi: Add links to macOS binary distributions.
26625 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26629 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26631 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26633 2017-05-26 Martin Liska <mliska@suse.cz>
26635 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26637 2017-05-26 Martin Liska <mliska@suse.cz>
26639 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26640 always leading ';; '.
26641 (dump_bb_info): Likewise.
26642 (brief_dump_cfg): Likewise.
26643 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26644 * dumpfile.c: Remove usage of TDF_VERBOSE.
26645 * dumpfile.h (enum dump_kind): Likewise.
26646 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26647 * print-tree.c (debug_verbose): Remove.
26648 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26649 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26650 * tree-diagnostic.c (default_tree_printer): Replace
26651 TDF_DIAGNOSTIC with TDF_SLIM.
26653 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26655 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26656 in parameter loop, rather than loop_vinfo.
26657 (create_intersect_range_checks): Ditto.
26658 (vect_create_cond_for_alias_checks): Update call to above functions.
26660 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26662 PR tree-optimization/80815
26663 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26664 for merging runtime alias checks. Handle negative DR_STEPs.
26666 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26668 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26670 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26671 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26672 out code pruning runtime alias checks.
26673 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26674 factored out from above.
26675 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26677 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26679 (prune_runtime_alias_test_list): New decalaration.
26681 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26683 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26684 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26685 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26686 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26687 (operator==, comp_dr_with_seg_len_pair): Ditto.
26688 (vect_prune_runtime_alias_test_list): Ditto.
26690 2017-05-26 Martin Liska <mliska@suse.cz>
26693 * params.def: Bound partial-inlining-entry-probability param.
26695 2017-05-26 Marek Polacek <polacek@redhat.com>
26698 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26701 2017-05-26 Richard Biener <rguenther@suse.de>
26703 PR tree-optimization/80842
26704 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26707 2017-05-26 Richard Biener <rguenther@suse.de>
26709 PR tree-optimization/80844
26710 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26712 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26714 * doc/md.texi (Machine Constraints): Update x86 family
26715 machine constraints section to match 'config/i386/constraints.md'.
26717 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26719 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26721 2017-05-25 Nathan Sidwell <nathan@acm.org>
26723 * doc/invoke.texi (--enable-languages): Update documentation.
26725 2017-05-25 Martin Liska <mliska@suse.cz>
26727 * dumpfile.c: Add TDF_FOLDING.
26728 * dumpfile.h (enum dump_kind): Likewise.
26729 * genmatch.c (dt_simplify::gen_1): Use it.
26731 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26733 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26735 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26737 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26738 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26740 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26742 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26743 * match.pd (X == C): Rewrite it here.
26744 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26745 with_certain_nonzero_bits2): New predicates.
26746 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26748 2017-05-24 Nathan Sidwell <nathan@acm.org>
26750 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26753 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26756 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26758 * config/powerpcspe: New port. Files are copied from the rs6000
26759 port, with "rs6000" in filenames replaced by "powerpcspe".
26761 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26763 PR rtl-optimization/80754
26764 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26766 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26768 * config/sparc/sparc.md (length): Return the correct value for -mflat
26769 sibcalls to match output_sibcall.
26771 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26775 * config/rs6000/rs6000.c (struct machine_function): Add new field
26777 (rs6000_get_separate_components): Init that field, use it.
26778 (rs6000_components_for_bb): Use the field.
26780 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26782 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26784 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26786 PR middle-end/80823
26787 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26789 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26792 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26793 address against address_operand predicate.
26794 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26796 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26798 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26799 parameters passed indirectly.
26801 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26803 * config/i386/i386.md (*movdi_internal): Remove SSE4
26804 alternative 18 (?r, *v). Update insn attributes.
26805 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26806 Update insn attributes.
26807 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26808 Update insn attributes.
26809 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26810 alternative 1 (r, v). Remove isa attribute.
26811 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26812 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26813 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26815 2017-05-23 Tom de Vries <tom@codesourcery.com>
26817 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26820 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26822 * cgraphunit.c (symbol_table::process_new_functions): Update.
26823 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26824 (inline_generate_summary): Rename to ...
26825 (ipa_fn_summary_generate): ... this one.
26826 (inline_read_summary): Rename to ...
26827 (ipa_fn_summary_read): ... this one.
26828 (inline_write_summary): Rename to ...
26829 (ipa_fn_summary_write): ... this one.
26830 (inline_free_summary): Rename to ...
26831 (ipa_free_fn_summary): ... this one.
26832 (pass_data_local_fn_summary, pass_local_fn_summary,
26833 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26834 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26835 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26836 make_pass_ipa_fn_summary): New.
26837 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26838 inline_write_summary, inline_free_summary): Remove.
26839 (ipa_free_fn_summary) : New.
26840 * ipa-inline.c (ipa_inline): Update.
26841 (pass_ipa_inline): Do not generate summaries.
26842 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26844 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26845 and add pass_ipa_fn_summary.
26846 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26848 (make_pass_inline_parameters): Remove.
26850 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26852 * omp-low.c (struct omp_context): Remove "default_kind" member.
26855 * omp-offload.c (execute_oacc_device_lower): Remove the
26856 parallelism dimensions function attributes for unparallelized
26857 OpenACC kernels constructs.
26859 2017-05-23 Martin Liska <mliska@suse.cz>
26861 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26863 (cgraph_edge::make_speculative): Likewise.
26864 (cgraph_edge::resolve_speculation): Likewise.
26865 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26866 (cgraph_node::dump): Likewise.
26867 * cgraph.h: Likewise.
26868 * cgraphunit.c (analyze_functions): Likewise.
26869 (symbol_table::compile): Likewise.
26870 * ipa-cp.c (print_all_lattices): Likewise.
26871 (determine_versionability): Likewise.
26872 (initialize_node_lattices): Likewise.
26873 (ipcp_verify_propagated_values): Likewise.
26874 (estimate_local_effects): Likewise.
26875 (update_profiling_info): Likewise.
26876 (create_specialized_node): Likewise.
26877 (perhaps_add_new_callers): Likewise.
26878 (decide_about_value): Likewise.
26879 (decide_whether_version_node): Likewise.
26880 (identify_dead_nodes): Likewise.
26881 (ipcp_store_bits_results): Likewise.
26882 * ipa-devirt.c (dump_targets): Likewise.
26883 (ipa_devirt): Likewise.
26884 * ipa-icf.c (sem_item::dump): Likewise.
26885 (sem_function::equals): Likewise.
26886 (sem_variable::equals): Likewise.
26887 (sem_item_optimizer::read_section): Likewise.
26888 (sem_item_optimizer::execute): Likewise.
26889 (congruence_class::dump): Likewise.
26890 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26891 (dump_inline_summary): Likewise.
26892 (estimate_node_size_and_time): Likewise.
26893 (inline_analyze_function): Likewise.
26894 * ipa-inline-transform.c (inline_call): Likewise.
26895 * ipa-inline.c (report_inline_failed_reason): Likewise.
26896 (want_early_inline_function_p): Likewise.
26897 (edge_badness): Likewise.
26898 (update_edge_key): Likewise.
26899 (inline_small_functions): Likewise.
26900 * ipa-profile.c (ipa_profile): Likewise.
26901 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26902 (ipa_make_edge_direct_to_target): Likewise.
26903 (remove_described_reference): Likewise.
26904 (ipa_impossible_devirt_target): Likewise.
26905 (propagate_controlled_uses): Likewise.
26906 (ipa_print_node_params): Likewise.
26907 (ipcp_transform_function): Likewise.
26908 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26909 (propagate_pure_const): Likewise.
26910 * ipa-reference.c (generate_summary): Likewise.
26911 (read_write_all_from_decl): Likewise.
26912 (propagate): Likewise.
26913 (ipa_reference_read_optimization_summary): Likewise.
26914 * ipa-utils.c (ipa_merge_profiles): Likewise.
26915 * ipa.c (walk_polymorphic_call_targets): Likewise.
26916 (symbol_table::remove_unreachable_nodes): Likewise.
26917 (ipa_single_use): Likewise.
26918 * passes.c (execute_todo): Likewise.
26919 * predict.c (drop_profile): Likewise.
26920 * symtab.c (symtab_node::get_dump_name): New function.
26921 (symtab_node::dump_name): Likewise.
26922 (symtab_node::dump_asm_name): Likewise.
26923 (symtab_node::dump_references): Likewise.
26924 (symtab_node::dump_referring): Likewise.
26925 (symtab_node::dump_base): Likewise.
26926 (symtab_node::debug_symtab): Likewise.
26927 * tree-sra.c (convert_callers_for_node): Likewise.
26928 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26929 * value-prof.c (init_node_map): Likewise.
26931 2017-05-23 Martin Liska <mliska@suse.cz>
26933 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26934 and symtab_node::debug_symtab to symbol_table::debug.
26935 * cgraphunit.c (analyze_functions): Use the renamed function.
26936 (symbol_table::compile): Likewise.
26937 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26938 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26939 * passes.c (execute_todo): Likewise.
26940 * symtab.c (symbol_table::dump): New function.
26941 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26943 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26945 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26946 that nonconst implies exec.
26948 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26950 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26951 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26952 (inline_edge_summary_vec): Turn into ...
26953 (ipa_call_summaries): ... this one.
26954 (redirect_to_unreachable, edge_set_predicate,
26955 evaluate_properties_for_edge, inline_summary_alloc,
26956 reset_ipa_call_summary, reset_inline_summary,
26957 inline_summary_t::duplicate): Update.
26958 (inline_edge_duplication_hook): Turn to ...
26959 (ipa_call_summary_t::duplicate): ... this one.
26960 (inline_edge_removal_hook): Turn to ...
26961 (ipa_call_summary_t::remove): ... this one.
26962 (dump_inline_edge_summary): Turn to ...
26963 (dump_ipa_call_summary): ... this one.
26964 (estimate_function_body_sizes): Update.
26965 (inline_update_callee_summaries): Update.
26966 (remap_edge_change_prob): Update.
26967 (remap_edge_summaries): Update.
26968 (inline_merge_summary): Update.
26969 (do_estimate_edge_time): Update.
26970 (inline_generate_summary): Update.
26971 (inline_read_section): Update.
26972 (inline_read_summary): Update.
26973 (inline_free_summary): Update.
26974 * ipa-inline.c (can_inline_edge_p): Update.
26975 (compute_inlined_call_time): Update.
26976 (want_inline_small_function_p): Update.
26977 (edge_badness): Update.
26978 (early_inliner): Update.
26979 * ipa-inline.h (inline_edge_summary): Turn to ...
26980 (ipa_call_summary): ... this one.
26981 (ipa_call_summary_t): New class.
26982 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26983 (ipa_call_summaries): New.
26984 (inline_edge_summary): Remove.
26985 (estimate_edge_growth): Update.
26986 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26987 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26988 * ipa-split.c (execute_split_functions): Update.
26989 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26991 2017-05-23 Tom de Vries <tom@codesourcery.com>
26993 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26994 attributes): Document rdrand effective target.
26996 2017-05-23 Tom de Vries <tom@codesourcery.com>
26998 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26999 attributes): Sort alphabetically.
27001 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27003 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27005 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27008 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27009 V2DF/V2DI splat into two separate patterns, one that handles
27010 registers, and the other that only handles memory. Drop support
27011 for splatting from a GPR on ISA 2.07 and then splitting the
27012 splat into direct move and splat.
27013 (vsx_splat_<mode>_reg): Likewise.
27014 (vsx_splat_<mode>_mem): Likewise.
27016 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27018 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27020 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27022 PR middle-end/80809
27023 * omp-low.c (finish_taskreg_remap): New function.
27024 (finish_taskreg_scan): If unit size of ctx->record_type
27025 is non-constant, unshare the size expression and replace
27026 decls in it with possible outer var refs.
27028 PR middle-end/80809
27029 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27030 GOVD_SHARED rather than GOVD_PRIVATE with it.
27031 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27032 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27034 PR middle-end/80853
27035 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27036 as last argument to build_outer_var_ref for pointer bases of array
27037 section reductions.
27039 2017-05-19 Martin Sebor <msebor@redhat.com>
27041 * print-tree.c (print_node): Print DECL_READ_P flag.
27043 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27045 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27046 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27047 * cgraph.c: Likewise.
27048 * cgraphunit.c: Likewise.
27049 * gengtype.c: Likewise.
27050 * ipa-cp.c: Likewise.
27051 * ipa-devirt.c: Likewise.
27052 * ipa-icf.c: Likewise.
27053 * ipa-predicate.c: Likewise.
27054 * ipa-profile.c: Likewise.
27055 * ipa-prop.c: Likewise.
27056 * ipa-split.c: Likewise.
27058 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27059 edge_predicate_pool, dump_inline_hints,
27060 inline_summary::account_size_time, redirect_to_unreachable,
27061 edge_set_predicate, set_hint_predicate,
27062 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27063 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27064 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27065 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27066 ipa_call_summary_t::remove, initialize_growth_caches,
27067 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27068 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27069 mark_modified, unmodified_parm_1, unmodified_parm,
27070 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27071 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27072 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27073 will_be_nonconstant_predicate, record_modified_bb_info,
27074 get_minimal_bb, record_modified, param_change_prob,
27075 phi_result_unknown_predicate, predicate_for_phi_result,
27076 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27077 estimate_function_body_sizes, compute_inline_parameters,
27078 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27079 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27080 inline_update_callee_summaries, remap_edge_change_prob,
27081 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27082 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27083 inline_analyze_function, inline_summary_t::insert,
27084 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27085 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27086 inline_free_summary): Move to ipa-fnsummary.h
27087 (predicate_t): Remove.
27088 * ipa-fnsummary.c: New file.
27089 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27090 (enum inline_hints_vals, inline_hints, agg_position_info,
27091 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27092 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27093 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27094 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27095 inline_read_summary, inline_write_summary, inline_free_summary,
27096 inline_analyze_function, initialize_inline_failed,
27097 inline_merge_summary, inline_update_overall_summary,
27098 compute_inline_parameters): Move to ipa-fnsummary.h
27099 * ipa-fnsummary.h: New file.
27100 * ipa-inline-transform.h: Include ipa-inline.h.
27101 * ipa-inline.c: LIkewise.
27103 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27105 * ipa-inline.c (edge_badness): Use inlined_time instead of
27106 inline_summaries->get.
27108 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27110 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27112 2017-05-22 Nathan Sidwell <nathan@acm.org>
27114 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27115 (fdump-lang): Document 'raw' option.
27116 * dumpfile.h (TDI_tu): Delete.
27117 * dumpfile.c (dump_files): Remove translation-unit.
27118 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27120 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27122 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27123 command option from $(AWK) call.
27124 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27125 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27126 [FORMAT]: Remove handling of variable.
27127 * config/avr/t-multilib: Regenerate.
27129 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27131 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27133 (dump_inline_summary): Do not print self_time.
27134 (estimate_function_body_sizes): Do not set self_time.
27135 (compute_inline_parameters): Likewise.
27136 (inline_read_section, inline_write_summary): Do not stream self_time.
27137 * ipa-inline.h (inline_summary): Drop self_time.
27139 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27141 * ipa-inline-analysis.c (account_size_time): Rename to ...
27142 (inline_summary::account_size_time): ... this one.
27143 (reset_ipa_call_summary): Turn to ...
27144 (ipa_call_summary::reset): ... this one.
27145 (reset_inline_summary): Turn to ...
27146 (inline_summary::reset): ... this one.
27147 (inline_summary_t::remove): Update.
27148 (inline_summary_t::duplicate): Update.
27149 (ipa_call_summary_t::remove): Update.
27150 (dump_inline_summary): Update.
27151 (estimate_function_body_sizes): Update.
27152 (compute_inline_parameters): Update.
27153 (estimate_node_size_and_time): Update.
27154 (inline_merge_summary): Update.
27155 (inline_update_overall_summary): Update.
27156 (inline_read_section): Update.
27157 (inline_write_summary): Update.
27158 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27159 add account_size_time and reset member functions.
27160 (ipa_call_summary): Add reset function.
27161 * ipa-predicate.h (predicate::operator &): Constify.
27163 2017-05-22 Richard Biener <rguenther@suse.de>
27165 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27167 2017-05-19 Jason Merrill <jason@redhat.com>
27169 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27171 2017-05-19 Marek Polacek <polacek@redhat.com>
27174 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27175 TYPE_OVERFLOW_WRAPS checks.
27177 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27179 * tree-core.h (enum omp_clause_default_kind): Add
27180 "OMP_CLAUSE_DEFAULT_PRESENT".
27181 * tree-pretty-print.c (dump_omp_clause): Handle it.
27182 * gimplify.c (enum gimplify_omp_var_data): Add
27183 "GOVD_MAP_FORCE_PRESENT".
27184 (gimplify_adjust_omp_clauses_1): Map it to
27185 "GOMP_MAP_FORCE_PRESENT".
27186 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27188 * gimplify.c (oacc_default_clause): Clarify.
27190 2017-05-19 Nathan Sidwell <nathan@acm.org>
27192 LANG_HOOK_REGISTER_DUMPS
27193 * toplev.c (general_init): Call register dump lang hook.
27194 * doc/invoke.texi: Document -fdump-lang option family.
27195 * dumpfile.c (dump_files): Remove class dump here.
27196 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27197 * dumpfile.h (tree_dump_index): Remove TDI_class.
27198 * langhooks-def.h (lhd_register_dumps): Declare.
27199 (LANG_HOOKS_REGISTER_DUMPS): Define.
27200 (LANG_HOOKS_INITIALIZER): Add it.
27201 * langhooks.c (lhd_register_dumps): Define.
27202 * langhooks.h (struct lang_hooks): Add register_dumps.
27204 2017-05-19 Nathan Sidwell <nathan@acm.org>
27206 * context.h (context::set_passes): New.
27207 * context.c (context::context): Do not create pass manager.
27208 * toplev.c (general_init): Create pass manager here.
27210 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27212 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27213 use this splitter if two add or or instructions would also work for
27214 the constant we want to generate.
27216 2017-05-19 Richard Biener <rguenther@suse.de>
27219 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27220 predicate evaluation.
27222 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27224 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27226 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27229 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27231 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27232 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27233 (inline_edge_summary_vec): Turn into ...
27234 (ipa_call_summaries): ... this one.
27235 (redirect_to_unreachable, edge_set_predicate,
27236 evaluate_properties_for_edge, inline_summary_alloc,
27237 reset_ipa_call_summary, reset_inline_summary,
27238 inline_summary_t::duplicate): Update.
27239 (inline_edge_duplication_hook): Turn to ...
27240 (ipa_call_summary_t::duplicate): ... this one.
27241 (inline_edge_removal_hook): Turn to ...
27242 (ipa_call_summary_t::remove): ... this one.
27243 (dump_inline_edge_summary): Turn to ...
27244 (dump_ipa_call_summary): ... this one.
27245 (estimate_function_body_sizes): Update.
27246 (inline_update_callee_summaries): Update.
27247 (remap_edge_change_prob): Update.
27248 (remap_edge_summaries): Update.
27249 (inline_merge_summary): Update.
27250 (do_estimate_edge_time): Update.
27251 (inline_generate_summary): Update.
27252 (inline_read_section): Update.
27253 (inline_read_summary): Update.
27254 (inline_free_summary): Update.
27255 * ipa-inline.c (can_inline_edge_p): Update.
27256 (compute_inlined_call_time): Update.
27257 (want_inline_small_function_p): Update.
27258 (edge_badness): Update.
27259 (early_inliner): Update.
27260 * ipa-inline.h (inline_edge_summary): Turn to ...
27261 (ipa_call_summary): ... this one.
27262 (ipa_call_summary_t): New class.
27263 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27264 (ipa_call_summaries): New.
27265 (inline_edge_summary): Remove.
27266 (estimate_edge_growth): Update.
27267 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27268 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27269 * ipa-split.c (execute_split_functions): Update.
27270 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27272 2017-05-19 Richard Biener <rguenther@suse.de>
27274 PR middle-end/80764
27275 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27277 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27279 * config/rs6000/rs6000.c (struct machine_function): Add field
27280 fpr_is_wrapped_separately.
27281 (rs6000_get_separate_components): Use 64 components. Handle the
27282 new FPR components.
27283 (rs6000_components_for_bb): Handle the FPR components.
27284 (rs6000_emit_prologue_components): Handle the FPR components.
27285 (rs6000_emit_epilogue_components): Handle the FPR components.
27286 (rs6000_set_handled_components): Handle the FPR components.
27287 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27288 that are already separately shrink-wrapped.
27289 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27290 that are already separately shrink-wrapped.
27292 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27295 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27298 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27299 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27300 cases where the register allocator uses a move and an offsettable
27301 memory operation to/from a FPR register on ISA 2.06/2.07.
27302 (define_peephole2 for Altivec d-form store): Likewise.
27304 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27307 * config/i386/mmx.md (*mov<mode>_internal): Enable
27308 alternatives 11, 12, 13 and 14 also for 32bit targets.
27309 Remove alternatives 15, 16, 17 and 18.
27310 * config/i386/sse.md (vec_concatv2di): Change
27311 alternative (!x, *y) to (x, ?!*Yn).
27313 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27315 * dumpfile.h (enum dump_kind): Remove stray comma.
27317 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27319 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27320 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27321 predicate::num_conditions
27322 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27323 (CHANGED): turn into predicate::changed.
27324 (agg_position_info): Move to ipa-predicate.h
27325 (add_condition, predicate::add_clause, predicate::operator &=,
27326 predicate::or_with, predicate::evaluate, predicate::probability,
27327 dump_condition, dump_clause, predicate::dump,
27328 predicate::remap_after_duplication, predicate::remap_after_inlining,
27329 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27330 (evaluate_conditions_for_known_args): Update.
27331 (set_cond_stmt_execution_predicate): Update.
27332 * ipa-inline.h: Include ipa-predicate.h
27333 (condition, inline_param_summary, conditions, agg_position_info,
27334 predicate): Move to ipa-predicate.h
27335 * ipa-predicate.c: New file.
27336 * ipa-predicate.h: New file.
27338 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27340 * final.c (leaf_function_p): Check we are not in a sequence.
27342 2017-05-18 Martin Liska <mliska@suse.cz>
27344 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27345 * dumpfile.c (dump_register): Use new enum dump_kind.
27346 (get_dump_file_name): Likewise.
27347 (dump_enable_all): Likewise.
27348 (dump_switch_p_1): Likewise.
27349 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27350 * dumpfile.h (enum dump_kind): New enum type.
27351 (struct dump_file_info): Create constructor and
27352 format fields and comments.
27353 * passes.c (pass_manager::register_one_dump_file):
27355 * statistics.c (statistics_early_init): Likewise.
27356 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27357 TDF_TREE with TDF_SLIM.
27358 (gather_memory_references_ref): Likewise.
27360 2017-05-18 Martin Liska <mliska@suse.cz>
27362 * vec.h (struct vnull): Use it.
27364 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27366 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27367 (true_predicate, false_predicate, true_predicate_p,
27368 false_predicate_p): Remove.
27369 (single_cond_predicate, not_inlined_predicate): Turn to member function
27371 (add_condition): Update.
27372 (add_clause): Turn to...
27373 (predicate::add_clause): ... this one; update; allow passing NULL
27375 (and_predicates): Turn to ...
27376 (predicate::operator &=): ... this one.
27377 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27378 (or_predicates): Turn to ...
27379 (predicate::or_with): ... this one.
27380 (evaluate_predicate): Turn to ...
27381 (predicate::evaluate): ... this one.
27382 (predicate_probability): Turn to ...
27383 (predicate::probability): ... this one.
27384 (dump_condition): Update.
27385 (dump_predicate): Turn to ...
27386 (predicate::dump): ... this one.
27387 (account_size_time): Update.
27388 (edge_set_predicate): Update.
27389 (set_hint_predicate): UPdate.
27390 (evaluate_conditions_for_known_args): Update.
27391 (evaluate_properties_for_edge): Update.
27392 (remap_predicate_after_duplication): Turn to...
27393 (predicate::remap_after_duplication): ... this one.
27394 (remap_hint_predicate_after_duplication): Update.
27395 (inline_summary_t::duplicate): UPdate.
27396 (dump_inline_edge_summary): Update.
27397 (dump_inline_summary): Update.
27398 (set_cond_stmt_execution_predicate): Update.
27399 (set_switch_stmt_execution_predicate): Update.
27400 (compute_bb_predicates): Update.
27401 (will_be_nonconstant_expr_predicate): Update.
27402 (will_be_nonconstant_predicate): Update.
27403 (phi_result_unknown_predicate): Update.
27404 (predicate_for_phi_result): Update.
27405 (array_index_predicate): Update.
27406 (estimate_function_body_sizes): Update.
27407 (estimate_node_size_and_time): Update.
27408 (estimate_ipcp_clone_size_and_time): Update.
27409 (remap_predicate): Rename to ...
27410 (predicate::remap_after_inlining): ... this one.
27411 (remap_hint_predicate): Update.
27412 (inline_merge_summary): Update.
27413 (inline_update_overall_summary): Update.
27414 (estimate_size_after_inlining): Update.
27415 (read_predicate): Rename to ...
27416 (predicate::stream_in): ... this one.
27417 (read_inline_edge_summary): Update.
27418 (write_predicate): Rename to ...
27419 (predicate::stream_out): ... this one.
27420 (write_inline_edge_summary): Update.
27421 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27422 (clause_t): Turn to uint32_t
27423 (predicate): Turn to class; implement constructor and operators
27425 (size_time_entry): Update.
27426 (inline_summary): Update.
27427 (inline_edge_summary): Update.
27429 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27431 * fold-const.c (fold_binary_loc): Move transformation...
27432 * match.pd (C - X CMP X): ... here.
27434 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27436 * config/sparc/sparc.c (sparc_option_override): Set function
27437 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27438 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27440 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27442 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27444 2017-05-18 Marek Polacek <polacek@redhat.com>
27447 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27448 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27449 gimple_assign_load_p.
27451 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27453 PR middle-end/80692
27454 * real.c (do_compare): Give decimal_do_compare preference over
27455 comparing just the signs.
27457 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27459 * doc/md.texi (Canonicalization of Instructions): Describe the
27460 canonical form of instructions that inherently set a condition
27463 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27465 PR middle-end/80775
27466 * tree-cfg.c: Move deletion of unreachable case statements to after
27467 the merging of consecutive case labels.
27469 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27471 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27472 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27473 restoring of callee-saved registers.
27475 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27477 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27478 * config/visium/visium.c (single_set_and_flags): Likewise.
27479 * config/visium/visium.md (Substitutions): Likewise.
27481 2017-05-17 Martin Liska <mliska@suse.cz>
27483 * cfg.c: Introduce dump_flags_t type and
27484 use it instead of int type.
27486 * cfghooks.c: Likewise.
27487 * cfghooks.h (struct cfg_hooks): Likewise.
27488 * cfgrtl.c: Likewise.
27489 * cfgrtl.h: Likewise.
27490 * cgraph.c (cgraph_node::get_body): Likewise.
27491 * coretypes.h: Likewise.
27492 * domwalk.c: Likewise.
27493 * domwalk.h: Likewise.
27494 * dumpfile.c (struct dump_option_value_info): Likewise.
27495 (dump_enable_all): Likewise.
27496 (dump_switch_p_1): Likewise.
27497 (opt_info_switch_p): Likewise.
27498 * dumpfile.h (enum tree_dump_index): Likewise.
27499 (struct dump_file_info): Likewise.
27500 * genemit.c: Likewise.
27501 * generic-match-head.c: Likewise.
27502 * gengtype.c (open_base_files): Likewise.
27503 * gimple-pretty-print.c: Likewise.
27504 * gimple-pretty-print.h: Likewise.
27505 * graph.c (print_graph_cfg): Likewise.
27506 * graphite-scop-detection.c (dot_all_sese): Likewise.
27507 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27508 * loop-unroll.c (report_unroll): Likewise.
27509 * passes.c (pass_manager::register_one_dump_file): Likewise.
27510 * print-tree.c: Likewise.
27511 * statistics.c: Likewise.
27512 * tree-cfg.c: Likewise.
27513 * tree-cfg.h: Likewise.
27514 * tree-dfa.c: Likewise.
27515 * tree-dfa.h: Likewise.
27516 * tree-dump.c (dump_function): Likewise.
27517 * tree-dump.h (struct dump_info): Likewise.
27518 * tree-pretty-print.c: Likewise.
27519 * tree-pretty-print.h: Likewise.
27520 * tree-ssa-live.c: Likewise.
27521 * tree-ssa-live.h: Likewise.
27522 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27523 * tree-vect-loop.c: Likewise.
27524 * tree-vect-slp.c: Likewise.
27526 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27527 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27529 PR tree-optimization/80457
27530 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27531 of all arguments to a statement as scalar_to_vec operations.
27532 (vectorizable_call): Adjust call to vect_model_simple_cost for
27534 (vectorizable_conversion): Likewise.
27535 (vectorizable_assignment): Likewise.
27536 (vectorizable_shift): Likewise.
27537 (vectorizable_operation): Likewise.
27538 (vectorizable_comparison): Likewise.
27539 (vect_is_simple_cond): Record the def types for operands.
27540 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27541 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27542 for statement argument count.
27544 2017-05-16 Carl Love <cel@us.ibm.com>
27546 * config/rs6000/rs6000-c: Add support for built-in functions
27547 vector unsigned long long vec_bperm (vector unsigned long long,
27548 vector unsigned char)
27549 vector signed long long vec_mule (vector signed int,
27551 vector unsigned long long vec_mule (vector unsigned int,
27552 vector unsigned int)
27553 vector signed long long vec_mulo (vector signed int,
27555 vector unsigned long long vec_mulo (vector unsigned int,
27556 vector unsigned int)
27557 vector signed char vec_sldw (vector signed char,
27558 vector signed char,
27560 vector unsigned char vec_sldw (vector unsigned char,
27561 vector unsigned char,
27563 vector signed short vec_sldw (vector signed short,
27564 vector signed short,
27566 vector unsigned short vec_sldw (vector unsigned short,
27567 vector unsigned short,
27569 vector signed int vec_sldw (vector signed int,
27572 vector unsigned int vec_sldw (vector unsigned int,
27573 vector unsigned int,
27575 vector signed long long vec_sldw (vector signed long long,
27576 vector signed long long,
27578 vector unsigned long long vec_sldw (vector unsigned long long,
27579 vector unsigned long long,
27581 * config/rs6000/rs6000-c: Add support for built-in functions
27582 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27583 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27584 * doc/extend.texi: Update the built-in documentation for the
27585 new built-in functions.
27587 2017-05-16 Marek Polacek <polacek@redhat.com>
27591 * tree.c (save_expr): Don't fold the expression.
27593 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27595 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27596 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27597 and (?*y,m). Update insn attributes.
27599 2017-05-16 Martin Liska <mliska@suse.cz>
27601 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27602 flags argument of print_gimple_stmt, print_gimple_expr,
27603 print_generic_stmt and print_generic_expr.
27604 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27605 * coretypes.h: Likewise.
27606 * except.c (dump_eh_tree): Likewise.
27607 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27608 * gimple-pretty-print.h: Likewise.
27609 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27610 (backprop::push_to_worklist): Likewise.
27611 (backprop::pop_from_worklist): Likewise.
27612 (backprop::process_use): Likewise.
27613 (backprop::intersect_uses): Likewise.
27614 (note_replacement): Likewise.
27615 * gimple-ssa-store-merging.c
27616 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27617 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27618 (pass_store_merging::execute): Likewise.
27619 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27620 (ssa_base_cand_dump_callback): Likewise.
27621 (dump_incr_vec): Likewise.
27622 (replace_refs): Likewise.
27623 (replace_mult_candidate): Likewise.
27624 (create_add_on_incoming_edge): Likewise.
27625 (create_phi_basis): Likewise.
27626 (insert_initializers): Likewise.
27627 (all_phi_incrs_profitable): Likewise.
27628 (introduce_cast_before_cand): Likewise.
27629 (replace_one_candidate): Likewise.
27630 * gimplify.c (gimplify_expr): Likewise.
27631 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27632 (set_rename): Likewise.
27633 (rename_uses): Likewise.
27634 (copy_loop_phi_nodes): Likewise.
27635 (add_close_phis_to_merge_points): Likewise.
27636 (copy_loop_close_phi_args): Likewise.
27637 (copy_cond_phi_args): Likewise.
27638 (graphite_copy_stmts_from_block): Likewise.
27639 (translate_pending_phi_nodes): Likewise.
27640 * graphite-poly.c (print_pdr): Likewise.
27641 (dump_gbb_cases): Likewise.
27642 (dump_gbb_conditions): Likewise.
27643 (print_scop_params): Likewise.
27644 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27645 (build_cross_bb_scalars_use): Likewise.
27646 (gather_bbs::before_dom_children): Likewise.
27647 * hsa-dump.c (dump_hsa_immed): Likewise.
27648 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27649 (get_replacement_map): Likewise.
27650 * ipa-inline-analysis.c (dump_condition): Likewise.
27651 (estimate_function_body_sizes): Likewise.
27652 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27653 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27654 * ipa-prop.c (ipa_dump_param): Likewise.
27655 (ipa_print_node_jump_functions_for_edge): Likewise.
27656 (ipa_modify_call_arguments): Likewise.
27657 (ipa_modify_expr): Likewise.
27658 (ipa_dump_param_adjustments): Likewise.
27659 (ipa_dump_agg_replacement_values): Likewise.
27660 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27661 * ipa-pure-const.c (check_stmt): Likewise.
27662 (pass_nothrow::execute): Likewise.
27663 * ipa-split.c (execute_split_functions): Likewise.
27664 * omp-offload.c (dump_oacc_loop_part): Likewise.
27665 (dump_oacc_loop): Likewise.
27666 * trans-mem.c (tm_log_emit): Likewise.
27667 (tm_memopt_accumulate_memops): Likewise.
27668 (dump_tm_memopt_set): Likewise.
27669 (dump_tm_memopt_transform): Likewise.
27670 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27671 (print_loop): Likewise.
27672 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27673 (chkp_gather_checks_info): Likewise.
27674 (chkp_get_check_result): Likewise.
27675 (chkp_remove_check_if_pass): Likewise.
27676 (chkp_use_outer_bounds_if_possible): Likewise.
27677 (chkp_reduce_bounds_lifetime): Likewise.
27678 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27679 (chkp_mark_completed_bounds): Likewise.
27680 (chkp_register_incomplete_bounds): Likewise.
27681 (chkp_mark_invalid_bounds): Likewise.
27682 (chkp_maybe_copy_and_register_bounds): Likewise.
27683 (chkp_build_returned_bound): Likewise.
27684 (chkp_get_bound_for_parm): Likewise.
27685 (chkp_build_bndldx): Likewise.
27686 (chkp_get_bounds_by_definition): Likewise.
27687 (chkp_generate_extern_var_bounds): Likewise.
27688 (chkp_get_bounds_for_decl_addr): Likewise.
27689 * tree-chrec.c (chrec_apply): Likewise.
27690 * tree-data-ref.c (dump_data_reference): Likewise.
27691 (dump_subscript): Likewise.
27692 (dump_data_dependence_relation): Likewise.
27693 (analyze_overlapping_iterations): Likewise.
27694 * tree-inline.c (expand_call_inline): Likewise.
27695 (tree_function_versioning): Likewise.
27696 * tree-into-ssa.c (dump_defs_stack): Likewise.
27697 (dump_currdefs): Likewise.
27698 (dump_names_replaced_by): Likewise.
27699 (dump_update_ssa): Likewise.
27700 (update_ssa): Likewise.
27701 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27702 * tree-parloops.c (build_new_reduction): Likewise.
27703 (try_create_reduction_list): Likewise.
27704 (ref_conflicts_with_region): Likewise.
27705 (oacc_entry_exit_ok_1): Likewise.
27706 (oacc_entry_exit_single_gang): Likewise.
27707 * tree-pretty-print.h: Likewise.
27708 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27709 (get_scalar_evolution): Likewise.
27710 (add_to_evolution): Likewise.
27711 (get_loop_exit_condition): Likewise.
27712 (analyze_evolution_in_loop): Likewise.
27713 (analyze_initial_condition): Likewise.
27714 (analyze_scalar_evolution): Likewise.
27715 (instantiate_scev): Likewise.
27716 (number_of_latch_executions): Likewise.
27717 (gather_chrec_stats): Likewise.
27718 (final_value_replacement_loop): Likewise.
27719 (scev_const_prop): Likewise.
27720 * tree-sra.c (dump_access): Likewise.
27721 (disqualify_candidate): Likewise.
27722 (create_access): Likewise.
27723 (reject): Likewise.
27724 (maybe_add_sra_candidate): Likewise.
27725 (create_access_replacement): Likewise.
27726 (analyze_access_subtree): Likewise.
27727 (analyze_all_variable_accesses): Likewise.
27728 (sra_modify_assign): Likewise.
27729 (initialize_constant_pool_replacements): Likewise.
27730 (find_param_candidates): Likewise.
27731 (decide_one_param_reduction): Likewise.
27732 (replace_removed_params_ssa_names): Likewise.
27733 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27734 * tree-ssa-copy.c (dump_copy_of): Likewise.
27735 (copy_prop_visit_cond_stmt): Likewise.
27736 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27737 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27738 (record_equivalences_from_stmt): Likewise.
27739 * tree-ssa-dse.c (compute_trims): Likewise.
27740 (delete_dead_call): Likewise.
27741 (delete_dead_assignment): Likewise.
27742 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27743 (forward_propagate_into_cond): Likewise.
27744 (pass_forwprop::execute): Likewise.
27745 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27746 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27748 (move_computations_worker): Likewise.
27749 (execute_sm): Likewise.
27750 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27751 (remove_exits_and_undefined_stmts): Likewise.
27752 (remove_redundant_iv_tests): Likewise.
27753 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27754 (adjust_iv_update_pos): Likewise.
27755 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27756 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27757 (value_replacement): Likewise.
27758 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27759 * tree-ssa-pre.c (print_pre_expr): Likewise.
27760 (get_representative_for): Likewise.
27761 (create_expression_by_pieces): Likewise.
27762 (insert_into_preds_of_block): Likewise.
27763 (eliminate_insert): Likewise.
27764 (eliminate_dom_walker::before_dom_children): Likewise.
27765 (eliminate): Likewise.
27766 (remove_dead_inserted_code): Likewise.
27767 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27768 * tree-ssa-reassoc.c (get_rank): Likewise.
27769 (eliminate_duplicate_pair): Likewise.
27770 (eliminate_plus_minus_pair): Likewise.
27771 (eliminate_not_pairs): Likewise.
27772 (undistribute_ops_list): Likewise.
27773 (eliminate_redundant_comparison): Likewise.
27774 (update_range_test): Likewise.
27775 (optimize_range_tests_var_bound): Likewise.
27776 (optimize_vec_cond_expr): Likewise.
27777 (rewrite_expr_tree): Likewise.
27778 (rewrite_expr_tree_parallel): Likewise.
27779 (linearize_expr): Likewise.
27780 (break_up_subtract): Likewise.
27781 (linearize_expr_tree): Likewise.
27782 (attempt_builtin_powi): Likewise.
27783 (attempt_builtin_copysign): Likewise.
27784 (transform_stmt_to_copy): Likewise.
27785 (transform_stmt_to_multiply): Likewise.
27786 (dump_ops_vector): Likewise.
27787 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27788 (print_scc): Likewise.
27789 (set_ssa_val_to): Likewise.
27790 (visit_reference_op_store): Likewise.
27791 (visit_use): Likewise.
27792 (sccvn_dom_walker::before_dom_children): Likewise.
27793 (run_scc_vn): Likewise.
27794 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27796 (expr_hash_elt::print): Likewise.
27797 (const_and_copies::pop_to_marker): Likewise.
27798 (const_and_copies::record_const_or_copy_raw): Likewise.
27799 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27800 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27801 (dump_predicates): Likewise.
27802 (find_uninit_use): Likewise.
27803 (warn_uninitialized_phi): Likewise.
27804 (pass_late_warn_uninitialized::execute): Likewise.
27805 * tree-ssa.c (verify_vssa): Likewise.
27806 (verify_ssa): Likewise.
27807 (maybe_optimize_var): Likewise.
27808 * tree-vrp.c (dump_value_range): Likewise.
27809 (dump_all_value_ranges): Likewise.
27810 (dump_asserts_for): Likewise.
27811 (register_edge_assert_for_2): Likewise.
27812 (vrp_visit_cond_stmt): Likewise.
27813 (vrp_visit_switch_stmt): Likewise.
27814 (vrp_visit_stmt): Likewise.
27815 (vrp_visit_phi_node): Likewise.
27816 (simplify_cond_using_ranges_1): Likewise.
27817 (fold_predicate_in): Likewise.
27818 (evrp_dom_walker::before_dom_children): Likewise.
27819 (evrp_dom_walker::push_value_range): Likewise.
27820 (evrp_dom_walker::pop_value_range): Likewise.
27821 (execute_early_vrp): Likewise.
27823 2017-05-16 Richard Biener <rguenther@suse.de>
27825 * dwarf2out.c (loc_list_from_tree_1): Do not create
27826 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27828 2017-05-16 Richard Biener <rguenther@suse.de>
27830 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27832 (note_variable_value_in_expr): If we resolved the decl ref
27833 do not push to the stack.
27835 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27837 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27838 operations in fast-math mode.
27839 (vaddq_f16): Likewise.
27840 (vmul_f16): Likewise.
27841 (vmulq_f16): Likewise.
27842 (vsub_f16): Likewise.
27843 (vsubq_f16): Likewise.
27844 * config/arm/neon.md (add<mode>3): New.
27846 (fma:<VH:mode>3): New. Also remove outdated comment.
27849 2017-05-16 Martin Liska <mliska@suse.cz>
27853 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27854 (odr_types_equivalent_p): Likewise.
27856 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27858 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27860 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27863 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27864 non-interunit SSE move alternatives with '?'.
27865 (zero-extendsidi peephole2): New peephole to skip intermediate
27866 general register in SSE zero-extend sequence.
27868 2017-05-15 Jeff Law <law@redhat.com>
27870 * reorg.c (relax_delay_slots): Create a new variable to hold
27871 the temporary target rather than clobbering TARGET_LABEL.
27873 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27874 missing argument to extract_bit_field call.
27875 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27877 2017-05-15 Martin Liska <mliska@suse.cz>
27880 * gcc.c (process_command): Do not allow empty argument of -o option.
27882 2017-05-15 Renlin Li <renlin.li@arm.com>
27884 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27885 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27886 * config/aarch64/constraints.md (Usf): Add long call check.
27887 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27888 (call_value): Likewise.
27889 (sibcall): Likewise.
27890 (sibcall_value): Likewise.
27892 (call_value_insn): New.
27893 (sibcall_insn): Update rtx pattern.
27894 (sibcall_value_insn): Likewise.
27895 (call_internal): Remove.
27896 (call_value_internal): Likewise.
27897 (sibcall_internal): Likewise.
27898 (sibcall_value_internal): Likewise.
27899 (call_reg): Likewise.
27900 (call_symbol): Likewise.
27901 (call_value_reg): Likewise.
27902 (call_value_symbol): Likewise.
27904 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27907 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27909 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27911 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27912 compatible with CCGOCmode and with CCZmode.
27914 2017-05-14 Martin Sebor <msebor@redhat.com>
27916 PR middle-end/77671
27917 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27918 (gimple_fold_builtin_snprintf): Same.
27919 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27920 (gimple_fold_builtin_snprintf): Same.
27921 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27922 of character types.
27923 (is_call_safe): New function.
27924 (try_substitute_return_value): Call it.
27925 (try_simplify_call): New function.
27926 (pass_sprintf_length::handle_gimple_call): Call it.
27928 2017-05-14 Martin Sebor <msebor@redhat.com>
27930 PR middle-end/80669
27931 * builtins.c (expand_builtin_stpncpy): Simplify.
27933 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27935 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27936 * config/i386/i386.h
27937 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27938 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27939 (struct machine_function): Add new members call_ms2sysv,
27940 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27941 (struct machine_frame_state): New fields sp_realigned and
27942 sp_realigned_offset.
27943 * config/i386/i386.c
27944 (enum xlogue_stub): New enum.
27945 (enum xlogue_stub_sets): New enum.
27946 (class xlogue_layout): New class.
27947 (struct ix86_frame): New fields stack_realign_allocate_offset,
27948 stack_realign_offset and outlined_save_offset. Modify comments to
27949 detail stack layout when using out-of-line stubs.
27950 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27951 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27952 -mcall-ms2sysv-xlogues.
27953 (stub_managed_regs): New static variable.
27954 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27955 registers managed by out-of-line stub.
27956 (disable_call_ms2sysv_xlogues): New function.
27957 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27958 m->call_ms2sysv when appropriate and compute frame layout for
27960 (sp_valid_at, fp_valid_at): New inline functions.
27961 (choose_basereg): New function.
27962 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27964 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27965 Use align parameter of choose_baseaddr to generated aligned SSE movs
27967 (pro_epilogue_adjust_stack): Modify to track
27968 machine_frame_state::sp_realigned.
27969 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27970 (ix86_nsaved_sseregs): Likewise.
27971 (ix86_emit_save_regs): Likewise.
27972 (ix86_emit_save_regs_using_mov): Likewise.
27973 (ix86_emit_save_sse_regs_using_mov): Likewise.
27974 (get_scratch_register_on_entry): Likewise.
27975 (gen_frame_set): New function.
27976 (gen_frame_load): Likewise.
27977 (gen_frame_store): Likewise.
27978 (emit_outlined_ms2sysv_save): Likewise.
27979 (emit_outlined_ms2sysv_restore): Likewise.
27980 (ix86_expand_prologue): Modify stack re-alignment code and call
27981 emit_outlined_ms2sysv_save when appropriate.
27982 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27983 parameter rtx_insn *insn, which allows the function to be used to only
27984 generate the notes.
27985 (ix86_expand_epilogue): Modify validity checks of frame and stack
27986 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27987 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27988 * config/i386/predicates.md
27989 (save_multiple): New predicate.
27990 (restore_multiple): Likewise.
27991 * config/i386/sse.md
27992 (save_multiple<mode>): New pattern.
27993 (save_multiple_realign<mode>): Likewise.
27994 (restore_multiple<mode>): Likewise.
27995 (restore_multiple_and_return<mode>): Likewise.
27996 (restore_multiple_leave_return<mode>): Likewise.
27997 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27999 2017-05-14 Julia Koval <julia.koval@intel.com>
28001 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28002 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28003 (__builtin_ia32_xsetbv): New builtins.
28004 * config/i386/i386.c (ix86_expand_special_args_builtin):
28006 (ix86_expand_builtin): Special expand for new intrinsics.
28007 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28008 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28009 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28011 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28013 * cfganal.c (inverted_post_order_compute): Change argument type
28015 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28016 * df-core.c (rest_of_handle_df_initialize): Adjust.
28017 (rest_of_handle_df_finish): Likewise.
28018 (df_analyze_1): Likewise.
28019 (df_analyze): Likewise.
28020 (loop_inverted_post_order_compute): Change argument to be a vec *.
28021 (df_analyze_loop): Adjust.
28022 (df_get_n_blocks): Likewise.
28023 (df_get_postorder): Likewise.
28024 * df.h (struct df_d): Change field to be a vec.
28025 * lcm.c (compute_laterin): Adjust.
28026 (compute_available): Likewise.
28027 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28028 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28029 * tree-ssa-pre.c (compute_antic): Likewise.
28031 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28033 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28034 (depth_first_search::depth_first_search): Change structure init
28035 function to this constructor.
28036 (depth_first_search::add_bb): Rename function to this member.
28037 (depth_first_search::execute): Likewise.
28038 (flow_dfs_compute_reverse_finish): Adjust.
28040 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28042 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28043 (longest_simple_path): Likewise.
28044 * shrink-wrap.c (spread_components): Likewise.
28045 (disqualify_problematic_components): Likewise.
28046 (emit_common_heads_for_components): Likewise.
28047 (emit_common_tails_for_components): Likewise.
28048 (insert_prologue_epilogue_for_components): Likewise.
28050 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28052 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28055 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28057 * df-core.c (df_set_blocks): Start using auto_bitmap.
28058 (df_compact_blocks): Likewise.
28059 * df-problems.c (df_rd_confluence_n): Likewise.
28060 * df-scan.c (df_insn_rescan_all): Likewise.
28061 (df_process_deferred_rescans): Likewise.
28062 (df_update_entry_block_defs): Likewise.
28063 (df_update_exit_block_uses): Likewise.
28064 (df_entry_block_bitmap_verify): Likewise.
28065 (df_exit_block_bitmap_verify): Likewise.
28066 (df_scan_verify): Likewise.
28067 * lra-constraints.c (lra_constraints): Likewise.
28068 (undo_optional_reloads): Likewise.
28069 (lra_undo_inheritance): Likewise.
28070 * lra-remat.c (calculate_gen_cands): Likewise.
28071 (do_remat): Likewise.
28072 * lra-spills.c (assign_spill_hard_regs): Likewise.
28073 (spill_pseudos): Likewise.
28074 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28075 (bitmap_set_subtract_values): Likewise.
28077 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28079 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28080 management with auto_bitmap.
28081 (fix_inter_tick): Likewise.
28082 (fix_recovery_deps): Likewise.
28083 * ira.c (add_store_equivs): Likewise.
28084 (find_moveable_pseudos): Likewise.
28085 (split_live_ranges_for_shrink_wrap): Likewise.
28086 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28087 (rtx_reuse_manager::seen_def_p): Likewise.
28088 (rtx_reuse_manager::set_seen_def): Likewise.
28089 * print-rtl.h (class rtx_reuse_manager): Likewise.
28091 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28093 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28095 (migrate_btr_def): Likewise.
28096 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28097 * df-core.c (loop_post_order_compute): Likewise.
28098 (loop_inverted_post_order_compute): Likewise.
28099 * hsa-common.h: Likewise.
28100 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28101 * init-regs.c (initialize_uninitialized_regs): Likewise.
28102 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28103 (inline_small_functions): Likewise.
28104 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28105 * ira.c (combine_and_move_insns): Likewise.
28106 (build_insn_chain): Likewise.
28107 * loop-invariant.c (find_invariants): Likewise.
28108 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28109 * predict.c (tree_predict_by_opcode): Likewise.
28110 (predict_paths_leading_to): Likewise.
28111 (predict_paths_leading_to_edge): Likewise.
28112 (estimate_loops_at_level): Likewise.
28113 (estimate_loops): Likewise.
28114 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28115 (spread_components): Likewise.
28116 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28117 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28118 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28119 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28120 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28121 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28122 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28123 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28124 (mark_threaded_blocks): Likewise.
28125 (thread_through_all_blocks): Likewise.
28126 * tree-ssa.c (verify_ssa): Likewise.
28127 (execute_update_addresses_taken): Likewise.
28128 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28130 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28132 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28134 (post_order_compute): Likewise.
28135 (inverted_post_order_compute): Likewise.
28136 (pre_and_rev_post_order_compute_fn): Likewise.
28138 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28140 * genrecog.c (int_set::int_set): Explicitly construct our
28141 auto_vec base class.
28142 * vec.h (auto_vec::auto_vec): New constructor.
28144 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28146 * bitmap.h (class auto_bitmap): New constructor taking
28147 bitmap_obstack * argument.
28149 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28151 * bitmap.h (class auto_bitmap): Change type of m_bits to
28152 bitmap_head, and adjust ctor / dtor and member operators.
28154 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28156 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28157 when returned register mode doesn't match original mode.
28159 2017-05-12 Jeff Law <law@redhat.com>
28160 Jakub Jelinek <jakub@redhat.com>
28162 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28163 we look for cc setter after the compare-elim changes.
28164 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28165 within the vector to match what compare-elim now expects.
28166 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28167 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28169 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28170 after the compare-elim changes.
28171 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28172 the vector to match what compare-elim now expects.
28173 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28174 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28175 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28176 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28177 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28179 * config/visium/visium.c (single_set_and_flags): Fix where
28180 we look for cc setter after the compare-elim changes.
28181 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28182 with the vector to match what compare-elim now expects.
28183 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28184 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28185 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28186 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28187 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28188 (neg<mode>2_insn_set_overflow): Likewise.
28190 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28192 PR middle-end/79794
28193 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28194 maybe_expand_insn call, set ops[0].target. If still set after call,
28195 set alt_rtl. Add extra arg to recursive calls.
28196 (extract_bit_field): Add alt_rtl argument. Pass to
28198 * expmed.h (extract_bit_field): Fix prototype.
28199 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28200 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28201 to extract_bit_field_calls.
28202 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28203 Pass alt_rtl to extract_bit_field calls.
28204 * calls.c (store_unaligned_arguments_into_psuedos)
28205 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28206 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28208 * optabs.h (struct expand_operand): Add target bitfield.
28210 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28212 * compare-elim.c (try_eliminate_compare): Canonicalize
28213 operation with embedded compare to
28214 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28215 (set (reg) (operation)].
28217 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28219 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28222 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28223 cost of adding a carry flag for ADC instruction.
28224 [case MINUS]: Ignore the cost of subtracting a carry flag
28225 for SBB instruction.
28227 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28229 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28231 * config/rs6000/bmiintrin.h: New file.
28232 * config/rs6000/bmi2intrin.h: New file.
28233 * config/rs6000/x86intrin.h: New file.
28235 2017-05-12 Jeff Law <law@redhat.com>
28237 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28240 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28242 PR middle-end/80707
28243 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28245 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28247 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28248 early expansion of vector divide builtins.
28249 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28250 builtins identified as having unsigned arguments.
28252 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28254 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28255 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28256 expansion of vector logical operations (and, andc, or, xor,
28259 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28261 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28262 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28264 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28266 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28267 early GIMPLE expansion of vector multiplies.
28269 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28271 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28272 TARGET_HAVE_MOVT conditional.
28273 (movt splitter): Likewise.
28275 2017-05-12 Richard Biener <rguenther@suse.de>
28277 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28278 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28279 Fold all stmts not inplace.
28281 2017-05-12 Richard Biener <rguenther@suse.de>
28283 PR tree-optimization/80713
28284 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28285 inserted_exprs bit for not removed stmts.
28287 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28289 PR middle-end/69921
28290 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28291 parallelized" attribute for parallelized OpenACC kernels.
28292 * omp-offload.c (execute_oacc_device_lower): Use it.
28294 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28295 Set "oacc kernels" attribute.
28296 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28297 parameter. Adjust all users.
28298 (oacc_fn_attrib_kernels_p): Remove function.
28299 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28300 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28301 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28302 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28303 assert "oacc kernels" attribute is set.
28305 2017-05-11 Carl Love <cel@us.ibm.com>
28307 * config/rs6000/rs6000-c: Add support for built-in functions
28308 vector unsigned char vec_popcnt (vector signed char)
28309 vector unsigned char vec_popcnt (vector unsigned char)
28310 vector unsigned short vec_popcnt (vector signed short)
28311 vector unsigned short vec_popcnt (vector unsigned short)
28312 vector unsigned int vec_popcnt (vector signed int)
28313 vector unsigned int vec_popcnt (vector unsigned int)
28314 vector unsigned long long vec_popcnt (vector signed long long)
28315 vector unsigned long long vec_popcnt (vector unsigned long long)
28316 vector signed long long vec_slo (vector signed long long,
28317 vector signed char)
28318 vector signed long long vec_slo (vector signed long long,
28319 vector unsigned char)
28320 vector unsigned long long vec_slo (vector unsigned long long,
28321 vector signed char)
28322 vector unsigned long long vec_slo (vector unsigned long long,
28323 vector unsigned char)
28324 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28325 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28326 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28327 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28328 * doc/extend.texi: Update the built-in documentation file for the
28329 new built-in functions.
28331 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28333 * attribs.h (sorted_attr_string): Move machine independent
28334 functions for target clone support from the i386 port to common
28335 code. Rename ix86_function_versions to common_function_versions.
28336 Rename make_name to make_unique_name.
28337 (common_function_versions): Likewise.
28338 (make_unique_name): Likewise.
28339 (make_dispatcher_decl): Likewise.
28340 (is_function_default_version): Likewise.
28341 * attribs.c (attr_strcmp): Likewise.
28342 (sorted_attr_string): Likewise.
28343 (common_function_versions): Likewise.
28344 (make_unique_name): Likewise.
28345 (make_dispatcher_decl): Likewise.
28346 (is_function_default_version): Likewise.
28347 * config/i386/i386.c (attr_strcmp): Likewise.
28348 (sorted_attr_string): Likewise.
28349 (ix86_function_versions): Likewise.
28350 (make_name): Likewise.
28351 (make_dispatcher_decl): Likewise.
28352 (is_function_default_version): Likewise.
28353 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28355 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28358 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28359 Account for direct move costs for vec_construct of integer
28362 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28365 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28366 (UNSPEC_STX_ATOMIC): Ditto.
28367 (loaddi_via_sse): New insn.
28368 (storedi_via_sse): Ditto.
28369 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28370 Update corresponding peephole2 patterns.
28371 (atomic_storedi_fpu): Ditto.
28373 2017-05-11 Julia Koval <julia.koval@intel.com>
28375 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28376 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28378 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28379 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28380 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28382 2017-05-11 Nathan Sidwell <nathan@acm.org>
28384 * graphite-poly.c: Include dumpfile.h.
28386 * dumpfle.h (dump_function): Declare here ...
28387 * tree-dump.h (dump_function): ... not here.
28388 * dumpfile.c: #include tree-cfg.h.
28389 (dump_function): Move here from ...
28390 * tree-dump.c (dump_function): ... here.
28391 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28392 * graphite-poly.c: Don't include tree-dump.h.
28393 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28394 * print-tree.c: Likewise.
28395 * stor-layout.c: Likewise.
28396 * tree-nested.c: Likewise.
28398 * dumpfile.c (dump_start): Use TDF_FLAGS.
28399 (dump_enable_all): Fix TDF_KIND check thinko.
28401 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28403 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28404 array entries to represent two legal parameterizations of the
28405 overloaded __builtin_cmpb function, as represented by the
28406 P6_OV_BUILTIN_CMPB constant.
28407 (altivec_resolve_overloaded_builtin): Add special case handling
28408 for the __builtin_cmpb function, as represented by the
28409 P6_OV_BUILTIN_CMPB constant.
28410 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28411 (BU_P6_64BIT_2): New macro.
28412 (BU_P6_OVERLOAD_2): New macro
28413 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28414 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28415 (CMPB): Add overload support to represent both 32-bit and 64-bit
28416 compare-bytes function.
28417 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28418 support for TARGET_CMPB.
28419 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28420 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28421 documentation of the __builtin_cmpb overloaded built-in function.
28423 2017-05-11 Richard Biener <rguenther@suse.de>
28425 PR tree-optimization/80705
28426 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28427 bases are not vectorizable.
28429 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28431 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28432 when counting register pressure.
28434 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28436 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28437 (iv_ca_more_deps): Renamed to ...
28438 (iv_ca_compare_deps): ... this.
28439 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28443 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28445 (determine_group_iv_costs): ... here.
28446 (find_inv_vars_cb): Record inv var if it's not recorded before.
28448 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28450 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28451 (get_shiftadd_cost): Ditto.
28453 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28455 * tree-ssa-address.c: Include header file.
28456 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28458 (add_to_parts): Refactor.
28459 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28460 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28463 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28465 PR tree-optimization/53090
28466 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28468 (extract_cond_operands): Detect condition with IV on both sides
28469 and return COMP_IV_EXPR_2.
28470 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28471 (rewrite_use_compare): Simplify by removing call to function
28472 extract_cond_operands.
28474 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28476 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28477 (extract_cond_operands): Detect condition comparing against non-
28478 invariant bound and return appropriate enum value.
28479 (find_interesting_uses_cond): Update use of extract_cond_operands.
28480 Handle its return value accordingly.
28481 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28483 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28485 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28486 nonlinear iv_use computation in loop invariant sensitive way.
28488 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28490 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28491 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28493 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28495 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28496 (dump_cand): Support iv_cand.inv_exprs.
28497 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28499 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28502 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28504 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28506 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28507 as local function. Include necessary header files.
28508 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28510 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28512 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28514 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28516 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28517 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28518 RSHIFT_EXPR and BIT_NOT_EXPR.
28520 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28522 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28523 (adjust_setup_cost): New parameter supporting round up adjustment.
28524 (struct address_cost_data): Delete.
28525 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28526 (split_address_cost, ptr_difference_cost): Delete.
28527 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28528 (struct ainc_cost_data): New struct.
28529 (get_address_cost_ainc): New function.
28530 (get_address_cost, get_computation_cost): Reimplement.
28531 (determine_group_iv_cost_address): Record inv_expr for all uses of
28533 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28534 (iv_ca_has_deps): Reimplemented to ...
28535 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28537 (iv_ca_extend): Call iv_ca_more_deps.
28539 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28541 * tree-ssa-address.c (struct mem_address): Move to header file.
28542 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28543 * tree-ssa-address.h (struct mem_address): Move from C file.
28544 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28546 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28548 * tree-affine.h (aff_combination_type): New interface.
28549 (aff_combination_zero_p): Remove static.
28550 (aff_combination_const_p): New interface.
28551 (aff_combination_singleton_var_p): New interfaces.
28553 2017-05-11 Richard Biener <rguenther@suse.de>
28555 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28556 Skip unreachable blocks and destinations.
28557 (eliminate): Move stmt removal and fixup ...
28558 (fini_eliminate): ... here. Skip inserted exprs.
28559 (pass_pre::execute): Move fini_pre after fini_eliminate.
28560 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28561 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28562 PRE to get rid of dead code that has invalid SSA form and
28563 split critical edges again.
28565 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28567 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28569 2017-05-11 Richard Biener <rguenther@suse.de>
28571 * passes.c (execute_function_todo): Verify loops if they are
28572 said to be up-to-date.
28573 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28574 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28576 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28579 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28580 handle calling assemble_external ourself.
28583 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28584 modes with zero size. Enhance comment.
28586 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28588 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28589 built-ins for vec_xl and vec_xst with short and char pointer
28592 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28594 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28595 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28596 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28597 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28598 (_mm_maskz_min_round_ss): New intrinsics.
28599 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28600 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28601 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28602 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28603 (__builtin_ia32_minss_mask_round): New builtins.
28604 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28605 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28606 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28608 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28609 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28611 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28614 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28616 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28617 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28618 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28619 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28620 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28621 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28622 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28623 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28624 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28625 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28626 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28627 (__builtin_ia32_mulss_mask_round): New builtins.
28628 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28629 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28630 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28632 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28633 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28635 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28638 2017-05-10 Julia Koval <julia.koval@intel.com>
28640 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28641 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28642 (_mm256_setr_m128i): New intrinsics.
28644 2017-05-10 Julia Koval <julia.koval@intel.com>
28646 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28647 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28648 (_mm_maskz_rcp14_ss): New intrinsics.
28649 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28650 (__builtin_ia32_rcp14ss_mask): New builtins.
28651 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28653 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28655 PR tree-optimization/51513
28656 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28657 (assert_unreachable_fallthru_edge_p): Use it.
28658 (group_case_labels_stmt): Likewise.
28659 * tree-cfg.h: Prototype it.
28660 * stmt.c: Include cfghooks.h and tree-cfg.h.
28661 (emit_case_dispatch_table) <gap_label>: New local variable.
28662 Use it to fill dispatch table gaps.
28663 Test for default_label before updating probabilities.
28664 (expand_case) <default_label>: Remove unneeded initialization.
28665 Test for unreachable default case statement and remove its edge.
28666 Set default_label accordingly.
28667 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28669 2017-05-10 Carl Love <cel@us.ibm.com>
28671 * config/rs6000/rs6000-c: Add support for built-in functions
28672 vector signed char vec_neg (vector signed char)
28673 vector signed short int vec_neg (vector short int)
28674 vector signed int vec_neg (vector signed int)
28675 vector signed long long vec_neg (vector signed long long)
28676 vector float vec_neg (vector float)
28677 vector double vec_neg (vector double)
28678 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28680 * config/rs6000/altivec.h: Add define for vec_neg
28681 * doc/extend.texi: Update the built-in documentation for the
28682 new built-in functions.
28684 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28686 PR tree-optimization/77644
28687 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28689 2017-05-10 Nathan Sidwell <nathan@acm.org>
28691 * dumpfile.h (TDI_lang_all): New.
28692 (TDF_KIND): New. Renumber others
28693 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28695 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28697 (get_dump_file_name): Adjust suffix generation.
28698 (dump_enable_all): Use TDF_KIND.
28699 * doc/invoke.texi (-fdump-lang-all): Document.
28701 * dumpfile.h: Tabify.
28703 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28706 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28707 Move member access before delete.
28709 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28711 * tree-inline.c (expand_call_inline): Split block at stmt
28714 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28717 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28718 are now unused after splitting mov{sf,sd}_hardfloat.
28719 (f32_lr2): Likewise.
28720 (f32_lm): Likewise.
28721 (f32_lm2): Likewise.
28722 (f32_li): Likewise.
28723 (f32_li2): Likewise.
28724 (f32_lv): Likewise.
28725 (f32_sr): Likewise.
28726 (f32_sr2): Likewise.
28727 (f32_sm): Likewise.
28728 (f32_sm2): Likewise.
28729 (f32_si): Likewise.
28730 (f32_si2): Likewise.
28731 (f32_sv): Likewise.
28732 (f32_dm): Likewise.
28733 (f32_vsx): Likewise.
28734 (f32_av): Likewise.
28735 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28736 For movsf, order stores so the VSX stores occur before the GPR
28737 store which encourages the register allocator to use a traditional
28738 FPR instead of a GPR. For movsd, order the stores so that the GPR
28739 store comes before the VSX stores to allow the power6 to work.
28740 This is due to the power6 not having a 32-bit integer store
28741 instruction from a FPR.
28742 (movsf_hardfloat): Likewise.
28743 (movsd_hardfloat): Likewise.
28745 2017-05-09 Martin Sebor <msebor@redhat.com>
28747 PR translation/80280
28748 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28751 PR translation/80280
28752 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28753 data member added in r247778.
28754 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28756 2017-05-09 Nathan Sidwell <nathan@acm.org>
28758 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28760 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28763 2017-05-09 Marek Polacek <polacek@redhat.com>
28765 * doc/invoke.texi: Fix typo.
28767 2017-05-09 Richard Biener <rguenther@suse.de>
28769 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28770 (vrp_val_is_min): Likewise.
28771 (set_value_range_to_value): Likewise.
28772 (set_value_range_to_nonnegative): Likewise.
28773 (gimple_assign_nonzero_p): Likewise.
28774 (gimple_stmt_nonzero_p): Likewise.
28775 (vrp_int_const_binop): Likewise. Remove unreachable case.
28776 (adjust_range_with_scev): Adjust comments.
28777 (compare_range_with_value): Likewise.
28778 (extract_range_from_phi_node): Likewise.
28779 (test_for_singularity): Likewise.
28781 2017-05-09 Richard Biener <rguenther@suse.de>
28783 * tree-vrp.c (get_single_symbol): Add assert that we don't
28784 get overflowed constants as invariant part.
28785 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28786 checks. Use wi::cmp instead of recursing for integer constants.
28787 (compare_values): Just ignore whether we assumed undefined
28788 overflow instead of failing the compare.
28789 (extract_range_for_var_from_comparison_expr): Add comment before the
28790 TREE_NO_WARNING sets.
28791 (test_for_singularity): Likewise.
28792 (extract_range_from_comparison): Do not disable optimization
28793 when we assumed undefined overflow.
28794 (extract_range_basic): Remove init of unused var.
28796 2017-05-09 Richard Biener <rguenther@suse.de>
28798 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28799 (extract_range_from_multiplicative_op_1): Adjust.
28800 (extract_range_from_binary_expr_1): Use int_const_binop.
28802 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28805 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28806 rs6000_store_data_bypass_p in seven define_bypass directives and
28807 in several comments.
28808 * config/rs6000/rs6000-protos.h: Add prototype for
28809 rs6000_store_data_bypass_p function.
28810 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28811 function implements slightly different (rs6000-specific) semantics
28812 than store_data_bypass_p, returning false rather than aborting
28813 with assertion error when arguments do not satisfy the
28814 requirements of store data bypass.
28815 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28816 rs6000_store_data_bypass_p.
28818 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28820 * config/xtensa/xtensa-protos.h
28821 (xtensa_initial_elimination_offset): New declaration.
28822 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28823 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28824 macro definition, add case for FRAME_POINTER_REGNUM when
28825 FRAME_GROWS_DOWNWARD.
28826 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28827 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28828 xtensa_initial_elimination_offset.
28830 2017-05-08 Nathan Sidwell <nathan@acm.org>
28832 * doc/invoke.texi: Alphabetize -fdump options.
28834 2017-05-08 Martin Sebor <msebor@redhat.com>
28836 PR translation/80280
28837 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28839 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28841 * target.def (compute_frame_layout): New optional target hook.
28842 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28843 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28844 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28846 * reload1.c (verify_initial_elim_offsets): Likewise.
28847 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28848 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28849 (arm_get_frame_offsets): Split up into this ...
28850 (arm_compute_frame_layout): ... and this function.
28852 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28854 * config/aarch64/constraints.md (Usa): New constraint.
28855 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28857 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28859 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28860 with_multilib_list after it has been checked.
28862 2017-05-08 Richard Biener <rguenther@suse.de>
28864 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28865 (bitmap_set_subtract_values): Likewise.
28867 2017-05-08 Richard Biener <rguenther@suse.de>
28869 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28870 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28872 (gimple_stmt_nonzero_warnv_p): Rename to ...
28873 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28875 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28876 (extract_range_basic): Adjust, do not disable propagation on
28877 strict overflow sensitive simplification.
28878 (vrp_visit_cond_stmt): Likewise.
28880 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28882 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28883 body size unconditionally.
28885 2017-05-07 Jeff Law <law@redhat.com>
28888 2017-05-06 Jeff Law <law@redhat.com>
28889 PR tree-optimization/78496
28890 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28893 PR tree-optimization/78496
28894 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28895 (simplify_stmt_using_ranges): Call it.
28896 (vrp_dom_walker::before_dom_children): Extract equivalences
28897 from an ASSERT_EXPR with an equality comparison against a
28900 2017-05-06 Jeff Law <law@redhat.com>
28902 PR tree-optimization/78496
28903 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28906 PR tree-optimization/78496
28907 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28908 (simplify_stmt_using_ranges): Call it.
28909 (vrp_dom_walker::before_dom_children): Extract equivalences
28910 from an ASSERT_EXPR with an equality comparison against a
28913 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28915 * lra-constraints.c (lra_copy_reg_equiv): New function.
28916 (split_reg): Use it to copy equivalence information from the
28917 original register to the spill register.
28919 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28921 PR rtl-optimization/75964
28922 * simplify-rtx.c (simplify_const_relational_operation): Remove
28923 invalid handling of comparisons of integer ABS.
28925 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28927 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28928 initialize to zero.
28929 (init_regs): Remove declaration.
28930 (function_arg_advance_32): Initialize error_p as boolean variable.
28932 2017-05-05 Nathan Sidwell <nathan@acm.org>
28934 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28935 lines. Use for (;;).
28937 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28939 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28940 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28941 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28942 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28943 VF=2 that require versioning.
28945 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28947 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28950 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28952 * diagnostic.h (diagnostic_override_option_index): Convert from
28953 macro to inline function.
28955 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28957 * diagnostic.c (last_module_changed_p): New function.
28958 (set_last_module): New function.
28959 (diagnostic_report_current_module): Convert macro usage to
28960 the above functions.
28961 * diagnostic.h (diagnostic_context::last_module): Strengthen
28962 from const line_map * to const line_map_ordinary *.
28963 (diagnostic_last_module_changed): Delete macro.
28964 (diagnostic_set_last_module): Delete macro.
28966 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28968 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28969 with diagnostic_report_diagnostic.
28970 (diagnostic_n_impl_richloc): Likewise.
28971 * diagnostic.h (report_diagnostic): Delete macro.
28972 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28973 with diagnostic_report_diagnostic.
28974 * substring-locations.c (format_warning_va): Likewise.
28976 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28978 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28979 save/restor of format_spec. Move option-printing code to...
28980 (print_option_information): ...this new function, and
28981 reimplement by simply printing to the pretty_printer,
28982 rather than appending to the format string.
28984 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28986 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28987 handling logic into...
28988 (update_effective_level_from_pragmas): ...this new function.
28990 2017-05-04 Andrew Waterman <andrew@sifive.com>
28992 * config/riscv/riscv.opt (mstrict-align): New option.
28993 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28994 (SLOW_UNALIGNED_ACCESS): Define.
28995 (riscv_slow_unaligned_access): Declare.
28996 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28998 (riscv_slow_unaligned_access): New variable.
28999 (rocket_tune_info): Set slow_unaligned_access to true.
29000 (optimize_size_tune_info): Set slow_unaligned_access to false.
29001 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29002 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29003 (riscv_option_override): Set riscv_slow_unaligned_access.
29004 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29006 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29008 * config/riscv/riscv.md: Unify indentation.
29010 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29015 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29017 (extendsi<mode>2): Add support for doing sign extension via
29018 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29019 don't have ISA 3.0 instructions.
29020 (extendsi<mode>2 splitter): Likewise.
29021 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29022 generate the normal insns since SImode can now go in vector
29023 registers. Disallow the special UNSPECs needed for previous
29024 machines to hide SImode being used. Add new insns
29025 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29026 (fix_trunc<mode>si2_stfiwx): Likewise.
29027 (fix_trunc<mode>si2_internal): Likewise.
29028 (fixuns_trunc<mode>si2): Likewise.
29029 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29030 (fctiw<u>z_<mode>_smallint): Likewise.
29031 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29032 of floating point to 32-bit integer from doing a direct move to
29033 the GPR registers to do a store.
29034 (fctiwz_<mode>): Break long line.
29036 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29038 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29039 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29040 (addr_list, addr_offset_valid_p): New.
29041 (split_address_groups): Check offset validity with above function.
29042 (gt-tree-ssa-loop-ivopts.h): Include header file.
29044 2017-05-05 Nathan Sidwell <nathan@acm.org>
29046 * config.gcc (arm*-*-*): Add missing 'fi'.
29048 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29050 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29051 included in -fopt-info does not matter.
29052 * doc/optinfo.texi (-fopt-info): Fix description of default
29053 behavour. Explicitly say order of options included in -fopt-info
29056 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29058 * config.gcc: Allow combinations of aprofile and rmprofile values for
29059 --with-multilib-list.
29060 * config/arm/t-multilib: New file.
29061 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29062 variables. Remove setting of ISA and floating-point ABI in
29063 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29064 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29065 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29066 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29068 * config/arm/t-rmprofile: Likewise except for the matches changes.
29069 * doc/install.texi (--with-multilib-list): Document the combination of
29070 aprofile and rmprofile values and warn about pitfalls in doing that.
29072 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29074 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29075 (movdi_aarch64): Likewise.
29077 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29079 PR tree-optimization/80632
29080 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29082 (build_arrays): Initialize it for virtual phis.
29083 (fix_phi_nodes): Use it for virtual phis.
29085 PR tree-optimization/80558
29086 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29087 [x, y] op z into [x op, y op z] for op & or | if conditions
29090 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29091 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29094 * config/arm/arm.md (use_literal_pool): Remove.
29095 (64-bit immediate split): No longer takes cost into consideration
29096 if arm_disable_literal_pool is enabled.
29097 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29098 used when arm_disable_literal_pool is enabled.
29099 (arm_max_const_double_inline_cost): Remove use of
29100 arm_disable_literal_pool.
29101 (push_minipool_fix): Add assert.
29102 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29103 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29104 (no_literal_pool_sf_immediate): New.
29106 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29108 PR tree-optimization/80613
29109 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29110 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29112 2017-05-05 Richard Biener <rguenther@suse.de>
29114 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29116 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29118 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29119 of this flag from insn conditions due to removal from r247495.
29121 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29123 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29125 (arm_early_store_addr_dep_ptr): Likewise.
29126 * config/arm/aarch-common-protos.h
29127 (arm_early_load_addr_dep_ptr): Add prototype.
29128 (arm_early_store_addr_dep_ptr): Likewise.
29129 * config/arm/cortex-a53.md: Add new bypasses.
29131 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29133 * tree.c (next_type_uid): Change type to unsigned.
29134 (type_hash_canon): Decrement back next_type_uid if
29135 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29136 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29139 2017-05-04 Martin Sebor <msebor@redhat.com>
29141 * builtins.c: Fix a trivial typo in a comment.
29143 PR middle-end/79234
29144 * builtins.c (check_sizes): Adjust to handle reading past the end.
29145 Avoid printing excessive upper bound of ranges. Use %E to print
29146 tree nodes instead of converting them to %wu.
29147 (expand_builtin_memchr): New function.
29148 (compute_dest_size): Rename...
29149 (compute_objsize): ...to this.
29150 (expand_builtin_memcpy): Adjust.
29151 (expand_builtin_mempcpy): Adjust.
29152 (expand_builtin_strcat): Adjust.
29153 (expand_builtin_strcpy): Adjust.
29154 (check_strncat_sizes): Adjust.
29155 (expand_builtin_strncat): Adjust.
29156 (expand_builtin_strncpy): Adjust and simplify.
29157 (expand_builtin_memset): Adjust.
29158 (expand_builtin_bzero): Adjust.
29159 (expand_builtin_memcmp): Adjust.
29160 (expand_builtin): Handle memcmp.
29161 (maybe_emit_chk_warning): Check strncat just once.
29163 2017-05-04 Martin Sebor <msebor@redhat.com>
29165 PR preprocessor/79214
29166 PR middle-end/79222
29167 PR middle-end/79223
29168 * builtins.c (check_sizes): Add inlining context and issue
29169 warnings even when -Wno-system-headers is set.
29170 (check_strncat_sizes): Same.
29171 (expand_builtin_strncat): Same.
29172 (expand_builtin_memmove): New function.
29173 (expand_builtin_stpncpy): Same.
29174 (expand_builtin): Handle memmove and stpncpy.
29176 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29178 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29179 which is not used any more.
29181 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29183 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29185 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29187 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29188 (cortexa53_tunings): Likewise.
29189 (cortexa57_tunings): Likewise.
29190 (cortexa72_tunings): Likewise.
29191 (cortexa73_tunings): Likewise.
29193 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29195 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29196 Set loop alignment to 8.
29198 2017-05-04 Martin Sebor <msebor@redhat.com>
29200 PR translation/80280
29201 * builtins.c (expand_builtin_object_size): Add missing quoting to
29202 %D and like directives.
29203 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29204 (hsa_type_for_tree_type): Same.
29205 (verify_function_arguments): Same.
29206 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29207 * varasm.c (get_section): Same.
29210 2017-05-04 Martin Sebor <msebor@redhat.com>
29212 PR translation/80280
29213 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29215 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29217 * config/aarch64/aarch64.c (generic_addrcost_table):
29218 Change HI/TI mode setting.
29220 2017-05-04 Martin Jambor <mjambor@suse.cz>
29222 PR tree-optimization/80622
29223 * tree-sra.c (comes_initialized_p): New function.
29224 (build_accesses_from_assign): Only set write lazily when
29225 comes_initialized_p is false.
29226 (analyze_access_subtree): Use comes_initialized_p.
29227 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29228 instead of testing for PARM_DECL.
29230 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29232 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29233 constraint on operand 0 to allow more general addressing modes.
29234 Adjust output template.
29235 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29237 * config/aarch64/aarch64-protos.h
29238 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29239 * config/aarch64/constraints.md (Dp): New address constraint.
29240 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29243 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29245 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29246 update use of estimate_ipcp_clone_size_and_time.
29247 (estimate_local_effects): Update use of
29248 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29249 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29250 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29251 Return nonspecialized time.
29253 2017-05-04 Richard Biener <rguenther@suse.de>
29255 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29256 for the last VUSE which def dominates the PHI. Directly call
29258 (get_continuation_for_phi_1): Remove.
29260 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29262 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29263 to explain the use of truncating division. Cap the number of
29264 iterations to the maximum given by nb_iterations_upper_bound,
29267 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29269 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29270 * configure: Regenerate.
29271 * config.in: Regenerate.
29272 * config/i386/driver-mingw32.c: new file.
29273 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29274 * config.host: Link driver-mingw32.o on MinGW host.
29275 * doc/install.texi: Document new --enable-mingw-wildcard configure
29278 2017-05-04 Marek Polacek <polacek@redhat.com>
29280 PR tree-optimization/80612
29281 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29283 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29284 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29286 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29287 (movt splitter): Likewise.
29288 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29289 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29290 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29291 block for Thumb-1 with MOVT.
29292 (thumb2_legitimate_address_p): Move code block ...
29293 (can_avoid_literal_pool_for_label_p): ... into this new function.
29294 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29296 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29297 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29298 "M-profile targets with the MOVT instruction".
29300 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29302 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29303 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29304 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29306 2017-05-04 Martin Liska <mliska@suse.cz>
29308 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29309 variable cond_code.
29311 2017-05-04 Richard Biener <rguenther@suse.de>
29313 * tree.c (array_at_struct_end_p): Handle arrays at struct
29314 end with flexarrays more conservatively. Refactor and treat
29315 arrays of arrays or aggregates more strict. Fix
29316 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29317 * tree.c (array_at_struct_end_p): Adjust prototype.
29318 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29319 * gimple-fold.c (get_range_strlen): Likewise.
29320 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29322 2017-05-04 Richard Biener <rguenther@suse.de>
29324 PR tree-optimization/31130
29325 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29327 (supports_overflow_infinity): Likewise.
29328 (is_negative_overflow_infinity): Likewise.
29329 (is_positive_overflow_infinity): Likewise.
29330 (is_overflow_infinity): Likewise.
29331 (stmt_overflow_infinity): Likewise.
29332 (overflow_infinity_range_p): Likewise.
29333 (usable_range_p): Remove as always returning true.
29334 (make_overflow_infinity): Remove.
29335 (negative_overflow_infinity): Likewise.
29336 (positive_overflow_infinity): Likewise.
29337 (avoid_overflow_infinity): Likewise.
29338 (set_value_range): Adjust accordingly.
29339 (set_value_range_to_nonnegative): Likewise, remove now unused
29340 overflow_infinity arg.
29341 (vrp_operand_equal_p): Adjust.
29342 (update_value_range): Likewise.
29343 (range_int_cst_singleton_p): Likewise.
29344 (operand_less_p): Likewise.
29345 (compare_values_warnv): Likewise.
29346 (extract_range_for_var_from_comparison_expr): Likewise.
29347 (vrp_int_const_binop): Likewise.
29348 (zero_nonzero_bits_from_vr): Likewise.
29349 (extract_range_from_multiplicative_op_1): Likewise.
29350 (extract_range_from_binary_expr_1): Likewise.
29351 (extract_range_from_unary_expr): Likewise.
29352 (extract_range_from_comparison): Likewise.
29353 (extract_range_basic): Likewise.
29354 (adjust_range_with_scev): Likewise.
29355 (compare_ranges): Likewise.
29356 (compare_range_with_value): Likewise.
29357 (dump_value_range): Likewise.
29358 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29360 (simplify_cond_using_ranges): Adjust.
29362 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29364 * brig-builtins.def: Added a builtin for class_f64.
29365 * builtin-types.def: Added a builtin type needed by class_f64.
29367 2017-05-03 Jason Merrill <jason@redhat.com>
29369 * timevar.def: Add TV_CONSTEXPR.
29371 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29373 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29375 2017-05-03 Martin Jambor <mjambor@suse.cz>
29377 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29378 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29379 * ipa-cp.c (ipcp_propagate_stage): Do not call
29380 ipa_update_after_lto_read.
29381 * ipa-inline.c (ipa_inline): Likewise.
29383 2017-05-03 Martin Jambor <mjambor@suse.cz>
29385 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29386 tag. Added a default constructor and a destructor.
29387 (ipa_edge_args_sum_t): New class;
29388 (ipa_edge_args_sum): Declare.
29389 (ipa_edge_args_vector): Remove declaration.
29390 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29391 (ipa_free_edge_args_substructures): Remove declaration.
29392 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29393 (ipa_edge_args_info_available_for_edge_p): Likewise.
29394 * ipa-prop.c (ipa_edge_args_vector): Removed.
29395 (edge_removal_hook_holder): Likewise.
29396 (edge_duplication_hook_holder): Likewise.
29397 (ipa_edge_args_sum): New variable.
29398 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29399 ipa_edge_args_vector.
29400 (ipa_free_edge_args_substructures): Likewise.
29401 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29402 ipa_edge_args_vector.
29403 (ipa_edge_removal_hook): Turned into method
29404 ipa_edge_args_sum_t::remove.
29405 (ipa_edge_duplication_hook): Turned into method
29406 ipa_edge_args_sum_t::duplicate.
29407 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29408 registering edge hooks.
29409 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29410 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29411 ipa_edge_args_sum instead of ipa_edge_args_vector.
29412 * ipa-profile.c (ipa_profile): Likewise.
29414 2017-05-03 Martin Jambor <mjambor@suse.cz>
29416 * symbol-summary.h (function_summary): New method exists.
29417 (function_summary::symtab_removal): Deallocate through release.
29418 (call_summary): New class.
29419 (gt_ggc_mx): New overload.
29420 (gt_pch_nx): Likewise.
29421 (gt_pch_nx): Likewise.
29423 2017-05-03 Jeff Law <law@redhat.com>
29425 PR tree-optimization/78496
29426 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29427 from simplify_cond_using_ranges. Split off code to walk
29428 backwards through casts into ...
29429 (simplify_cond_using_ranges_2): New function.
29430 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29431 (execute_vrp): After identifying jump threads, call
29432 simplify_cond_using_ranges_2.
29434 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29437 * ipa-inline.h (inline_summary): Add ctor.
29438 (create_ggc): Do not use ggc_cleared_alloc.
29440 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29441 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29443 * gcc.c (handle_braces): Support escaping in switch matching
29445 * doc/invoke.texi (Spec Files): Document it.
29446 Remove superfluous @code markup in items.
29448 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29450 * diagnostic-show-locus.c (struct column_range): New struct.
29451 (get_affected_columns): New function.
29452 (get_printed_columns): New function.
29453 (struct correction): New struct.
29454 (correction::ensure_capacity): New function.
29455 (correction::ensure_terminated): New function.
29456 (struct line_corrections): New struct.
29457 (line_corrections::~line_corrections): New dtor.
29458 (line_corrections::add_hint): New function.
29459 (layout::print_trailing_fixits): Reimplement in terms of the new
29461 (selftest::test_overlapped_fixit_printing): New function.
29462 (selftest::diagnostic_show_locus_c_tests): Call it.
29464 2017-05-03 Nathan Sidwell <nathan@acm.org>
29466 Canonicalize canonical type hashing
29467 * tree.h (type_hash_canon_hash): Declare.
29468 * tree.c (type_hash_list, attribute_hash_list): Move into
29469 type_hash_canon_hash.
29470 (build_type_attribute_qual_variant): Break out hash code calc into
29471 type_hash_canon_hash.
29472 (type_hash_canon_hash): New. Generic type hash computation.
29473 (build_range_type_1, build_array_type_1, build_function_type,
29474 build_method_type_directly, build_offset_type, build_complex_type,
29475 make_vector_type): Call it.
29477 2017-05-03 Richard Biener <rguenther@suse.de>
29479 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29480 When all DRs have unknown misaligned do not always peel
29481 when there is a store but apply the same costing model as if
29482 there were only loads.
29484 2017-05-03 Richard Biener <rguenther@suse.de>
29487 PR tree-optimization/80492
29488 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29489 compare_base_decls returning dont-know properly.
29491 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29493 * config/arm/iterators.md (CCSI): New mode iterator.
29494 (arch): New mode attribute.
29495 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29496 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29497 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29498 code iterator for success result mode.
29499 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29500 the corresponding new insn generators.
29502 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29505 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29506 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29508 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29510 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29511 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29512 (DDR_A): Wrap DDR argument in brackets.
29513 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29514 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29515 (DDR_REVERSED_P): Likewise.
29517 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29519 PR tree-optimization/79472
29520 * tree-switch-conversion.c (struct switch_conv_info): Add
29521 contiguous_range and default_case_nonstandard fields.
29522 (collect_switch_conv_info): Compute contiguous_range and
29523 default_case_nonstandard fields, don't clear final_bb if
29524 contiguous_range and only the default case doesn't have the required
29526 (check_all_empty_except_final): Set default_case_nonstandard instead
29527 of failing if contiguous_range and the default case doesn't have empty
29529 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29530 and only the default case doesn't have the required constants. Skip
29532 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29533 if default_case_nonstandard.
29534 (build_constructors): Build constant 1 just once. Assert that default
29535 values aren't inserted in between cases if contiguous_range. Skip
29537 (build_arrays): Skip virtual phis.
29538 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29539 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29540 Handle virtual phis.
29541 (gen_inbound_check): Handle default_case_nonstandard case.
29542 (process_switch): Adjust check_final_bb caller. Call
29543 gather_default_values with the first non-default case instead of
29544 default case if default_case_nonstandard.
29546 2017-05-02 Nathan Sidwell <nathan@acm.org>
29548 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29549 check. Fix formatting.
29551 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29553 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29554 errors when comparing specialized and unspecialized times.
29556 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29558 * diagnostic-show-locus.c
29559 (layout::should_print_annotation_line_p): Make private.
29560 (layout::print_annotation_line): Make private.
29561 (layout::annotation_line_showed_range_p): Make private.
29562 (layout::show_ruler): Make private.
29563 (layout::print_source_line): Make private. Pass in line and
29564 line_width, rather than calling location_get_source_line. Drop
29566 (layout::print_leading_fixits): New method.
29567 (layout::print_any_fixits): Rename to...
29568 (layout::print_trailing_fixits): ...this, and make private.
29569 Don't print newline fixits.
29570 (diagnostic_show_locus): Move logic for printing one row into...
29571 (layout::print_line): ...this new function. Move the
29572 location_get_source_line call and error-handling from
29573 print_source_line to here. Call print_leading_fixits, and rename
29574 print_any_fixits to print_trailing_fixits.
29575 (selftest::test_fixit_insert_containing_newline): Update now that
29576 newlines are partially supported.
29577 (selftest::test_fixit_insert_containing_newline_2): New test.
29578 (selftest::test_fixit_replace_containing_newline): Update comments.
29579 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29580 * edit-context.c (class added_line): New class.
29581 (class edited_line): Describe newline handling in comment.
29582 (edited_line::actually_edited_p): New method.
29583 (edited_line::print_content): Delete redundant decl.
29584 (edited_line::m_predecessors): New field.
29585 (edited_file::print_content): Call edited_line::print_content.
29586 (edited_file::print_diff): Update to support newlines.
29587 (edited_file::print_diff_hunk): Likewise.
29588 (edited_file::print_run_of_changed_lines): New function.
29589 (edited_file::print_diff_line): Convert to...
29590 (print_diff_line): ...this.
29591 (edited_file::get_effective_line_count): New function.
29592 (edited_line::edited_line): Initialize new field m_predecessors.
29593 (edited_line::~edited_line): Clean up m_predecessors.
29594 (edited_line::apply_fixit): Handle newlines.
29595 (edited_line::get_effective_line_count): New function.
29596 (edited_line::print_content): New function.
29597 (edited_line::print_diff_lines): New function.
29598 (selftest::test_applying_fixits_insert_containing_newline): New
29600 (selftest::test_applying_fixits_replace_containing_newline): New
29602 (selftest::insert_line): New function.
29603 (selftest::test_applying_fixits_multiple_lines): Add example of
29605 (selftest::edit_context_c_tests): Call the new tests.
29607 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29609 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29610 parameter cand. Update dump information.
29611 (get_computation_cost): Update uses.
29613 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29615 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29616 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29617 (get_computation_at, rewrite_use_address): Update use of
29618 get_computation_aff.
29620 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29622 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29623 (get_computation): Delete.
29624 (get_computation_cost): Implement like get_computation_cost_at.
29625 Use get_computation_at.
29626 (get_computation_cost_at): Delete.
29627 (rewrite_use_nonlinear_expr): Use get_computation_at.
29628 (rewrite_use_compare, remove_unused_ivs): Ditto.
29630 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29632 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29634 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29636 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29637 (ivopts_global_cost_for_size): Rename parameter and update uses.
29638 (iv_ca_recount_cost): Update uses.
29639 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29640 candidates seperately in n_invs and n_cands.
29641 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29643 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29645 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29646 (find_inv_vars_cb): New.
29647 (find_depends): Renamed to ...
29648 (find_inv_vars): ... this.
29649 (add_candidate_1, force_var_cost): Call find_inv_vars.
29650 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29652 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29654 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29655 inv_vars. Add inv_exprs.
29656 (struct iv_cand): Rename depends_on to inv_vars.
29657 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29658 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29659 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29660 (dump_cand): Dump inv_vars.
29661 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29662 (record_invariant, find_depends, add_candidate_1): Ditto.
29663 (set_group_iv_cost, force_var_cost): Ditto.
29664 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29665 (get_computation_cost_at, get_computation_cost): Ditto.
29666 (determine_group_iv_cost_generic): Ditto.
29667 (determine_group_iv_cost_address): Ditto.
29668 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29669 (determine_group_iv_costs): Ditto.
29670 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29671 (iv_ca_set_remove_invariants): Renamed to ...
29672 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29673 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29674 (iv_ca_set_add_invariants): Renamed to ...
29675 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29676 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29677 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29678 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29679 (create_new_ivs): Remove useless dump.
29681 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29683 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29685 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29686 (iv_ca_set_no_cp, create_new_iv): Ditto.
29688 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29690 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29692 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29694 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29695 function tree_check2.
29697 2017-05-02 Martin Liska <mliska@suse.cz>
29699 * doc/gcov.texi: Add missing preposition.
29700 * gcov.c (function_info::function_info): Properly fill up
29701 all member variables.
29703 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29705 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29707 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29709 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29711 2017-05-02 Martin Liska <mliska@suse.cz>
29714 * lto-streamer-in.c (lto_read_tree_1): Remove
29715 LTO_STREAMER_DEBUG.
29716 * lto-streamer.c (struct tree_hash_entry): Likewise.
29717 (struct tree_entry_hasher): Likewise.
29718 (tree_entry_hasher::hash): Likewise.
29719 (tree_entry_hasher::equal): Likewise.
29720 (lto_streamer_init): Likewise.
29721 (lto_orig_address_map): Likewise.
29722 (lto_orig_address_get): Likewise.
29723 (lto_orig_address_remove): Likewise.
29724 * lto-streamer.h: Likewise.
29725 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29726 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29728 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29730 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29731 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29732 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29733 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29734 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29735 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29736 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29737 (mm_maskz_sub_ss): New intrinsics.
29738 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29739 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29740 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29741 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29742 (__builtin_ia32_subss_mask_round): New builtins.
29743 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29744 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29745 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29747 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29748 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29750 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29753 2017-05-02 Martin Jambor <mjambor@suse.cz>
29755 PR tree-optimization/78687
29756 * tree-sra.c (access): New field parent.
29757 (process_subtree_disqualification): New function.
29758 (disqualify_candidate): Call it.
29759 (build_accesses_from_assign): Reset write flag if creating an
29761 (build_access_subtree): Fill in parent field and also prpagate
29762 down grp_write flag.
29763 (create_artificial_child_access): New parameter set_grp_write, set
29764 grp_write to its value.
29765 (propagate_subaccesses_across_link): Also propagate grp_write flag
29767 (propagate_all_subaccesses): Push the closest parent back to work
29768 queue if add_access_to_work_queue returned true.
29770 2017-05-02 Richard Biener <rguenther@suse.de>
29772 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29773 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29774 -fstrict-overflow documentation.
29775 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29776 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29777 flag_strict_overflow.
29778 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29779 * lto-opts.c (lto_write_options): Do not stream it.
29780 * lto-wrapper.c (merge_and_complain): Do not handle it.
29781 * opts.c (default_options_table): Do not set -fstrict-overflow.
29782 (finish_options): Likewise do not clear it when sanitizing.
29783 * simplify-rtx.c (simplify_const_relational_operation): Do not
29784 test flag_strict_overflow.
29786 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29788 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29789 using enabled attribute.
29790 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29791 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29792 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29793 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29794 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29795 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29796 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29797 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29798 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29799 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29801 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29803 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29805 2017-05-02 Richard Biener <rguenther@suse.de>
29807 PR tree-optimization/80591
29809 2017-04-10 Richard Biener <rguenther@suse.de>
29811 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29814 2017-05-02 Richard Biener <rguenther@suse.de>
29816 PR tree-optimization/80549
29817 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29818 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29819 headers if they do not have a preheader.
29821 2017-05-02 Martin Liska <mliska@suse.cz>
29824 * common.opt: Fix typo.
29825 * doc/invoke.texi: Likewise.
29827 2017-05-01 Jan Beulich <jbeulich@suse.com>
29829 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29830 swapping, add (x,x,m,x,n) alternative.
29832 2017-05-01 Nathan Sidwell <nathan@acm.org>
29834 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29835 unnecessary unadjusted_alignment check.
29837 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29840 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29842 * gimplify.c (gimplify_cilk_detach): New function.
29843 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29844 * tree-core.h: Document EXPR_CILK_SPAWN.
29845 * tree.h (EXPR_CILK_SPAWN): Define.
29847 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29849 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29850 to use new fixit_hint representation, using the "replace" logic.
29851 (get_line_span_for_fixit_hint): Likewise.
29852 (layout::print_any_fixits): Likewise.
29853 (selftest::test_one_liner_many_fixits): Rename to...
29854 (selftest::test_one_liner_many_fixits_1): ...this, and update
29855 comment and expected output to reflect that the multiple fix-it
29856 hints are now consolidated into one insertion.
29857 (selftest::test_one_liner_many_fixits_2): New test.
29858 (selftest::test_diagnostic_show_locus_one_liner): Update for
29860 (selftest::test_fixit_consolidation): Update for fix-it API
29862 * diagnostic.c (print_parseable_fixits): Likewise.
29863 * edit-context.c (edited_line::m_line_events): Convert from
29864 auto_vec <line_event *> to auto_vec <line_event>.
29865 (class line_event): Convert from abstract base class to a concrete
29866 class, taking over the role of replace_event.
29867 (class insert_event): Delete.
29868 (class replace_event): Rename to class line_event. Convert to
29870 (edit_context::add_fixits): Reimplement.
29871 (edit_context::apply_insert): Delete.
29872 (edit_context::apply_replace): Rename to...
29873 (edit_context::apply_fixit): ...this. Convert to half-open range.
29874 (edited_file::apply_insert): Delete.
29875 (edited_file::apply_replace): Rename to...
29876 (edited_file::apply_fixit): ...this.
29877 (edited_line::~edited_line): Drop deletion of events.
29878 (edited_line::apply_insert): Delete.
29879 (edited_line::apply_replace): Rename to...
29880 (edited_line::apply_fixit): ...this. Convert to half-open range.
29881 Update for change to type of m_line_events.
29882 * edit-context.h (edit_context::apply_insert): Delete.
29883 (edit_context::apply_replace): Rename to...
29884 (edit_context::apply_fixit): ...this.
29886 2017-05-01 Martin Sebor <msebor@redhat.com>
29888 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29891 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29894 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29895 __get_cpuid_max returns 0.
29896 (__get_cpuid_count): Ditto.
29898 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29900 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29901 replacement expression is another instance of one of its arguments.
29903 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29906 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29907 check for stack push/pop autoinc.
29908 * config/i386/i386.c (ix86_agi_dependent): Return false
29909 if the only reason why modified_in_p returned true is that
29910 addr is SP based and set_insn is a push or pop.
29912 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29914 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29917 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29920 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29921 (account_size_time): Use two predicates - exec_pred and
29923 (evaluate_conditions_for_known_args): Compute both clause and
29925 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29926 (inline_summary_t::duplicate): Update.
29927 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29929 (compute_inline_parameters): Likewise.
29930 (estimate_edge_size_and_time): Update caluclation of time.
29931 (estimate_node_size_and_time): Compute both time and nonspecialized
29933 (estimate_ipcp_clone_size_and_time): Update.
29934 (inline_merge_summary): Update.
29935 (do_estimate_edge_time): Update.
29936 (do_estimate_edge_size): Update.
29937 (do_estimate_edge_hints): Update.
29938 (inline_read_section, inline_write_summary): Stream both new predicates.
29939 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29941 (compute_inlined_call_time): Cleanup.
29942 (big_speedup_p): Update.
29943 (edge_badness): Update.
29944 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29945 (size_time_entry): Replace predicate by exec_predicate and
29946 nonconst_predicate.
29947 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29948 (estimate_edge_time): Return also nonspec_time.
29949 (reset_edge_growth_cache): Update.
29951 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29953 PR rtl-optimization/80491
29954 * ifcvt.c (noce_process_if_block): When looking for x setter
29955 with missing else_bb, don't check only the insn right before
29956 cond_earliest, but look for the last insn that x is modified in
29957 within the same bb.
29959 PR rtl-optimization/80491
29960 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29962 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29964 PR tree-optimization/80487
29965 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29967 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29969 PR tree-optimization/79697
29970 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29971 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29972 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29974 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29975 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29977 2017-04-28 Martin Sebor <msebor@redhat.com>
29979 PR tree-optimization/80523
29980 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29981 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29983 (maybe_warn, format_directive, parse_directive): Use new functions.
29984 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29986 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29988 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29990 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29992 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29993 target_header_dir): Set correctly.
29994 * configure: Regenerated.
29995 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29996 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29997 instead of SYSTEM_HEADER_DIR.
29999 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30001 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30002 (estimate_local_effects): Likewise.
30003 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30004 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30005 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30006 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30007 do_estimate_edge_time, estimate_edge_time): Likewise.
30008 * ipa-inline-analysis.c (estimate_node_size_and_time,
30009 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30010 (estimate_time_after_inlining): Remove.
30012 2017-04-28 Martin Liska <mliska@suse.cz>
30014 * doc/gcov.texi: Enhance documentation of gcov.
30016 2017-04-28 Martin Liska <mliska@suse.cz>
30018 * doc/gcov.texi: Sort options in alphabetic order.
30019 * doc/gcov-dump.texi: Likewise.
30020 * doc/gcov-tool.texi: Likewise.
30021 * gcov.c (print_usage): Likewise.
30022 * gcov-dump.c (print_usage): Likewise.
30023 * gcov-tool.c (print_merge_usage_message): Likewise.
30024 (print_rewrite_usage_message): Likewise.
30025 (print_overlap_usage_message): Likewise.
30027 2017-04-28 Martin Liska <mliska@suse.cz>
30029 PR gcov-profile/53915
30030 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30032 2017-04-28 Martin Liska <mliska@suse.cz>
30034 PR gcov-profile/79891
30035 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30036 is marked by compiler as living on a line.
30037 (get_cycles_count): Remove usage of the union.
30038 (output_intermediate_file): Likewise.
30039 (find_source): Fix GNU coding style.
30040 (accumulate_line_counts): Remove old non-all block mode.
30041 (output_lines): Remove usage of the union.
30042 * profile.c (output_location): Include all BBs, even if
30043 belonging to a same line (and file) as a previous BB.
30045 2017-04-28 Martin Liska <mliska@suse.cz>
30047 * gcov.c (process_args): Handle new argument 'w'.
30048 (read_graph_file): Assign ID to BBs.
30049 (output_branch_count): Display BB # if verbose flag is set.
30050 (output_lines): Likewise for arcs.
30051 (print_usage): Add '--verbose' option help.
30052 * doc/gcov.texi: Document --verbose (-w) option.
30054 2017-04-28 Martin Liska <mliska@suse.cz>
30056 * gcov.c (struct block_location_info): New struct.
30057 (process_file): Fill up the new structure.
30058 (read_graph_file): Replace usage of encoding by the newly added
30060 (add_line_counts): Likewise.
30061 (accumulate_line_counts): Remove usage of the union.
30062 (function_info::function_info): New function.
30063 (function_info::~function_info): Likewise.
30064 (process_file): Call delete instead of release_function.
30065 (release_function): Release the function.
30066 (release_structures): Call delete instead of release_function.
30067 (solve_flow_graph): Replace usage of num_blocks.
30068 (find_exception_blocks): Likewise.
30069 (output_lines): Fix GNU coding style.
30071 2017-04-28 Martin Liska <mliska@suse.cz>
30074 * coverage.c (coverage_remove_note_file): New function.
30075 * coverage.h: Declare the function.
30076 * toplev.c (finalize): Clean if an error has been seen.
30078 2017-04-28 Martin Liska <mliska@suse.cz>
30080 PR gcov-profile/80031
30081 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30082 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30083 * gcov.c (read_graph_file): Read just number of blocks.
30084 * profile.c (branch_prob): Do not stream 0 flags per a basic
30087 2017-04-28 Martin Liska <mliska@suse.cz>
30089 * gcov-dump.c (tag_*): Add new argument to declarations.
30090 (dump_gcov_file): Likewise.
30091 (tag_blocks): Add and use new argument depth.
30092 (tag_arcs): Likewise.
30093 (tag_lines): Likewise.
30094 (tag_counters): Likewise.
30095 (tag_summary): Likewise.
30096 (dump_working_sets): Use depth to do a proper indentation.
30098 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30101 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30102 * symtab.c (symtab_node::debug_symtab): Move definition here.
30104 2017-04-28 Richard Biener <rguenther@suse.de>
30106 * lto-streamer.h (LTO_major_version): Bump to 7.
30108 2017-04-28 Richard Biener <rguenther@suse.de>
30110 * tree-vrp.c (assert_info): New struct.
30111 (add_assert_info): New helper.
30112 (register_edge_assert_for_2): Refactor to add asserts to a vector
30114 (register_edge_assert_for_1): Likewise.
30115 (register_edge_assert_for): Likewise.
30116 (finish_register_edge_assert_for): New helper actually registering
30117 asserts where live on edge.
30118 (find_conditional_asserts): Adjust.
30119 (find_switch_asserts): Likewise.
30120 (evrp_dom_walker::try_find_new_range): Generalize.
30121 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30123 2017-04-27 Marek Polacek <polacek@redhat.com>
30126 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30127 arg10 and arg11 to itype.
30129 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30131 * doc/extend.texi (Object Size Checking): Improve grammar.
30133 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30136 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30137 that the logic for permitting reciprocal estimates matches that
30140 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30143 * tree.c (type_cache_hasher::equal): Only compare
30144 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30145 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30146 non-aggregate element types.
30147 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30148 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30150 2017-04-27 Richard Biener <rguenther@suse.de>
30152 PR middle-end/80533
30153 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30154 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30155 keeping a reference to a trailing array.
30157 2017-04-27 Richard Biener <rguenther@suse.de>
30159 PR middle-end/80539
30160 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30161 being in loop-closed SSA form conservatively.
30162 (chrec_fold_multiply_poly_poly): Likewise.
30164 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30166 PR middle-end/79665
30167 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30168 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30170 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30173 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30174 (aarch64_function_arg_alignment): Return unsigned int again, but still
30175 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30176 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30177 Don't emit -Wpsabi note.
30178 (aarch64_function_arg_boundary): Likewise.
30179 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30182 2017-04-26 Nathan Sidwell <nathan@acm.org>
30184 * tree.h (crc32_unsigned_n): Declare.
30185 (crc32_unsigned, crc32_unsigned): Make inline.
30186 * tree.c (crc32_unsigned_bits): Replace with ...
30187 (crc32_unsigned_n): ... this.
30188 (crc32_unsigned, crc32_byte): Remove.
30189 (crc32_string): Remove unnecessary braces.
30191 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30193 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30194 * ipa-inline-analysis.c (MAX_TIME): Remove.
30195 (account_size_time): Use sreal for time.
30196 (dump_inline_summary): Update.
30197 (estimate_function_body_sizes): Update.
30198 (estimate_edge_size_and_time): Update.
30199 (estimate_calls_size_and_time): Update.
30200 (estimate_node_size_and_time): Update.
30201 (inline_merge_summary): Update.
30202 (inline_update_overall_summary): Update.
30203 (estimate_time_after_inlining): Update.
30204 (inline_read_section): Update.
30205 (inline_write_summary): Update.
30206 * ipa-inline.c (compute_uninlined_call_time): Update.
30207 (compute_inlined_call_time): Update.
30208 (recursive_inlining): Update.
30209 (inline_small_functions): Update.
30210 (dump_overall_stats): Update.
30211 * ipa-inline.h: Include sreal.h.
30212 (size_time_entry): Turn time to sreal.
30213 (inline_summary): Turn self_time nad time to sreal.
30215 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30217 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30219 (sreal::stream_out, sreal::stream_in): New.
30220 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30222 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30224 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30227 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30230 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30232 (dimode_scalar_chain::compute_convert_gain): Ditto.
30233 (dimode_scalar_chain::make_vector_copies): Ditto.
30234 (dimode_scalar_chain::convert_reg): Ditto.
30235 (dimode_scalar_chain::convert_insn): Ditto.
30236 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30237 (VI248_AVX512BW_1): New mode iterator.
30238 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30239 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30242 2017-04-25 Martin Sebor <msebor@redhat.com>
30244 PR tree-optimization/80497
30245 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30246 constants are representable in HOST_WIDE_INT.
30247 (parse_directive): Ditto.
30249 2017-04-25 Martin Sebor <msebor@redhat.com>
30252 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30253 (new_zero_array): Adjust signature.
30254 (dom_info::dom_init): Used unsigned rather that size_t.
30255 (dom_info::dom_info): Same.
30257 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30258 Jakub Jelinek <jakub@redhat.com>
30261 * config/arm/arm.c: Include gimple.h.
30262 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30263 returns negative, increment ncrn only if it returned positive.
30264 (arm_needs_doubleword_align): Return int instead of bool,
30265 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30266 members, but if there is any such non-FIELD_DECL
30267 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30268 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30269 returns negative, increment nregs only if it returned positive.
30270 (arm_setup_incoming_varargs): Likewise.
30271 (arm_function_arg_boundary): Emit -Wpsabi note if
30272 arm_needs_doubleword_align returns negative, return
30273 DOUBLEWORD_ALIGNMENT only if it returned positive.
30275 2017-04-25 Marek Polacek <polacek@redhat.com>
30278 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30279 first argument to type.
30281 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30284 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30285 type checks to test for compatibility instead of equality.
30287 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30288 Jakub Jelinek <jakub@redhat.com>
30291 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30293 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30294 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30295 the alignment computation, but return their maximum in warn_alignment.
30296 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30297 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30299 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30300 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30303 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30305 * config/arc/simdext.md (dmpyh): Fix typo.
30307 2017-04-25 Richard Biener <rguenther@suse.de>
30309 PR tree-optimization/80492
30310 * alias.c (compare_base_decls): Handle registers with asm
30311 specification conservatively.
30312 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30313 compare_base_decls returning dont-know properly.
30315 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30317 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30318 (legitimate_offset_address_p): New function.
30319 (arc_legitimate_address_p): Use above function.
30321 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30323 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30325 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30327 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30328 ACCH registers whenever they are available.
30330 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30332 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30333 double regs fix when not used.
30335 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30337 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30339 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30340 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30342 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30344 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30345 when emitting short ADD instructions.
30347 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30349 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30351 (cmpsi_cc_c_insn): Likewise.
30352 (cbranchsi4_scratch): Compute proper instruction length using
30353 compact_hreg_operand.
30354 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30356 2017-04-25 Richard Biener <rguenther@suse.de>
30358 PR middle-end/80509
30359 * passes.c (pass_manager::pass_manager): Initialize
30360 m_name_to_pass_map.
30362 2017-04-25 Richard Biener <rguenther@suse.de>
30364 PR tree-optimization/79201
30365 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30367 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30370 * config/s390/vector.md: Split MEM->GPR vector moves for
30371 non-s_operand addresses.
30373 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30376 * config/s390/predicates.md (reload_const_wide_int_operand): New
30378 * config/s390/s390.md ("movti"): Remove d/P alternative.
30379 ("movti_bigconst"): New pattern definition.
30381 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30384 * s390-protos.h (s390_expand_cs_hqi): Removed.
30385 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30386 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30387 modes as well as CCZ1mode and CCZmode.
30388 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30389 signature of s390_emit_compare_and_swap.
30390 (s390_expand_cs_hqi): Likewise, make static.
30391 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30392 compare-and-swap, in some cases.
30393 (s390_expand_cs): Wrapper function.
30394 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30396 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30397 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30398 patterns for small and large integers. Forbid symref memory operands.
30399 Move expander to s390.c. Require cc register.
30400 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30401 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30402 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30403 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30404 symref memory operands. Remove CC mode and call s390_match_ccmode
30406 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30408 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30410 * config/s390/s390.md (define_peephole2): New peephole to help
30411 combining the load-and-test pattern with volatile memory.
30413 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30415 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30416 with CCZmode for TARGET_Z196.
30418 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30420 PR rtl-optimization/80501
30421 * combine.c (make_compound_operation_int): Set subreg_code to SET
30422 even for AND with mask of the sign bit of mode.
30424 PR rtl-optimization/80500
30425 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30426 sum's initial value.
30428 2017-04-25 Julian Brown <julian@codesourcery.com>
30429 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30431 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30433 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30435 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30437 2017-04-25 Julian Brown <julian@codesourcery.com>
30438 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30440 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30441 (thunderx2t99_sha): New Reservation.
30443 2017-04-25 Julian Brown <julian@codesourcery.com>
30444 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30446 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30447 type for 1-element load.
30449 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30451 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30453 2017-04-24 Martin Jambor <mjambor@suse.cz>
30455 PR tree-optimization/80293
30456 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30457 char arrays not totally scalarizable if it is false.
30458 (analyze_all_variable_accesses): Pass correct value in the new
30459 parameter. Add a statistics counter.
30461 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30463 PR middle-end/79931
30464 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30466 2017-04-24 Richard Biener <rguenther@suse.de>
30468 PR tree-optimization/80494
30469 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30470 out for complex types.
30472 2017-04-24 Richard Biener <rguenther@suse.de>
30474 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30475 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30476 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30477 (DFS): Adjust and never fail.
30478 (sccvn_dom_walker::fail): Remove.
30479 (sccvn_dom_walker::before_dom_children): Adjust.
30480 (run_scc_vn): Likewise and never fail.
30481 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30482 (pass_fre::execute): Likewise.
30484 2017-04-24 Richard Biener <rguenther@suse.de>
30486 PR tree-optimization/79725
30487 * tree-ssa-sink.c (statement_sink_location): Return whether
30488 failure reason was zero uses. Move that check later.
30489 (sink_code_in_bb): Deal with zero uses by removing the stmt
30492 2017-04-24 Richard Biener <rguenther@suse.de>
30495 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30496 pointer-based references.
30498 2017-04-24 Richard Biener <rguenther@suse.de>
30501 * pass_manager.h (pass_manager::operator new): Remove.
30502 (pass_manager::operator delete): Likewise.
30503 * passes.c (pass_manager::operator new): Remove.
30504 (pass_manager::operator delete): Likewise.
30505 (pass_manager::pass_manager): Zero individual pass members.
30507 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30510 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30511 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30512 Check "XEXP (src, 1)" operand here.
30513 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30514 Check "XEXP (src, 1)" operand here.
30515 (dimode_scalar_chain::make_vector_copies): Detect count register
30516 of a shift instruction. Zero extend count register from QImode
30517 to DImode to satisfy vector shift pattern count operand predicate.
30518 Substitute vector shift count operand with a DImode copy.
30519 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30522 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30524 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30525 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30526 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30527 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30528 (UNSPEC_NOREX_MEM): Remove definition.
30530 2017-04-21 Richard Biener <rguenther@suse.de>
30532 PR tree-optimization/79547
30533 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30534 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30535 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30536 without any constraints.
30538 2017-04-21 Richard Biener <rguenther@suse.de>
30540 PR tree-optimization/78847
30541 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30543 2017-04-21 Richard Biener <rguenther@suse.de>
30545 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30546 (build_distinct_type_copy): Likewise.
30547 (build_variant_type_copy): Likewise.
30548 * tree.c (build_qualified_type): Pass down mem-stat info.
30549 (build_distinct_type_copy): Likewise.
30550 (build_variant_type_copy): Likewise.
30552 2017-04-21 Richard Biener <rguenther@suse.de>
30554 PR tree-optimization/80237
30555 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30557 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30558 for a simplified result.
30560 2016-04-21 Richard Biener <rguenther@suse.de>
30562 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30563 sth as strict as a simple_iv but a chrec without symbols and an
30564 operand defined in the loop we are peeling (and not some subloop).
30565 (propagate_constants_for_unrolling): Propagate all constants.
30567 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30570 * config/i386/i386.c (print_reg): Remove assert for disalowed
30571 regno values, call output_operand_lossage instead.
30573 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30576 * config/i386/constraints.md (Yc): New register constraint.
30577 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30578 Use Yc constraint for alternative 2 of operand 0. Remove
30579 preferred_for_speed attribute.
30581 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30583 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30584 lastprivate clauses in SIMT case.
30586 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30588 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30590 2017-04-20 Richard Biener <rguenther@suse.de>
30592 PR tree-optimization/57796
30593 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30594 as N scalar stores.
30595 (vect_model_load_cost): Cost gathers as N scalar loads.
30597 2017-04-20 Richard Biener <rguenther@suse.de>
30599 * ggc-page.c (ggc_allocated_p): Rename to ...
30600 (safe_lookup_page_table_entry): ... this and return the lookup
30602 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30604 2017-04-20 Richard Biener <rguenther@suse.de>
30606 PR tree-optimization/80453
30607 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30608 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30609 from the conditions.
30610 (vn_phi_eq): Pass them down.
30611 (vn_phi_lookup): Record them.
30612 (vn_phi_insert): Likewise.
30614 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30616 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30617 uninitialized variable warning to avoid buffer overrun.
30619 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30622 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30623 is suppressed for '{ 0 }' in C.
30625 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30627 * BASE-VER: Set to 8.0.0.
30629 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30631 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30632 priority .init_array and .fini_array section with SECTION_NOTYPE
30635 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30637 PR middle-end/80423
30638 * tree.h (build_array_type): Add typeless_storage default argument.
30639 * tree.c (type_cache_hasher::equal): Also compare
30640 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30641 (build_array_type): Add typeless_storage argument, set
30642 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30644 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30645 (build_array_type): Likewise. Add typeless_storage argument.
30647 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30648 Jakub Jelinek <jakub@redhat.com>
30650 PR tree-optimization/80426
30651 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30652 operation on symbolic operands, also compute the overflow for the
30653 invariant part when the operation degenerates into a negation.
30655 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30658 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30659 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30662 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30664 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30667 * config/avr/avr.c (tree.h): Include it.
30668 (cgraph.h): Include it.
30669 (avr_encode_section_info): Don't warn for uninitialized progmem
30670 variable if it's just an alias.
30672 2017-04-19 Richard Biener <rguenther@suse.de>
30675 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30676 when needed by AutoPGO.
30678 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30681 * doc/lto.texi: Remove an extra 'that'.
30683 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30685 PR rtl-optimization/80429
30686 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30687 are only used in debug insns.
30689 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30690 Vladimir Makarov <vmakarov@redhat.com>
30692 * config/sparc/predicates.md (input_operand): Add comment. Return
30693 true for any memory operand when LRA is in progress.
30694 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30696 2017-04-18 Jeff Law <law@redhat.com>
30699 * mips.md ({return,simple_return}_internal): Do not overwrite
30702 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30704 PR tree-optimization/80443
30705 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30706 instead of adding 1, subtract -1 and similarly instead of subtracting
30709 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30711 PR rtl-optimization/80357
30712 * haifa-sched.c (tmp_bitmap): New variable.
30713 (model_recompute): Handle duplicate use records.
30714 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30715 (free_global_sched_pressure_data): Free it.
30717 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30720 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30721 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30722 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30723 instead of SYSTEM_HEADER_DIR.
30725 2017-04-18 Jeff Law <law@redhat.com>
30727 PR middle-end/80422
30728 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30729 predecessors after walking up the insn chain.
30731 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30734 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30735 sizetype type into debug info.
30737 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30740 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30741 unneeded test for TARGET_UPPER_REGS_SF.
30742 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30744 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30747 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30748 instead of gsi_after_labels.
30750 2017-04-18 Jeff Law <law@redhat.com>
30752 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30756 2017-04-13 Jeff Law <law@redhat.com>
30757 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30758 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30760 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30763 * config/avr/avr.c (intl.h): Include it.
30764 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30766 2017-04-18 Martin Liska <mliska@suse.cz>
30768 PR gcov-profile/78783
30769 * gcov-tool.c (gcov_output_files): Validate that destination
30770 file is either removed by the tool or by a user.
30772 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30773 Guy Benyei <guybe@mellanox.com>
30775 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30776 block, and do not negate it, the stored id is already negative.
30778 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30780 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30782 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30785 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30786 masks of options that should be turned off if the VSX vector
30787 options are turned off.
30788 (OTHER_P8_VECTOR_MASKS): Likewise.
30789 (OTHER_VSX_VECTOR_MASKS): Likewise.
30790 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30791 rs6000_disable_incompatible_switches to validate no type switches
30793 (rs6000_incompatible_switch): New function to disallow turning on
30794 other vector options if -mno-vsx, -mno-power8-vector, or
30795 -mno-power9-vector are specified.
30797 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30799 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30801 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30803 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30804 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30805 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30806 (ARG_POINTER_CFA_OFFSET): Likewise.
30808 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30810 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30811 conditions to take advantage of various optimizations.
30813 2017-04-13 Jeff Law <law@redhat.com>
30815 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30816 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30817 (zero_extendsidi2_dext): Likewise.
30819 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30822 * fold-const.c (fold_ternary_loc): Revert
30823 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30826 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30828 PR rtl-optimization/80343
30829 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30830 new scratch pseudo.
30832 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30835 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30836 to ubsan_encode_value.
30838 2017-04-13 Jeff Law <law@redhat.com>
30840 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30841 appearing in DEBUG_INSNs.
30843 2017-04-13 Martin Liska <mliska@suse.cz>
30845 PR gcov-profile/80413
30846 * gcov-io.c (gcov_write_string): Copy to buffer just when
30847 allocated size is greater than zero.
30849 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30852 * dwarf2out.c (decls_for_scope): Ignore declarations of
30853 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30855 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30858 * ipa-visibility.c (non_local_p): Fix typos.
30859 (localize_node): When localizing symbol in same comdat group,
30860 dissolve the group only when we know external symbols are going
30862 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30864 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30866 PR tree-optimization/79390
30867 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30868 order does not result in usable sequence, retry with reversed operand
30874 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30875 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30876 op0 instead of fold_convert_loc (loc, type, arg0).
30878 2017-04-12 Jeff Law <law@redhat.com>
30880 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30881 has a delay slot in the generated code.
30883 * config/cris/cris.md (cris_preferred_reload_class): Return
30884 GENNONACR_REGS rather than GENERAL_REGS.
30886 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30889 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30890 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30891 signedness of the result type.
30893 2017-04-12 Richard Biener <rguenther@suse.de>
30894 Jeff Law <law@redhat.com>
30896 PR tree-optimization/80359
30897 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30898 trim stores to TARGET_MEM_REFs.
30900 2017-04-12 Richard Biener <rguenther@suse.de>
30902 PR tree-optimization/79390
30903 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30904 threading case even more.
30906 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30909 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30910 for quad_address_p for TImode, instead of just not indexed_address.
30912 2017-04-12 Richard Biener <rguenther@suse.de>
30913 Bernd Edlinger <bernd.edlinger@hotmail.de>
30915 PR middle-end/79671
30916 * alias.c (component_uses_parent_alias_set_from): Handle
30917 TYPE_TYPELESS_STORAGE.
30918 (get_alias_set): Likewise.
30919 * tree-core.h (tree_type_common): Add typeless_storage flag.
30920 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30921 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30922 for types containing members with TYPE_TYPELESS_STORAGE.
30923 (place_field): Likewise.
30924 (layout_type): Likewise for ARRAY_TYPE.
30925 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30926 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30927 TYPE_TYPELESS_STORAGE.
30928 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30930 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30933 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30934 first argument to type.
30936 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30940 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30941 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30942 (rs6000_expand_binop_builtin): Likewise.
30943 (rs6000_expand_ternop_builtin): Likewise; also add missing
30944 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30945 vshasigma built-ins.
30946 * doc/extend.texi: Document that vec_xxpermdi's third argument
30947 must be a constant.
30949 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30951 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30952 Use shift_const cost parameter when calculating gain of STV shifts.
30954 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30956 PR rtl-optimization/70478
30957 * lra-constraints.c (process_alt_operands): Check memory for
30958 disfavoring memory insn operand.
30960 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30962 PR middle-end/80100
30963 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30964 left shift in unsigned HOST_WIDE_INT type.
30966 PR rtl-optimization/80385
30967 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30968 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30971 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30972 if they have any depend clauses.
30974 2017-04-11 Martin Liska <mliska@suse.cz>
30977 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30978 * ipa-split.c (split_function): Create a local comdat symbol
30979 if caller is in a comdat group.
30981 2017-04-11 Martin Liska <mliska@suse.cz>
30984 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30987 2017-04-11 Martin Sebor <msebor@redhat.com>
30989 PR middle-end/80364
30990 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30991 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30993 (directive::set_width, directive::set_precision, format_character):
30995 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30998 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31001 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31002 conflict, set target->arch_name instead of target->cpu_name.
31004 2017-04-11 Richard Biener <rguenther@suse.de>
31006 PR tree-optimization/80374
31007 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31008 build_zero_cst, remove fold_convertible_p check again.
31010 2017-04-11 Martin Liska <mliska@suse.cz>
31013 * ubsan.c (instrument_object_size): Do not instrument register
31016 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31019 * config/i386/i386-builtin-types.def
31020 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31021 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31022 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31023 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31024 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31025 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31026 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31027 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31028 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31029 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31030 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31031 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31032 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31033 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31034 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31035 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31036 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31037 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31038 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31039 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31040 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31041 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31042 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31043 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31044 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31045 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31046 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31047 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31048 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31049 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31050 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31051 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31052 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31053 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31054 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31055 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31056 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31057 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31058 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31059 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31060 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31061 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31062 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31063 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31064 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31065 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31067 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31068 flag to second_arg_count, handle 4 argument function type _COUNT
31069 aliases, handle second_arg_count on second argument rather than last.
31071 2017-04-10 Jeff Law <law@redhat.com>
31073 PR tree-optimization/80374
31074 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31075 record anything if we can not convert integer_zero_node to the
31078 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31081 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31082 Enhance special handling given to the TARGET_P9_MINMAX option in
31083 relation to certain other options.
31085 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31087 PR tree-optimization/80153
31088 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31089 remove POINTER_PLUS_EXPR's base part directly, rather than through
31092 2017-04-10 Richard Biener <rguenther@suse.de>
31093 Bin Cheng <bin.cheng@arm.com>
31095 PR tree-optimization/80153
31096 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31097 the first element of pointer type aff_tree. Build result expr in
31099 (add_elt_to_tree): Convert to type unconditionally. Remove other
31100 fold_convert calls.
31101 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31102 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31104 2017-04-10 Richard Biener <rguenther@suse.de>
31106 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31109 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31111 PR rtl-optimization/70478
31112 * lra-constraints.c (curr_small_class_check): New.
31113 (update_and_check_small_class_inputs): New.
31114 (process_alt_operands): Update curr_small_class_check. Disfavor
31115 alternative insn memory operands. Check available regs for small
31118 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31121 * config/mips/mips.opt (-mvirt): Update description.
31122 * doc/invoke.texi (-mvirt): Likewise.
31124 2017-04-10 Richard Biener <rguenther@suse.de>
31126 PR middle-end/80362
31127 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31128 looking for NEGATE_EXPR in -A / -B to A / B folding.
31130 2017-04-10 Martin Liska <mliska@suse.cz>
31132 PR gcov-profile/80224
31133 * gcov.c (print_usage): Fix usage string.
31134 (get_gcov_intermediate_filename): Remove.
31135 (output_gcov_file): Use both for normal and intermediate format.
31136 (generate_results): Do not initialize special file for
31137 intermediate format.
31139 2017-04-10 Richard Biener <rguenther@suse.de>
31141 PR tree-optimization/80304
31142 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31145 2017-04-10 Nathan Sidwell <nathan@acm.org>
31148 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31149 (rs6000_init_builtins): Use it.
31151 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31153 * config/arm/arm.md (<mrc>): Add mode to SET source.
31154 (<mrrc>): Likewise.
31156 2017-04-10 Richard Biener <rguenther@suse.de>
31158 PR middle-end/80344
31159 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31161 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31164 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31165 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31166 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31167 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31168 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31169 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31170 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31171 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31172 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31173 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31174 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31175 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31176 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31177 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31178 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31179 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31180 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31181 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31182 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31183 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31184 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31185 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31186 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31188 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31190 PR rtl-optimization/70478
31191 * lra-constraints.c: Reverse the last patch.
31193 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31195 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31196 Add comment for WCHAR_T.
31198 2017-04-08 Martin Liska <mliska@suse.cz>
31201 2017-04-07 Martin Liska <mliska@suse.cz>
31204 * ipa-split.c (split_function): Add function part to a same comdat
31207 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31210 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31212 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31214 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31215 (vec_pairing): ...to this.
31216 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31217 (rs6000_sched_init): Adjust for name change.
31218 (struct rs6000_sched_context): Likewise.
31219 (rs6000_init_sched_context): Likewise.
31220 (rs6000_set_sched_context): Likewise.
31222 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31228 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31230 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31231 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31232 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31234 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31236 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31238 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31240 PR rtl-optimization/70703
31241 * ira-color.c (update_conflict_hard_regno_costs): Use
31242 int64_t instead of HOST_WIDE_INT.
31244 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31246 PR rtl-optimization/70478
31247 * lra-constraints.c (process_alt_operands): Disfavor alternative
31248 insn memory operands.
31250 2017-04-07 Jeff Law <law@redhat.com>
31252 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31253 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31255 2017-04-07 Martin Liska <mliska@suse.cz>
31258 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31259 Show error message instead of an ICE.
31261 2017-04-07 Martin Liska <mliska@suse.cz>
31264 * ipa-split.c (split_function): Add function part to a same comdat
31267 2017-04-07 Richard Biener <rguenther@suse.de>
31269 PR middle-end/80341
31270 * tree.c (get_unwidened): Also handle ! for_type case for
31272 * convert.c (do_narrow): Split out from ...
31273 (convert_to_integer_1): ... here. Do not pass final truncation
31274 type to get_unwidened for TRUNC_DIV_EXPR.
31276 2017-04-07 Richard Biener <rguenther@suse.de>
31278 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31280 (aff_combination_const): Adjust.
31281 (aff_combination_scale): Likewise.
31282 (aff_combination_add_elt): Likewise.
31283 (aff_combination_add_cst): Likewise.
31284 (aff_combination_convert): Likewise.
31285 (add_elt_to_tree): Likewise. Remove unused argument.
31286 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31288 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31290 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31292 * config/arm/arm.c (arm_default_short_enums): Use
31293 ARM_DEFAULT_SHORT_ENUMS.
31294 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31296 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31299 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31300 members with redundant out-of-class redeclaration.
31302 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31305 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31306 * config/i386/i386.md (*zero_extendsidi2):
31307 Add (?*x,*x) and (?*v,*v) alternatives.
31309 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31312 * config/i386/i386.c (ix86_expand_builtin)
31313 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31314 mode from insn data. Convert operands to insn operand mode.
31315 Copy operands that don't satisfy insn predicate to a register.
31317 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31319 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31322 2017-04-06 Richard Biener <rguenther@suse.de>
31324 PR tree-optimization/80334
31325 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31326 preserve alignment of accesses.
31328 2017-04-06 Richard Biener <rguenther@suse.de>
31330 PR tree-optimization/80262
31331 * tree-sra.c (build_ref_for_offset): Preserve address-space
31333 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31334 Drop useless address-space information on MEM_REF offsets.
31336 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31338 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31340 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31342 PR rtl-optimization/70703
31343 * ira-color.c (update_conflict_hard_regno_costs): Use
31344 HOST_WIDE_INT instead of long.
31346 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31349 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31350 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31352 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31353 for x86_64 target. Handle -m3dnowa option.
31355 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31357 PR rtl-optimization/70703
31358 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31359 (update_conflict_hard_regno_costs): Use long instead of unsigned
31360 arithmetic for cost calculation.
31362 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31363 Bernd Edlinger <bernd.edlinger@hotmail.de>
31366 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31369 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31372 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31373 ptr_mode with Pmode throughout.
31374 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31375 into probe_stack_range and use DImode.
31377 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31380 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31381 call_eh_return is true.
31383 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31385 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31386 Initialize last_match_fntype_index.
31388 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31391 * tree-nvr.c: Include internal-fn.h.
31392 (pass_return_slot::execute): Ignore internal calls without
31395 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31396 Richard Biener <rguenther@suse.de>
31399 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31400 captures used multiple times, except for the last use.
31401 * generic-match-head.c: Include gimplify.h.
31403 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31405 PR tree-optimization/79390
31406 * target.h (struct noce_if_info): Declare.
31407 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31408 * target.def (noce_conversion_profitable_p): New target hook.
31409 * ifcvt.h (struct noce_if_info): New type, moved from ...
31410 * ifcvt.c (struct noce_if_info): ... here.
31411 (noce_conversion_profitable_p): Renamed to ...
31412 (default_noce_conversion_profitable_p): ... this. No longer
31414 (noce_try_store_flag_constants, noce_try_addcc,
31415 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31416 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31417 instead of noce_conversion_profitable_p.
31418 * config/i386/i386.c: Include ifcvt.h.
31419 (ix86_option_override_internal): Don't override
31420 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31421 (ix86_noce_conversion_profitable_p): New function.
31422 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31423 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31424 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31425 * doc/tm.texi: Regenerated.
31427 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31429 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31432 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31435 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31436 instructions for small multiply cores.
31438 2017-04-04 Jeff Law <law@redhat.com>
31440 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31442 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31443 that are not set by the loop over the elements.
31445 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31448 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31449 int mode, convert_modes it to mode as unsigned, otherwise use
31450 lowpart_subreg to mode rather than SImode.
31451 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31452 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31453 Use DImode instead of SImode for the shift count operand.
31454 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31457 2017-04-04 Richard Biener <rguenther@suse.de>
31459 PR middle-end/80281
31460 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31461 arithmetic done for the negate or the plus. Simplify.
31462 (A - (-B) -> A + B): Likewise.
31463 * fold-const.c (split_tree): Make sure to not negate pointers.
31465 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31467 PR rtl-optimization/60818
31468 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31469 a compare of comparisons with the thing compared if this results
31470 in a different machine mode.
31472 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31474 * alias.c (base_alias_check): Fix typo in comment.
31475 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31476 * cgraphunit.c (symbol_table::compile): Likewise.
31477 * collect2.c (maybe_run_lto_and_relink): Likewise.
31478 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31479 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31480 * config/avr/avr.c (avr_map_op_t): Likewise.
31481 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31482 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31483 * config/epiphany/epiphany.md (movcc): Likewise.
31484 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31485 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31487 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31488 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31489 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31490 * config/sh/sh.c (sh_rtx_costs): Likewise.
31491 * fold-const.c (fold_truth_andor): Likewise.
31492 * genautomata.c (collapse_flag): Likewise.
31493 * gengtype.h (struct type::u::s): Likewise.
31494 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31495 * input.c (FORMAT_AMOUNT): Likewise.
31496 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31497 (known_aggs_to_agg_replacement_list): Likewise.
31498 * ipa-inline-analysis.c: Likewise.
31499 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31500 * ipa-polymorphic-call.c
31501 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31502 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31503 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31505 * modulo-sched.c (apply_reg_moves): Likewise.
31506 * omp-expand.c (build_omp_regions_1): Likewise.
31507 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31508 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31509 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31510 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31511 * value-prof.c: Likewise.
31512 * var-tracking.c (val_reset): Likewise.
31514 2017-04-03 Richard Biener <rguenther@suse.de>
31516 PR tree-optimization/80275
31517 * fold-const.c (split_address_to_core_and_offset): Handle
31520 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31522 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31523 descriptors is at least equal to that of functions.
31525 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31527 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31529 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31532 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31533 (mov<IMOD4:mode>): New expander.
31534 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31536 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31538 PR rtl-optimization/79405
31539 * fwprop.c (propagations_left): New variable.
31540 (forward_propagate_into): Decrement it.
31541 (fwprop_init): Initialize it.
31542 (fw_prop): If the variable has reached zero, stop propagating.
31543 (fwprop_addr): Ditto.
31545 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31548 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31549 a FUNCTION_DECL, pass it as decl instead of origin to
31552 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31554 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31557 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31560 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31561 TARGET_VSX_SMALL_INTEGER.
31563 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31565 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31566 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31568 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31570 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31571 extraction from odd-numbered MSA register.
31573 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31575 PR middle-end/80173
31576 * expmed.c (store_bit_field_1): Don't attempt to create
31577 a word subreg out of hard registers wider than word if they
31578 have HARD_REGNO_NREGS of 1 for their mode.
31580 PR middle-end/80163
31581 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31582 conversions to integer types wider than word and pointer.
31585 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31586 (rtx_equal_for_cselib_p): Pass 0 to it.
31587 * cselib.c (cselib_hasher::equal): Likewise.
31588 (rtx_equal_for_cselib_1): Add depth argument. If depth
31589 is 128, don't look up VALUE locs and punt. Increment
31590 depth in recursive calls when walking VALUE locs.
31592 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31594 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31595 (make_gcov_file_name): Use the canonical path name for generating
31597 (read_line): Fix handling of files with ascii null bytes.
31599 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31601 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31602 to initialise a vector register instead
31603 of using a const_int.
31605 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31607 PR translation/80189
31608 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31609 diagnostic messages.
31611 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31614 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31615 (dfp_diex_<mode>): Update mode of operand 1.
31616 * doc/extend.texi (dxex, dxexq): Document change to return type.
31617 (diex, diexq): Document change to argument type.
31619 2017-03-30 Martin Jambor <mjambor@suse.cz>
31622 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31623 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31624 it reflects the signature changes performed at the callee side.
31625 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31626 to cgraph_build_function_type_skip_args.
31627 (build_function_decl_skip_args): Adjust call to the above function.
31629 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31632 * config/i386/sse.md
31633 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31634 register as dest whenever it is a MEM not rtx_equal_p to the
31635 corresponding dup operand, and when forcing into reg move the
31636 reg into the memory afterwards.
31637 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31638 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31639 for the force_reg mode.
31640 (avx512vl_vextractf128<mode>): Use register as dest either
31641 always when a MEM, or when it is a MEM not rtx_equal_p to the
31642 corresponding dup operand, or even not when it is a CONST_VECTOR
31643 depending on the mode and lo vs. hi.
31644 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31646 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31647 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31648 Likewise. Require that operands[2] is even.
31649 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31650 Remove extraneous parens. Require that operands[2] is a multiple
31652 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31653 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31654 disallow memory then.
31656 2017-03-30 Richard Biener <rguenther@suse.de>
31658 PR tree-optimization/77498
31659 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31660 to non-constants over backedges.
31662 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31664 PR rtl-optimization/80233
31665 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31666 as last_combined_insn. Do not test for BARRIER_P separately.
31668 2017-03-29 Andreas Schwab <schwab@suse.de>
31671 * calls.c (prepare_call_address): Convert funexp to Pmode before
31672 copying to temp reg.
31674 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31676 PR tree-optimization/80158
31677 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31678 Handle possible future case of more than one alternate
31680 (replace_rhs_if_not_dup): Likewise.
31681 (replace_one_candidate): Likewise.
31683 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31685 PR rtl-optimization/80193
31686 * ira.c (ira): Do not check allocation for LRA.
31688 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31690 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31691 (nvptx_output_simt_exit): Declare.
31692 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31693 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31694 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31695 (nvptx_declare_function_name): ...here. Emit declaration of local
31696 memory space buffer for omp_simt_enter insn.
31697 (nvptx_output_unisimt_switch): New.
31698 (nvptx_output_softstack_switch): New.
31699 (nvptx_output_simt_enter): New.
31700 (nvptx_output_simt_exit): New.
31701 * config/nvptx/nvptx.h (struct machine_function): New fields
31702 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31703 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31704 (UNSPECV_SIMT_EXIT): Ditto.
31705 (omp_simt_enter_insn): New insn.
31706 (omp_simt_enter): New expansion.
31707 (omp_simt_exit): New insn.
31708 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31710 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31711 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31712 (expand_GOMP_SIMT_EXIT): New.
31713 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31714 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31715 (GOMP_SIMT_EXIT): Ditto.
31716 * target-insns.def (omp_simt_enter): New insn.
31717 (omp_simt_exit): Ditto.
31718 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31720 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31721 (lower_rec_input_clauses): Likewise.
31722 (lower_lastprivate_clauses): Handle SIMT privatization.
31724 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31725 (ompdevlow_adjust_simt_enter): New.
31726 (find_simtpriv_var_op): New.
31727 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31728 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31730 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31731 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31732 (copy_decl_for_dup_finish): Ditto.
31734 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31736 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31739 * config/i386/i386.c (ix86_option_override_internal): Always
31740 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31742 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31744 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31746 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31748 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31749 mark new edge's irreducible flag accordign to it.
31750 (vect_do_peeling): Check loop preheader edge's irreducible flag
31751 and pass it to function slpeel_add_loop_guard.
31753 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31755 PR tree-optimization/80218
31756 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31757 Update block frequencies and counts.
31759 2017-03-28 Richard Biener <rguenther@suse.de>
31761 PR tree-optimization/78644
31762 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31763 of a simplification result we may not use it at all.
31765 2017-03-28 Richard Biener <rguenther@suse.de>
31768 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31769 without arguments, generate default definition of a SSA name.
31771 2017-03-28 Richard Biener <rguenther@suse.de>
31773 PR middle-end/80222
31774 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31775 TYPE_REF_CAN_ALIAS_ALL references.
31776 * fold-const.c (fold_indirect_ref_1): Likewise.
31778 2017-03-28 Martin Liska <mliska@suse.cz>
31781 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31782 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31784 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31785 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31787 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31788 (EXTRA_SPECS): Define.
31789 (SUBTARGET_EXTRA_SPECS): Likewise.
31790 (SUBTARGET_CPP_SPEC): Likewise.
31791 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31792 SUBTARGET_EXTRA_SPECS.
31793 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31795 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31797 * config/arc/simdext.md (vst64_insn): Update pattern.
31798 (vld32wh_insn): Likewise.
31799 (vld32wl_insn): Likewise.
31800 (vld64_insn): Likewise.
31801 (vld32_insn): Likewise.
31803 2017-03-28 Marek Polacek <polacek@redhat.com>
31806 * fold-const.c (fold_comparison): Use protected_set_expr_location
31807 instead of SET_EXPR_LOCATION.
31809 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31811 * tree.c (add_expr): Avoid name lookup warning.
31813 2017-03-27 Jeff Law <law@redhat.com>
31815 PR tree-optimization/80216
31816 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31817 function name. Limit recursion depth.
31818 (record_temporary_equivalences): Corresponding changes.
31820 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31822 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31825 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31828 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31830 * cfgcleanup.c (reg_note_cfa_p): New array.
31831 (insns_have_identical_cfa_notes): New function.
31832 (old_insns_match_p): Don't cross-jump in between /f
31833 and non-/f instructions. If both i1 and i2 are frame related,
31834 verify all CFA notes, their order and content.
31836 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31839 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31840 HImode and SImode with zero extend to DImode to one insn.
31841 (bswap<mode>2_extenddi): Likewise.
31842 (bswapsi2_extenddi): Likewise.
31843 (bswaphi2_extendsi): Likewise.
31844 (bswaphi2): Combine bswap HImode and SImode into one insn.
31845 Separate memory insns from swapping register.
31846 (bswapsi2): Likewise.
31847 (bswap<mode>2): Likewise.
31848 (bswaphi2_internal): Delete, no longer used.
31849 (bswapsi2_internal): Likewise.
31850 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31851 store, and gpr<-gpr swap insns.
31852 (bswap<mode>2_store): Likewise.
31853 (bswaphi2_reg): Register only splitter, combine with the splitter.
31854 (bswaphi2 splitter): Likewise.
31855 (bswapsi2_reg): Likewise.
31856 (bswapsi2 splitter): Likewise.
31857 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31858 the insns into load, store, and register/register insns.
31859 (bswapdi2_ldbrx): Likewise.
31860 (bswapdi2_load): Likewise.
31861 (bswapdi2_store): Likewise.
31862 (bswapdi2_reg): Likewise.
31864 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31866 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31867 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31869 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31872 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31874 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31875 special handling for target option conflicts between dform
31876 options (-mpower9-dform, -mpower9-dform-vector,
31877 -mpower9-dform-scalar) and -mno-direct-move.
31879 2017-03-27 Richard Biener <rguenther@suse.de>
31881 PR tree-optimization/80181
31882 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31884 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31886 * config/arc/predicates.md (move_double_src_operand): Replace the
31887 call to move_double_src_operand with a call to address_operand.
31889 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31891 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31892 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31893 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31895 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31897 * config/arc/predicates.md (long_immediate_loadstore_operand):
31898 Consider scaled addresses cases.
31900 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31902 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31903 restored when in interrupt.
31904 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31905 doesn't have delay slot.
31907 2017-03-27 Richard Biener <rguenther@suse.de>
31910 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31911 inlined thunk clones.
31913 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31916 * asan.c (instrument_derefs): Copy over last operand from
31917 original COMPONENT_REF to the new COMPONENT_REF with
31918 DECL_BIT_FIELD_REPRESENTATIVE.
31919 * ubsan.c (instrument_object_size): Likewise.
31921 2017-03-27 Richard Biener <rguenther@suse.de>
31923 PR tree-optimization/80170
31924 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31925 sure DR/SCEV didnt fold in constants we do not see when looking
31926 at the reference base alignment.
31928 2017-03-27 Richard Biener <rguenther@suse.de>
31930 PR middle-end/80171
31931 * gimple-fold.c (fold_ctor_reference): Properly guard against
31932 NULL return value from canonicalize_constructor_val.
31934 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31937 * config/i386/i386.c (ix86_expand_builtin)
31938 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31939 flags reg setting and flags reg using instructions.
31940 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31941 clobbering instructions to zero extend op2.
31943 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31945 * doc/install.texi (Configuration) <--with-aix-soname>:
31946 Update link to AIX ld.
31948 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31950 PR rtl-optimization/80160
31951 PR rtl-optimization/80159
31952 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31953 reg_alternate_class into account.
31955 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31958 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31959 to consider in curr_insn_transform.
31961 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31963 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31964 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31965 and emit_mode_inner.
31967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31969 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31970 argument to the overloaded builtin variants. Use the new flag to
31971 deprecate certain builtin variants.
31972 * config/s390/s390-builtin-types.def: Add new builtin types.
31973 * config/s390/s390-builtins.h: Support new flags field for
31974 overloaded builtins.
31975 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31976 (s390_macro_to_expand): Enable vector float data type.
31977 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31978 builtins by incrementing the __VEC__ version number.
31979 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31981 (s390_resolve_overloaded_builtin): Emit error messages depending
31982 on the builtin flags.
31983 * config/s390/s390.c (s390_expand_builtin): Support additional
31984 flags argument. Change error message to match the messages
31985 emitted in s390-c.c.
31986 * config/s390/s390.md: New UNSPEC_* constants.
31987 (op_type): Add new instruction types.
31988 * config/s390/vecintrin.h: Add new builtins and test data class
31990 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31991 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31992 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31993 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31994 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31995 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31997 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31998 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31999 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32000 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32002 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32003 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32004 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32005 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32007 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32008 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32009 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32010 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32011 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32012 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32013 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32015 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32016 ("vec_scatter_element<V_HW_4:mode>_DI")
32017 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32018 ("vec_fpint<mode>", "vflls")
32019 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32020 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32021 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32022 ("*vec_cmphe<mode>_cc"): ... these.
32024 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32025 mode constant instead of magic value.
32027 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32029 * config/s390/s390.c (s390_expand_vec_compare): Support other
32030 vector floating point modes than just V2DF.
32031 (s390_expand_vcond): Likewise.
32032 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32033 (s390_cannot_change_mode_class): Prevent mode changes between TF
32034 and V1TF in vector registers.
32035 * config/s390/s390.md (DF, SF): New mode attributes.
32036 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32037 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32038 SFmode support for VRs.
32039 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32041 (VFT, VF_HW): New mode iterators.
32042 (vw, sdx): New mode attributes.
32043 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32044 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32045 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32046 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32047 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32048 also the new vector floating point modes. Renaming to ...
32050 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32051 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32052 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32053 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32054 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32055 ("vec_unordered<mode>"): ... these.
32057 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32058 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32059 ("*vec_extendv2df"): New insn definitions.
32061 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32063 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32064 ("mulditi3_2", "*muldi3_sign"): New patterns.
32065 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32066 rename the pattern definition.
32068 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32070 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32072 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32074 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32076 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32077 instruction if possible.
32078 * config/s390/vector.md (vec_halfnumelts): New mode
32080 ("*vec_vllezlf<mode>"): New pattern.
32082 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32084 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32085 ("popcountv4si2", "popcountv2di2"): Rename to ...
32086 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32087 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32089 ("popcount<mode>2_vxe"): New pattern.
32091 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32093 * common/config/s390/s390-common.c (processor_flags_table): Add
32095 * config.gcc: Add arch12.
32096 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32097 Default to arch12 for unknown CPU model numbers.
32098 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32099 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32100 PROCESSOR_max sanity check.
32101 * config/s390/s390-opts.h (enum processor_type): Add
32103 * config/s390/s390.c (processor_table): Add arch12.
32104 (s390_expand_builtin): Add check for B_VXE flag.
32105 (s390_issue_rate): Add PROCESSOR_ARCH12.
32106 (s390_get_sched_attrmask): Likewise.
32107 (s390_get_unit_mask): Likewise.
32108 (s390_sched_score): Enable z13 scheduling for arch12.
32109 (s390_sched_reorder): Likewise.
32110 (s390_sched_variable_issue): Likewise.
32111 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32113 (s390_tune_attr): Use z13 scheduling also for arch12.
32114 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32115 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32116 (TARGET_VXE_P): New macros.
32117 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32118 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32119 * config/s390/s390.opt: Add arch12 as processor_type.
32121 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32123 * config/s390/s390.md
32124 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32125 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32126 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32128 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32129 Rename expanders to ...
32131 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32132 ("fixuns_truncdddi2_emu"): ... these.
32134 ("fixuns_trunc<mode>si2_emu"): New expander.
32136 ("*fixuns_truncdfdi2_z13"): Rename to ...
32137 ("*fixuns_truncdfdi2_vx"): ... this.
32139 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32141 * config/s390/2964.md: Remove the single element vector compare
32142 instructions which are no longer used.
32143 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32145 (s390_canonicalize_comparison): Remove handling of DFmode
32147 (s390_expand_vec_compare_scalar): Remove function.
32148 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32149 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32151 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32153 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32155 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32157 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32158 will anyway by matched by mov<mode>_64dfp.
32160 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32162 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32163 vlef/vstef. Add missing operand to vleif.
32165 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32167 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32168 pair for all vector types with 64 bit elements.
32169 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32170 * config/s390/vector.md (V_HW_64): ... here.
32171 (V_128_NOSINGLE): New mode iterator.
32172 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32173 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32174 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32175 ("*vec_load_pairv2di"): Change to ...
32176 ("*vec_load_pair<mode>"): ... this one.
32178 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32180 * config/s390/constraints.md: Add comments.
32181 (jKK): Reject element sizes > 8 bytes.
32182 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32184 * config/s390/s390.md: Add the s_operand checks formerly in
32185 s390_split_ok_p to various splitters where they are still
32187 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32188 for 128 bit vectors. Plus two splitters.
32190 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32192 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32195 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32198 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32199 error if the boundary argument is not constant.
32201 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32203 PR rtl-optimization/80112
32204 * loop-doloop.c (doloop_condition_get): Don't check condition
32205 if cmp isn't SET with IF_THEN_ELSE src.
32207 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32209 PR tree-optimization/80158
32210 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32211 replacing a candidate statement, also replace it for the
32212 candidate's alternate interpretation.
32213 (replace_rhs_if_not_dup): Likewise.
32214 (replace_one_candidate): Likewise.
32216 2017-03-24 Richard Biener <rguenther@suse.de>
32218 PR tree-optimization/80167
32219 * graphite-isl-ast-to-gimple.c
32220 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32222 (translate_isl_ast_to_gimple::get_rename): Likewise.
32224 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32226 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32227 handling of certain combinations of target options, including the
32228 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32229 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32231 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32234 * config/arm/arm.md (*load_multiple): Add reload_completed to
32235 matching condition.
32237 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32238 Richard Biener <rguenth@suse.de>
32240 PR tree-optimization/79908
32241 PR tree-optimization/80136
32242 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32243 been cast away, gimplify_and_add suffices.
32245 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32247 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32249 2017-03-23 Richard Biener <rguenther@suse.de>
32251 PR tree-optimization/80032
32252 * gimplify.c (gimple_push_cleanup): Forced unconditional
32253 cleanups still have to go to the conditional_cleanups
32256 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32258 PR tree-optimization/80072
32259 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32261 (next_operand_entry_id): Change type to unsigned int.
32262 (sort_by_operand_rank): Make sure to return the right return value
32263 even if unsigned fields are bigger than INT_MAX.
32264 (struct oecount): Change cnt and id type to unsigned int.
32265 (oecount_hasher::equal): Formatting fix.
32266 (oecount_cmp): Make sure to return the right return value
32267 even if unsigned fields are bigger than INT_MAX.
32268 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32271 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32272 TREE_READONLY on result if writing it more than once.
32275 * doc/invoke.texi (-fsanitize=thread): Document that with
32276 -fnon-call-exceptions atomics are not able to throw
32280 * tsan.c: Include tree-eh.h.
32281 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32282 maybe_clean_or_replace_eh_stmt where needed.
32283 (instrument_memory_accesses): Add cfg_changed argument.
32284 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32285 if it returned true.
32286 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32288 PR rtl-optimization/63191
32289 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32290 wrapper function, moved the whole old content into ...
32291 (ix86_delegitimize_address_1): ... this. New inline function.
32292 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32293 true as last argument instead of ix86_delegitimize_address.
32295 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32297 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32298 cortexa57_branch_cost.
32300 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32302 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32304 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32307 * doc/md.texi (Constraints): Document wA constraint.
32308 * config/rs6000/constraints.md (wA): New.
32309 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32310 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32311 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32312 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32314 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32317 * gimplify.c (is_oacc_declared): New function.
32318 (oacc_default_clause): Use it to set default flags for acc declared
32319 variables inside parallel regions.
32320 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32321 declared variables.
32322 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32323 declare attribute to any decl as necessary.
32325 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32328 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32329 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32330 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32331 (arm_arch_lpae): This.
32332 * config/arm/arm.c (arm_arch7ve): Rename into ...
32333 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32334 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32337 2017-03-22 Martin Liska <mliska@suse.cz>
32340 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32341 error message instead of an ICE.
32343 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32345 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32347 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32349 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32351 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32354 2017-03-21 Martin Sebor <msebor@redhat.com>
32356 * doc/extend.texi: Use "cannot" instead of "can't."
32357 * doc/hostconfig.texi: Same.
32358 * doc/install.texi: Same.
32359 * doc/invoke.texi: Same.
32360 * doc/loop.texi: Same.
32361 * doc/md.texi: Same.
32362 * doc/objc.texi: Same.
32363 * doc/rtl.texi: Same.
32364 * doc/tm.texi: Same.
32365 * doc/tm.texi.in: Same.
32366 * doc/trouble.texi: Same.
32368 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32371 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32372 (collect_checksum_attributes): Set it.
32373 (die_checksum_ordered): Use it.
32375 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32377 PR tree-optimization/79908
32378 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32379 change: For a VA_ARG whose LHS has been cast away, use
32380 force_gimple_operand to construct the side effects.
32382 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32384 PR translation/80001
32385 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32386 more amenable to translation.
32387 (oacc_loop_auto_partitions): Likewise.
32389 2017-03-21 Marek Polacek <polacek@redhat.com>
32390 Martin Sebor <msebor@redhat.com>
32392 PR tree-optimization/80109
32393 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32394 on INTEGRAL_TYPE_P.
32396 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32397 Segher Boessenkool <segher@kernel.crashing.org>
32400 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32401 check reg_used_between_p between insn and one of succ or succ2
32402 depending on if succ is artificial insn not inserted into insn
32405 2017-03-21 Martin Liska <mliska@suse.cz>
32407 PR gcov-profile/80081
32408 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32409 * doc/gcc.texi: Include gcov-dump stuff.
32410 * doc/gcov-dump.texi: New file.
32412 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32414 PR rtl-optimization/79150
32415 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32416 conditional jump, if the jump is the last insn of the loop.
32418 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32419 Richard Biener <rguenth@suse.de>
32421 PR tree-optimization/79908
32422 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32423 been cast away, use force_gimple_operand to construct the side
32426 2017-03-21 Martin Liska <mliska@suse.cz>
32428 PR libfortran/79956
32429 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32432 2017-03-21 Brad Spengler <spender@grsecurity.net>
32435 * plugin.c (htab_hash_plugin): New function.
32436 (add_new_plugin): Use it and adjust.
32437 (parse_plugin_arg_opt): Adjust.
32438 (init_one_plugin): Likewise.
32440 2017-03-21 Richard Biener <rguenther@suse.de>
32442 PR tree-optimization/80032
32443 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32444 if set force the cleanup to happen unconditionally.
32445 (gimplify_target_expr): Push inserted clobbers with force_uncond
32446 to avoid them being removed by control-dependent DCE.
32448 2017-03-21 Richard Biener <rguenther@suse.de>
32450 PR tree-optimization/80122
32451 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32452 va_arg_pack_len when the inlined call stmt requires pack
32454 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32456 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32459 * tsan.c (instrument_builtin_call): If the memory model argument
32460 is not a constant, assume it is valid.
32463 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32466 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32468 PR rtl-optimization/79910
32469 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32470 if its dest is used by an insn before I2 (other than the combined
32471 insns themselves, which are properly handled already).
32473 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32476 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32478 * combine.c (record_used_regs): New static function.
32479 (try_combine): Handle situations where there is an additional
32480 instruction between I2 and I3 which needs to have a LOG_LINK
32484 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32486 * combine.c (try_combine): Delete redundant i1 test. Call
32487 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32489 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32492 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32493 alternatives 13/14.
32495 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32497 PR tree-optimization/80054
32498 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32499 the optimization if a PHI or any of its arguments is not dominated
32500 by the candidate's basis. Use gphi* rather than gimple* as
32502 (replace_profitable_candidates): Clean up a gimple* variable that
32503 should be a gphi* variable.
32505 2017-03-20 Martin Sebor <msebor@redhat.com>
32508 * doc/extend.texi (attribute constructor): Document present limitation.
32510 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32513 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32514 __POWER9_VECTOR__ #ifdef control, change template definition to
32515 use Power9-specific built-in function.
32516 (vec_any_eq): Likewise.
32517 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32518 to control outcomes from this test.
32519 (vector_ae_<mode>p): For VEC_F modes, likewise.
32521 2017-03-20 Ian Lance Taylor <iant@google.com>
32523 * config/i386/i386.c (ix86_function_regparm): Save an extra
32524 register for -fsplit-stack with DECL_STATIC_CHAIN.
32526 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32529 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32530 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32532 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32534 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32536 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32539 2017-03-20 Marek Polacek <polacek@redhat.com>
32542 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32544 2017-03-20 Richard Biener <rguenther@suse.de>
32546 PR tree-optimization/80113
32547 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32548 allocate extra SSA name for PHI def.
32549 (add_close_phis_to_outer_loops): Likewise.
32550 (add_close_phis_to_merge_points): Likewise.
32551 (copy_loop_close_phi_args): Likewise.
32552 (copy_cond_phi_nodes): Likewise.
32554 2017-03-20 Martin Liska <mliska@suse.cz>
32556 PR middle-end/79753
32557 * tree-chkp.c (chkp_build_returned_bound): Do not build
32558 returned bounds for a LHS that's not a BOUNDED_P type.
32560 2017-03-20 Martin Liska <mliska@suse.cz>
32564 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32565 COMPLEX_CST and VECTOR_CST.
32567 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32570 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32571 target operand. A new splitter adds the clobber statement in case
32572 the target operand is dead anyway.
32574 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32576 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32577 to age-old versions of binutils and glibc.
32579 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32581 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32583 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32585 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32587 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32589 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32590 requirement for binutils 2.13.
32592 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32594 * combine.c (try_combine): Delete redundant i1 test. Call
32595 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32597 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32599 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32600 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32602 <riscv64-*-elf>: Re-arrange section
32603 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32604 <riscv32-*-linux>: Likewise.
32605 <riscv64-*-elf>: Likewise
32606 <riscv64-*-linux>: Likewise.
32608 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32611 * aarch64.opt(verbose-cost-dump): Fix typo.
32613 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32616 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32617 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32619 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32621 * reload.c (find_reloads): When reloading a nonoffsettable address,
32622 use RELOAD_OTHER for it and its address reloads.
32624 PR rtl-optimization/79910
32625 * combine.c (record_used_regs): New static function.
32626 (try_combine): Handle situations where there is an additional
32627 instruction between I2 and I3 which needs to have a LOG_LINK
32630 2017-03-17 Jeff Law <law@redhat.com>
32632 PR tree-optimization/71437
32633 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32634 conditional in the hash table first.
32635 (vrp_dom_walker::before_dom_children): Extract condition from
32636 ASSERT_EXPR. Record condition, its inverion and any implied
32637 conditions as well.
32639 2017-03-17 Marek Polacek <polacek@redhat.com>
32640 Markus Trippelsdorf <markus@trippelsdorf.de>
32642 PR tree-optimization/80079
32643 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32646 2017-03-17 Richard Biener <rguenther@suse.de>
32648 PR middle-end/80075
32649 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32650 Properly verify the LHS before the RHS possibly claims to be
32652 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32655 2017-03-17 Martin Jambor <mjambor@suse.cz>
32657 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32658 (List of -O2 options): Likewise.
32659 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32662 2017-03-17 Tom de Vries <tom@codesourcery.com>
32664 * gcov-dump.c (print_usage): Print bug_report_url.
32666 2017-03-17 Richard Biener <rguenther@suse.de>
32668 PR middle-end/80050
32669 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32670 (parser::peek): Likewise.
32672 2017-03-17 Richard Biener <rguenther@suse.de>
32674 PR tree-optimization/80048
32675 * sese.c (free_sese_info): Properly release rename_map and
32676 copied_bb_map elements.
32678 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32680 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32681 Add linked-list forward and backlinks. Insert on
32682 construction, remove on destruction.
32683 (class pass_store_merging): Add m_stores_head field.
32684 (pass_store_merging::terminate_and_process_all_chains):
32685 Iterate over m_stores_head list.
32686 (pass_store_merging::terminate_all_aliasing_chains):
32688 (pass_store_merging::execute): Check for debug stmts first.
32689 Push new chains onto the m_stores_head stack.
32691 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32694 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32695 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32696 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32698 2017-03-16 Jeff Law <law@redhat.com>
32700 PR tree-optimization/71437
32701 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32702 member function. Implementation moved into after_dom_children
32703 member function and into the threader's thread_outgoing_edges
32705 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32706 some code into new thread_outgoing_edges.
32707 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32708 definition. Simplify marker handling (do it here). Assume we always
32709 have the available expression and the const/copies tables.
32710 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32712 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32713 * tree-vrp.c (equiv_stack): No longer file scoped.
32714 (vrp_dom_walker): New class.
32715 (vrp_dom_walker::before_dom_children): New member function.
32716 (vrp_dom_walker::after_dom_children): Likewise.
32717 (identify_jump_threads): Setup domwalker. Use it rather than
32718 walking edges in a random order by hand. Simplify setup/finalization.
32719 (finalize_jump_threads): Remove.
32720 (vrp_finalize): Do not call identify_jump_threads here.
32721 (execute_vrp): Do it here instead and call thread_through_all_blocks
32724 PR tree-optimization/71437
32725 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32727 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32729 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32730 (dom_opt_dom_walker::thread_across_edge): Remove
32731 handle_dominating_asserts argument. All callers changed.
32732 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32733 changes. Remove calls to lhs_of_dominating_assert. Other
32734 uses of handle_dominating_asserts turn into unconditional code
32735 (simplify_control_stmt_condition_1): Likewise.
32736 (simplify_control_stmt_condition): Likewise.
32737 (thread_through_normal_block, thread_across_edge): Likewise.
32738 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32739 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32740 object if it is not an SSA_NAME.
32741 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32742 before calling into the VRP specific simplifiers.
32743 (identify_jump_threads): Remove handle_dominating_asserts
32746 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32749 * tree-diagnostic.c (default_tree_printer): No longer static.
32750 * tree-diagnostic.h (default_tree_printer): New prototype.
32752 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32754 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32755 Change ins into fmov.
32757 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32759 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32760 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32761 Use h_con constraint for operand 1.
32762 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32763 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32765 2017-03-15 Jeff Law <law@redhat.com>
32767 PR tree-optimization/71437
32768 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32769 (record_temporary_equivalences): Use it.
32771 PR tree-optimization/71437
32772 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32773 tree-ssa-scopedtables.
32774 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32775 (record_conditions, record_cond, vuse_eq): Likewise.
32776 (record_edge_info): Adjust to API tweak of record_conditions.
32777 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32778 (record_temporary_equivalences, optimize_stmt): Likewise.
32779 (eliminate_redundant_computations): Likewise.
32780 (record_equivalences_from_stmt): Likewise.
32781 * tree-ssa-scopedtables.c: Include options.h and params.h.
32782 (vuse_eq): New function, moved from tree-ssa-dom.c
32783 (build_and_record_new_cond): Likewise.
32784 (record_conditions): Likewise. Accept vector of conditions rather
32785 than edge_equivalence structure for first argument.
32786 for the first argument.
32787 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32788 from tree-ssa-dom.c.
32789 (avail_exprs_stack::record_cond): Likewise.
32790 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32791 from tree-ssa-dom.c.
32792 (avail_exprs_stack): Add new member functions lookup_avail_expr
32794 (record_conditions): Declare.
32796 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32799 * lra-constraints.c (process_alt_operands): Increase reject for
32800 reloading an input/output operand.
32802 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32805 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32806 insns to convert from signed/unsigned char/short to IEEE 128-bit
32808 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32810 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32813 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32814 subreg of inner mode for values already in registers.
32816 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32818 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32819 iteration reg is used after the loop.
32821 2017-03-14 Martin Sebor <msebor@redhat.com>
32823 PR tree-optimization/79800
32824 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32825 precision in negative-positive range.
32826 (format_floating): Call non-const overload with adjusted precision.
32828 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32831 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32834 2017-03-14 Martin Sebor <msebor@redhat.com>
32836 PR middle-end/80020
32837 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32838 * builtins.def (aligned_alloc): Use it.
32841 * Makefile.in (GTFILES): Add calls.c.
32842 * calls.c: Include "gt-calls.h".
32844 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32846 PR rtl-optimization/79728
32847 * regs.h (struct target_regs): New field
32848 x_contains_allocatable_regs_of_mode.
32849 (contains_allocatable_regs_of_mode): New macro.
32850 * reginfo.c (init_reg_sets_1): Initialize it, and change
32851 contains_reg_of_mode so it includes global regs as well.
32852 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32853 rather than contains_regs_of_mode.
32855 2017-03-14 Martin Liska <mliska@suse.cz>
32857 * doc/invoke.texi: Document options that can't be combined with
32858 -fcheck-pointer-bounds.
32860 2017-03-14 Martin Liska <mliska@suse.cz>
32862 PR middle-end/79831
32863 * doc/invoke.texi (-Wchkp): Document the option.
32865 2017-03-14 Martin Liska <mliska@suse.cz>
32867 * Makefile.in: Install gcov-dump.
32869 2017-03-14 Martin Liska <mliska@suse.cz>
32871 * multiple_target.c (expand_target_clones): Bail out for
32872 an invalid attribute.
32874 2017-03-14 Richard Biener <rguenther@suse.de>
32876 * alias.c (struct alias_set_entry): Pack properly.
32877 * cfgloop.h (struct loop): Likewise.
32878 * cse.c (struct set): Likewise.
32879 * ipa-utils.c (struct searchc_env): Likewise.
32880 * loop-invariant.c (struct invariant): Likewise.
32881 * lra-remat.c (struct cand): Likewise.
32882 * recog.c (struct change_t): Likewise.
32883 * rtl.h (struct address_info): Likewise.
32884 * symbol-summary.h (function_summary): Likewise.
32885 * tree-loop-distribution.c (struct partition): Likewise.
32886 * tree-object-size.c (struct object_size_info): Likewise.
32887 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32888 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32889 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32890 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32891 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32892 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32893 (struct _stmt_vec_info): Likewise.
32895 2017-03-14 Martin Liska <mliska@suse.cz>
32898 * multiple_target.c (create_dispatcher_calls): Check that
32899 a target can create a function dispatcher.
32901 2017-03-14 Martin Liska <mliska@suse.cz>
32904 * multiple_target.c (expand_target_clones): Drop local.local
32905 flag for default implementation.
32907 2017-03-14 Richard Biener <rguenther@suse.de>
32909 PR tree-optimization/80030
32910 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32912 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32914 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32915 gcc_fallthrough() instead of __attribute__((fallthrough));
32917 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32919 * doc/gcc.texi: Remove "up" link to (DIR).
32920 * doc/gccint.texi: Ditto.
32922 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32924 * doc/install.texi (Specific) <avr>: Remove reference to
32927 2017-03-13 Jeff Law <law@redhat.com>
32929 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32930 attribute rather than comments.
32932 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32933 match_scratch operand is highest.
32935 2017-03-13 Martin Liska <mliska@suse.cz>
32937 PR middle-end/78339
32938 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32939 is a CHKP clone, use original declaration.
32941 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32943 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32944 (arc_conditional_register_usage): Use a different allocation order
32945 when optimizing for size.
32946 * common/config/arc/arc-common.c (arc_option_optimization_table):
32947 Section anchors default on when optimizing for size.
32949 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32951 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32953 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32955 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32956 * config/arc/arc.md (cpu_facility): Add cd variant.
32957 (*movqi_insn): Add code density variant.
32958 (*movhi_insn): Likewise.
32959 (*movqi_insn): Likewise.
32960 (*addsi3_mixed): Likewise.
32961 (subsi3_insn): Likewise.
32963 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32965 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32967 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32969 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32970 expressions with MINUS and UNARY ops.
32972 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32975 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32977 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32978 between vec_select and vector argument.
32979 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32980 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32981 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32982 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32983 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32984 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32986 2017-03-13 Richard Biener <rguenther@suse.de>
32989 * params.def (vect-max-peeling-for-alignment): Fix typo.
32991 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32993 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32994 issue that only occurred with binutils below 2.18.
32996 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32998 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32999 refer to binutils 2.11/2.12 minimum.
33001 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33003 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33004 ftp.kernel.org and simplify binutils requirement.
33006 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33008 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33010 (Optimize Options): Ditto. Also remove redundancy.
33012 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33014 PR translation/79848
33015 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33017 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33018 to G_ to avoid double translation.
33020 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33022 PR translation/79923
33023 * auto-profile.c (get_combined_location): Convert leading
33024 character of diagnostics to lower case and remove trailing period.
33025 (read_profile): Likewise for various diagnostics.
33026 * config/arm/arm.c (arm_option_override): Remove trailing period
33027 from various diagnostics.
33028 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33029 (msp430_expand_delay_cycles): Likewise.
33031 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33034 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33035 full command-line argument, rather than just "str".
33036 (aarch64_validate_march): Likewise.
33037 (aarch64_validate_mtune): Likewise.
33039 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33041 PR rtl-optimization/78911
33042 * lra-assigns.c (must_not_spill_p): New function.
33043 (spill_for): Use it.
33045 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33047 PR tree-optimization/79981
33048 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33049 ATOMIC_COMPARE_EXCHANGE ifn result.
33050 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33051 IFN_ATOMIC_COMPARE_EXCHANGE.
33053 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33056 * opts.c (parse_sanitizer_options): Add missing question mark to
33057 "did you mean" message.
33059 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33061 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33063 (VMULEUH_UNS): Likewise.
33064 (VMULOUB_UNS): Likewise.
33065 (VMULOUH_UNS): Likewise.
33066 * config/rs6000/rs6000.c (builtin_function_type): Remove
33067 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33069 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33072 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33073 x with result of extra_parsing_for_operand_code_0.
33074 (function_reader::extra_parsing_for_operand_code_0): Convert
33075 return type from void to rtx, returning x. When reading
33076 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33077 larger size containing struct block_symbol.
33079 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33081 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33082 -mfloat128-hardware without -m64.
33084 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33087 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33088 entries to the case statement that marks unsigned arguments to
33089 overloaded functions.
33091 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33093 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33094 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33096 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33099 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33100 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33102 2017-03-10 Martin Liska <mliska@suse.cz>
33106 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33107 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33110 2017-03-10 Olivier Hainque <hainque@adacore.com>
33112 * tree-switch-conversion (array_value_type): Start by resetting
33113 candidate type to it's main variant.
33115 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33117 PR rtl-optimization/79909
33118 * combine.c (try_combine): Use simplify_replace_rtx on individual
33119 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33120 of the whole CALL_INSN_FUNCTION_USAGE.
33122 PR tree-optimization/79972
33123 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33124 get_range_info on SSA_NAMEs. Formatting fixes.
33126 2017-03-10 Richard Biener <rguenther@suse.de>
33127 Jakub Jelinek <jakub@redhat.com>
33129 PR tree-optimization/77975
33130 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33131 edge to be constant.
33132 (get_val_for): For constant x return it. Formatting fix.
33133 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33134 would use the same bases as the current one.
33136 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33138 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33139 instead of vec_select for V1TImode.
33140 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33142 (VSX_LE_128): Add V1TI to this mode iterator.
33143 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33144 (*vsx_le_perm_store_<mode>): Likewise.
33145 (pre-reload splitter for VSX stores): Likewise.
33146 (post-reload splitter for VSX stores): Likewise.
33147 (*vsx_xxpermdi2_le_<mode>): Likewise.
33148 (*vsx_lxvd2x2_le_<mode>): Likewise.
33149 (*vsx_stxvd2x2_le_<mode>): Likewise.
33151 2017-03-09 Michael Eager <eager@eagercon.com>
33153 Correct failures with --enable-checking=yes,rtl.
33155 * config/microblaze/microblaze.c (microblaze_expand_shift):
33156 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33157 test for const0_rtx.
33158 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33159 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33161 2017-03-09 Richard Biener <rguenther@suse.de>
33163 PR tree-optimization/79977
33164 * graphite-scop-detection.c (scop_detection::merge_sese):
33165 Handle the case of extra exits to blocks dominating the entry.
33167 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33169 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33172 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33174 PR rtl-optimization/79949
33175 * lra-constraints.c (process_alt_operands): Check memory when
33176 trying to predict a cycle. Print about the overall increase.
33178 2017-03-09 Richard Biener <rguenther@suse.de>
33180 PR middle-end/79971
33181 * gimple-expr.c (useless_type_conversion_p): Preserve
33182 TYPE_SATURATING for fixed-point types.
33184 2017-03-09 Richard Biener <rguenther@suse.de>
33187 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33188 alignment of BLKmode params.
33190 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33193 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33194 (VALL_NO_V2Q): Likewise.
33196 * config/aarch64/aarch64-simd.md
33197 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33199 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33200 VALL_NO_V2Q mode iterator.
33201 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33203 2017-03-09 Martin Liska <mliska@suse.cz>
33205 PR tree-optimization/79631
33206 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33207 tree_int_cst_sign_bit just for INTEGER constants.
33209 2017-03-09 Martin Liska <mliska@suse.cz>
33213 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33216 2017-03-09 Marek Polacek <polacek@redhat.com>
33219 * tree.c (inchash::add_expr): Handle TREE_VEC.
33221 2017-03-09 Martin Liska <mliska@suse.cz>
33224 (chkp_narrow_size_and_offset): New function.
33225 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33226 (void chkp_parse_bit_field_ref): New function.
33227 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33228 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33230 2017-03-09 Martin Liska <mliska@suse.cz>
33233 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33234 (chkp_find_bounds_1): Remove gcc_unreachable.
33236 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33239 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33240 BUILT_IN_SYNC*, determine the access type from the size suffix and
33241 always build a MEM_REF with that type. Handle forgotten
33242 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33245 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33246 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33247 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33248 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33249 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33250 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33251 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33252 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33253 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33254 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33255 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33256 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33257 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33258 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33259 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33260 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33261 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33262 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33263 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33264 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33265 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33266 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33267 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33268 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33269 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33270 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33271 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33272 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33273 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33274 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33275 definitions outside of __OPTIMIZE__ guarded section.
33278 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33279 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33280 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33281 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33284 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33286 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33287 ("vfenez<mode>"): Add missing constraints.
33289 2017-03-08 Martin Sebor <msebor@redhat.com>
33292 * config/nds32/nds32.c (nds32_option_override):
33293 Fix misspelled diagnostic.
33295 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33298 * gimplify.c (gimplify_omp_for): Replace index var in outer
33299 taskloop statement with an artificial variable and add
33300 OMP_CLAUSE_PRIVATE clause for it.
33302 2017-03-08 Richard Biener <rguenther@suse.de>
33304 PR tree-optimization/79955
33305 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33306 for accesses that are completely outside of the variable.
33308 2017-03-08 Andrew Haley <aph@redhat.com>
33310 PR tree-optimization/79943
33311 * tree-ssa-loop-split.c (compute_new_first_bound): When
33312 calculating the new upper bound, (END-BEG) should be added, not
33315 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33317 * config/avr/avr.md (setmemhi): Make sure match_dup
33318 operand number comes before match_scratch.
33320 2017-03-08 Richard Biener <rguenther@suse.de>
33322 PR tree-optimization/79920
33323 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33324 with ncopies == 1 to ...
33325 (vect_transform_slp_perm_load): ... here. Properly compute
33326 all element loads by iterating VF times over the group. Do
33327 not handle ncopies (computed in a broken way) in
33328 vect_create_mask_and_perm.
33330 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33333 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33334 is a uniform vector, use uniform_vector_p return value instead of
33335 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33337 2017-03-07 Marek Polacek <polacek@redhat.com>
33339 PR middle-end/79809
33340 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33341 (alloca_call_type): Likewise.
33343 2017-03-07 Martin Liska <mliska@suse.cz>
33345 * gcov.c (process_args): Put comment to correct location.
33347 2017-03-07 Martin Liska <mliska@suse.cz>
33349 PR middle-end/68270
33350 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33351 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33352 (chkp_narrow_bounds_for_field): Likewise.
33353 (chkp_parse_array_and_component_ref): Pass one more argument to
33356 2017-03-07 Richard Biener <rguenther@suse.de>
33358 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33361 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33363 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33364 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33366 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33369 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33370 to end of description.
33371 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33373 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33375 PR rtl-optimization/79901
33376 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33378 (*avx512f_<code><mode>3<mask_name>): ... this.
33379 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33380 iterator instead of VI8_AVX2_AVX512BW.
33382 PR rtl-optimization/79901
33383 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33384 min/max expander, expand it using expand_vec_cond_expr.
33387 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33390 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33393 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33394 to void * for PCH reasons.
33395 * dwarf2out.c (output_loc_operands, output_die): Cast
33396 v.val_vec.array to unsigned char *.
33398 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33401 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33404 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33406 PR rtl-optimization/79571
33407 * lra-constraints.c (process_alt_operands): Calculate static
33408 reject and subtract it from overall when only addresses will be
33411 2017-03-06 Julia Koval <julia.koval@intel.com>
33414 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33415 incoming stack boundary to 128 for 64-bit targets.
33417 2017-03-06 Richard Biener <rguenther@suse.de>
33419 PR tree-optimization/79894
33420 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33421 to NULL after folding it.
33423 2017-03-06 Richard Biener <rguenther@suse.de>
33425 PR tree-optimization/79824
33426 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33427 check disabling peeling for gaps.
33429 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33431 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33432 attributes): Document gettimeofday.
33434 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33436 * config/s390/s390.c (s390_option_override_internal): Set
33437 PARAM_MIN_VECT_LOOP_BOUND
33439 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33441 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33442 * config/s390/s390.md: Likewise.
33444 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33447 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33448 (<avx2_avx512>_perm<mode>): Rename to ...
33449 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33451 (<avx512>_perm<mode>_mask): Rename to ...
33452 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33454 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33455 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33456 instead of VI8F_256_512.
33457 (avx512f_perm<mode>): New define_expand.
33458 (avx512f_perm<mode>_mask): Likewise.
33459 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33460 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33462 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33464 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33465 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33467 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33469 2017-03-06 Martin Liska <mliska@suse.cz>
33472 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33473 when having a SSA NAME w/o VAR_DECL assigned to it.
33475 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33477 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33478 msa_dpsub_<su>_d): Fix MODE for vec_select.
33480 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33482 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33484 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33486 2017-03-06 Richard Biener <rguenther@suse.de>
33488 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33489 * plugin.c (register_plugin_info): Likewise.
33490 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33492 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33494 * config/i386/sse.md (sse_storehps, sse_storelps,
33495 avx_<castmode><avxsizesuffix>_<castmode>,
33496 avx512f_<castmode><avxsizesuffix>_<castmode>,
33497 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33498 in condition that at least one operand is not a MEM.
33500 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33502 PR middle-end/79805
33503 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33504 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33506 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33507 gimple_call_nothrow_p flag based on whether original builtin can throw.
33508 If it can, emit following stmts on the fallthrough edge.
33509 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33510 don't create new bb if inserting just debug stmts on the edge, try to
33511 insert them on the fallthru bb or just reset debug stmts.
33513 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33516 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33517 restore recog_data (including the operand rtxes inside it) around
33518 the call to get_insn_template.
33520 2017-03-03 Martin Sebor <msebor@redhat.com>
33522 PR tree-optimization/79699
33523 * context.c (context::~context): Free MPFR caches to avoid
33524 a memory leak on program exit.
33526 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33528 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33529 Use wide_int::ulow () instead of .elt (0).
33531 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33533 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33534 (*pushxf): Limit oF constraint to 32bit targets and add oC
33535 constraint for 64bit targets.
33536 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33537 (*pushdf): Change rmF constraint to rmC.
33539 2017-03-03 Martin Liska <mliska@suse.cz>
33541 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33542 Remove unused variable.
33544 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33547 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33548 is a memory operand, increase num_memory.
33549 (ix86_expand_args_builtin): Likewise.
33551 2017-03-03 Jan Hubicka <jh@suse.cz>
33554 * ipa-devirt.c (maybe_record_node): Properly handle
33555 __cxa_pure_virtual visibility.
33557 2017-03-03 Martin Liska <mliska@suse.cz>
33559 PR tree-optimization/79803
33560 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33562 (pass_loop_prefetch::execute): Disabled optimization if an
33563 assumption about L1 cache size is not met.
33565 2017-03-03 Martin Liska <mliska@suse.cz>
33567 PR rtl-optimization/79574
33568 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33569 (hash_scan_set): Likewise.
33570 (dump_hash_table): Likewise.
33571 (hoist_code): Likewise.
33573 2017-03-03 Richard Biener <rguenther@suse.de>
33575 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33576 in place of uhigh ().
33577 (fixed_convert_from_real): Likewise.
33579 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33582 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33584 2017-03-03 Richard Biener <rguenther@suse.de>
33586 PR middle-end/79818
33587 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33588 TYPE_OVERFLOW_UNDEFINED check.
33590 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33592 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33594 (vector_ae_<mode>_p): Likewise.
33595 (vector_nez_<mode>_p): Likewise.
33596 (vector_ne_v2di_p): Likewise.
33597 (vector_ae_v2di_p): Likewise.
33598 (vector_ne_<mode>_p): Likewise.
33599 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33601 (vsx_tsqrt<mode>2_fe): Likewise.
33603 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33606 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33608 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33610 PR rtl-optimization/79780
33611 * cprop.c (one_cprop_pass): When second and further conditional trap
33612 in a single basic block is turned into an unconditional trap, turn it
33613 into a deleted note to avoid RTL verification failures.
33615 2017-03-02 Richard Biener <rguenther@suse.de>
33617 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33619 2017-03-02 Richard Biener <rguenther@suse.de>
33621 PR tree-optimization/79345
33623 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33624 param and abort the walk, returning -1 if it is hit.
33625 (walk_aliased_vdefs): Take a limit param and pass it on.
33626 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33627 defaulting to 0 and return a signed int.
33628 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33629 (check_defs): New helper.
33630 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33631 about uninitialized memory.
33632 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33633 bogus uninitialized warning.
33634 (fixed_convert_from_real): Likewise.
33636 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33638 PR tree-optimization/66768
33639 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33640 iv_use if base object can't be determined.
33642 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33644 PR tree-optimization/79345
33645 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33646 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33647 (get_pattern_stats): Initialize it.
33648 * genemit.c (gen_expand): Verify match_scratch numbers come after
33649 match_operand/match_dup numbers.
33650 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33651 match_scratch numbers.
33652 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33654 * config/s390/s390.md (trunctdsd2): Likewise.
33656 2017-03-02 Richard Biener <rguenther@suse.de>
33658 * wide-int.h (wide_int_storage::operator=): Implement in terms
33661 2017-03-02 Richard Biener <rguenther@suse.de>
33663 PR tree-optimization/79777
33664 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33665 the to insert expression to sth existing.
33667 2017-03-01 Martin Sebor <msebor@redhat.com>
33669 PR middle-end/79692
33670 * gimple-ssa-sprintf.c
33671 (directive::known_width_and_precision): New function.
33672 (format_integer): Use it.
33673 (get_mpfr_format_length): Consider the full range of precision
33674 when computing %g output with the # flag. Set the likely byte
33675 count to 3 rather than 1 when precision is indeterminate.
33676 (format_floating): Correct the lower bound of precision.
33678 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33680 * doc/invoke.texi: Document default code model for 64-bit Linux.
33682 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33685 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33686 udiv rather than div since input pattern is unsigned.
33688 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33690 * config/i386/i386.c (print_reg): Warn for values of
33691 unsupported size in integer register.
33693 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33696 * config/rs6000/predicates.md (current_file_function_operand): Do
33697 not allow self calls to be local if the function is replaceable.
33699 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33702 * config/rs6000/altivec.h (vec_ctz and others): Change the
33703 preprocessor macro that controls conditional compilation from
33704 _ARCH_PWR9 to __POWER9_VECTOR__.
33705 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33706 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33707 control (instead of _ARCH_PWR9 control) so that template
33708 definition uses power9-specific function.
33709 (vec_any_eq): Likewise.
33710 (vec_all_ne): Change macro definition to use a power9-specific
33711 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33712 _ARCH_PWR9 control).
33713 (vec_any_eq) Likewise.
33714 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33715 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33716 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33717 support for xvcmpnedp instruction.
33718 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33719 macro expansion so that Power9 implementation of vec_all_ne does
33720 not use the AltiVec predicate framework.
33721 (VCMPNEH_P): Likewise.
33722 (VCMPNEW_P): Likewise.
33723 (VCMPNED_P): Likewise.
33724 (VCMPNEFP_P): Likewise.
33725 (VCMPNEDP_P): Likewise.
33726 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33727 implementation of vec_any_eq to not use AltiVec predicate
33729 (VCMPAEH_P): Likewise.
33730 (VCMPAEW_P): Likewise.
33731 (VCMPAED_P): Likewise.
33732 (VCMPAEFP_P): Likewise.
33733 (VCMPAEDP_P): Likewise.
33734 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33735 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33736 not use the AltiVec predicate framework.
33737 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33738 of vec_any_eq to not use AltiVec predicate framework.
33739 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33740 support for predefined __POWER9_VECTOR__ macro to indicate that
33741 Power9 instruction selection is enabled.
33742 (altivec_overloaded_builtins): Remove extraneous
33743 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33744 function argument types RS6000_BTI_bool_V16QI and
33745 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33746 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33747 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33748 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33749 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33750 Power9 for implementations of vec_cmpne. Change the signature for
33751 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33752 (representing vec_all_ne) to remove the previously described first
33753 argument of type RS6000_BTI_INTSI, as this was an artifact of
33754 reliance on the AltiVec predicate framework, which is no longer
33755 used in the implementation of these functions. Add
33756 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33757 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33758 since, unlike the AltiVec predicate framework implementation, we
33759 do not share function descriptors between vec_alle and vec_anyeq.
33760 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33761 set of modes that receive special treatment even when
33762 TARGET_P9_VECTOR is true. The special treatment emits code that
33763 does not depend on Power9 instructions.
33764 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33765 define_expand to not rely on AltiVec predicate framework.
33766 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33768 (vector_ne_v2di_p): Change this define_expand to not rely on
33769 AltiVec predicate framework.
33770 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33772 (vector_ne_<mode>_p): Change this define_expand to not rely on
33773 AltiVec predicate framework.
33774 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33776 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33777 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33778 define_insn pattern.
33779 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33780 define_insn pattern because the xvcmpne<VSs>. instruction is not
33782 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33783 instruction is not supported.
33785 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33787 * config/nvptx/nvptx.c: Include intl.h.
33789 2017-03-01 Martin Jambor <mjambor@suse.cz>
33792 * ipa-prop.h (ipa_bits): Removed field known.
33793 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33794 to pointers. Adjusted their comments to warn about their sharing.
33795 (ipcp_transformation_summary): Change bits to a vector of pointers.
33796 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33797 (ipa_get_ipa_bits_for_value): Declare.
33798 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33799 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33800 (ipa_bits_hash_table): Likewise.
33801 (ipa_vr_ggc_hash_traits): Likewise.
33802 (ipa_vr_hash_table): Likewise.
33803 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33804 being pointers and vr_known being removed.
33805 (ipa_set_jf_unknown): Likewise.
33806 (ipa_get_ipa_bits_for_value): New function.
33807 (ipa_set_jfunc_bits): Likewise.
33808 (ipa_get_value_range): New overloaded functions.
33809 (ipa_set_jfunc_vr): Likewise.
33810 (ipa_compute_jump_functions_for_edge): Use the above functions to
33811 construct bits and vr parts of jump functions.
33812 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33813 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33815 (ipcp_grow_transformations_if_necessary): Also allocate
33816 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33818 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33819 them. Fix too long lines.
33820 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33821 vr_known being removed.
33822 (ipa_read_jump_function): Use new setter functions to construct bits
33823 and vr parts of jump functions or set them to NULL.
33824 (write_ipcp_transformation_info): Adjust for bits being pointers.
33825 (read_ipcp_transformation_info): Likewise.
33826 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33828 Include gt-ipa-prop.h.
33829 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33831 (ipcp_store_bits_results): Likewise.
33832 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33833 Do not write to existing jump functions but use a temporary instead.
33835 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33838 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33839 attempt to use its first operand as BIT_FIELD_REF base.
33841 2017-03-01 Richard Biener <rguenther@suse.de>
33843 PR middle-end/79721
33844 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33845 interpolating formula in wrapping arithmetic.
33846 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33848 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33850 PR tree-optimization/79734
33851 * tree-vect-generic.c (expand_vector_condition): Optimize
33852 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33853 Handle VEC_COND_EXPR where comparison has different inner width from
33854 type's inner width.
33856 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33858 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33859 markup, and similar issues. Remove @opindex entries for things
33860 that aren't options. Add missing -mmpy-option entries.
33862 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33864 PR tree-optimization/79737
33865 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33866 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33867 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33868 instead of byte_size. Formatting fix.
33869 (shift_bytes_in_array_right): Formatting fix.
33871 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33874 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33875 condition on optimize for the leaf function test.
33877 2017-02-28 Martin Liska <mliska@suse.cz>
33880 * read-rtl-function.c (function_reader::handle_unknown_directive):
33881 Bail out when one uses -flto.
33883 2017-02-28 Martin Liska <mliska@suse.cz>
33885 * common.opt: Replace space with tabular for options of <number>
33887 * config/i386/i386.opt: Show <number> value for
33888 -mlarge-data-threshold.
33889 * opts.c (print_filtered_help): Do not display number in hexadecimal
33892 2017-02-28 Martin Liska <mliska@suse.cz>
33894 * common.opt: Fix --help=option -Q for options which are of
33897 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33899 * config/i386/i386.c (print_reg): Error out for values
33900 of 8-bit size in invalid integer register.
33902 2017-02-28 Martin Sebor <msebor@redhat.com>
33904 PR tree-optimization/79691
33905 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33907 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33910 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33911 gcc_unreachable with output_operand_lossage.
33913 2017-02-28 Richard Biener <rguenther@suse.de>
33915 PR tree-optimization/79740
33916 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33918 (visit_nary_op): Insert the nary into the hashtable if we
33919 pattern-matched sth.
33920 * tree-ssa-pre.c (eliminate_insert): Robustify.
33922 2017-02-28 Richard Biener <rguenther@suse.de>
33924 PR middle-end/79731
33925 * fold-const.c (decode_field_reference): Reject out-of-bound
33928 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33930 * config/i386/i386.c: Include intl.h.
33931 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33932 instead of just cond ? "..." : "...".
33933 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33934 * coverage.c (read_counts_file): Likewise.
33935 * omp-offload.c: Include intl.h.
33936 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33937 of just cond ? "..." : "...".
33938 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33939 of just cond ? "..." : "...".
33941 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33944 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33946 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33947 'tune for' CPU name.
33948 * config/arm/arm-cpu-data.h: Regenerated.
33950 2017-02-28 Richard Biener <rguenther@suse.de>
33952 PR tree-optimization/79732
33953 * tree-inline.c (expand_call_inline): Do not shadow var.
33955 2017-02-28 Richard Biener <rguenther@suse.de>
33957 PR tree-optimization/79723
33958 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33959 address-space properly.
33961 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33963 * doc/optinfo.texi (Optimization groups): Fix option used for
33965 * doc/invoke.texi (-fopt-info): Document "omp".
33966 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33967 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33968 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33969 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33970 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33972 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33974 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33975 OPTGROUP_OMP with "omp".
33977 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33980 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33981 for arithmetic shift of unsigned V2DI.
33983 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33985 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33986 arc/linux.h headers.
33987 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33988 (LINK_SPEC): Likewise.
33989 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33990 (EXTRA_SPECS): Likewise.
33991 (STARTFILE_SPEC): Likewise.
33992 (ENDFILE_SPEC): Likewise.
33993 (LIB_SPEC): Likewise.
33994 (TARGET_SDATA_DEFAULT): Likewise.
33995 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33996 (MULTILIB_DEFAULTS): Likewise.
33997 (DWARF2_UNWIND_INFO): Likewise.
33998 * config/arc/big.h: New file.
33999 * config/arc/elf.h: Likewise.
34000 * config/arc/linux.h: Likewise.
34001 * config/arc/t-uClibc: Remove.
34003 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34005 PR tree-optimization/77536
34006 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34007 (tree_transform_and_unroll_loop): Use above function to compute the
34008 estimated niter of unrolled loop and use it when scaling profile.
34009 Also use count info rather than frequency if it's non-zero.
34010 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34011 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34012 (vect_transform_loop): Call above function.
34014 2017-02-27 Richard Biener <rguenther@suse.de>
34016 PR tree-optimization/45397
34017 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34018 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34019 (visit_nary_op): Add pattern matching for CSEing sign-changed
34020 or truncated operations with wider ones.
34022 2017-02-27 Richard Biener <rguenther@suse.de>
34024 PR tree-optimization/79690
34025 * tree-vect-stmts.c (vectorizable_store): Use vector type
34026 built from the DR with address-space.
34028 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34030 * doc/invoke.texi (Optimize Options): Refine the description
34031 of asan-use-after-return.
34033 2017-02-25 Alan Modra <amodra@gmail.com>
34035 PR rtl-optimization/79584
34036 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34037 base, not ad->base_term, the reg within base. Remove assertion
34038 that ad->base == ad->base_term. Replace gen_int_mode using
34039 bogus mode with const0_rtx.
34041 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34043 PR middle-end/79396
34044 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34045 FMA_EXPR like tcc_binary or tcc_unary.
34047 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34050 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34052 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34053 (output_loc_operands): Handle DW_OP_call_ref and
34054 DW_OP_GNU_variable_value.
34055 (struct variable_value_struct): New type.
34056 (struct variable_value_hasher): Likewise.
34057 (variable_value_hash): New variable.
34058 (string_types): Remove.
34059 (copy_loc_descr): New function.
34060 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34061 (prepend_loc_descr_to_each): New function.
34062 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34063 instead of add_loc_descr_to_each if the first argument is single
34064 location list and the second has multiple.
34065 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34066 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34067 when looking for variable value which doesn't have other location info.
34068 (loc_list_from_tree): Formatting fix.
34069 (gen_array_type_die): Simplify DW_AT_string_length handling.
34070 (adjust_string_types): Remove.
34071 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34072 string_types. Call resolve_variable_values.
34073 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34074 (resolve_addr_in_expr): Likewise. Add A argument.
34075 (copy_deref_exprloc): Remove deref argument. Adjust for the
34076 original expression being DW_OP_GNU_variable_value with optionally
34077 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34078 optionally after it.
34079 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34080 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34081 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34082 (variable_value_hasher::hash, variable_value_hasher::equal): New
34084 (resolve_variable_value_in_expr, resolve_variable_value,
34085 resolve_variable_values, note_variable_value_in_expr,
34086 note_variable_value): New functions.
34087 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34090 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34093 * opts.h (handle_generated_option): Add GENERATED_P argument.
34094 * opts-common.c (handle_option): Adjust function comment.
34095 (handle_generated_option): Add GENERATED_P argument, pass it to
34097 (control_warning_option): Pass false to handle_generated_option
34099 * opts.c (maybe_default_option): Pass true to handle_generated_option
34101 * optc-gen.awk: Likewise.
34103 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34105 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34106 a REG, look at the REG it is a SUBREG of.
34107 (splitter for cmpeqsi_t): Ditto.
34109 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34111 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34112 the special USEs with the pattern of the insn, not the insn itself.
34114 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34117 * doc/invoke.texi: Document -mload-store-pairs.
34119 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34120 Sandra Loosemore <sandra@codesourcery.com>
34122 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34123 argument isn't a CONST_INT.
34124 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34125 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34126 (nios2_validate_compare): Bypass alternate compare logic if *op2
34127 is not a CONST_INT.
34128 (ldstwm_operation_p): Return false if first_base is not a REG or
34129 if first_offset is not a CONST_INT.
34131 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34133 * config/cris/cris.md: Use correct operand in a define_peephole2.
34135 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34137 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34139 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34141 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34142 this_insn if it is an INSN or JUMP_INSN.
34143 (force_offsettable): Look at base, not at addr.
34144 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34145 on things that aren't necessarily CONST_INTs.
34147 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34149 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34150 -mfpmath=sse is the default also for x86-32 targets with SSE2
34151 instruction set when @option{-ffast-math} is enabled
34153 2017-02-24 Jeff Law <law@redhat.com>
34155 PR rtl-optimizatoin/79286
34156 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34159 2017-02-24 Richard Biener <rguenther@suse.de>
34161 PR tree-optimization/79389
34162 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34165 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34167 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34168 function comment to reflect reality.
34169 (loop_exits_before_overflow): Fix typo in function description.
34171 2017-02-24 Richard Biener <rguenther@suse.de>
34173 PR tree-optimization/79389
34174 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34175 properly that a threading opportunity exists. Detect conditional
34176 copy/constant propagation opportunities.
34178 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34180 * config/visium/visium.md (type): Add trap.
34181 (b): New mode attribute.
34182 (*btst): Rename into...
34183 (*btst<mode>): ...this and adjust.
34184 (*cbranchsi4_btst_insn): Rename into...
34185 (*cbranch<mode>4_btst_insn): ...this and adjust.
34186 (trap): New define_insn.
34188 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34190 PR tree-optimization/79389
34191 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34192 (noce_reversed_cond_code): New function.
34193 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34194 reversed_comparison_code. Formatting fix.
34195 (noce_try_store_flag): Test rev_cond != NULL in addition to
34196 reversed_comparison_code.
34197 (noce_try_store_flag_constants): Likewise.
34198 (noce_try_store_flag_mask): Likewise.
34199 (noce_try_addcc): Use rev_cond if non-NULL instead of
34200 reversed_comparison_code.
34201 (noce_try_cmove_arith): Likewise. Formatting fixes.
34202 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34203 (noce_find_if_block): Initialize rev_cond.
34204 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34205 instead of false as last argument never attempt to reverse it
34208 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34210 PR tree-optimization/79663
34211 * tree-predcom.c (combine_chains): Process refs in reverse order
34212 only for ZERO length chains, and add explaining comment.
34214 2017-02-23 Jeff Law <law@redhat.com>
34216 PR tree-optimization/79578
34217 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34218 in call to operand_equal_p.
34220 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34223 * config/i386/cpuid.h: Fix another undefined behavior.
34225 2017-02-23 Richard Biener <rguenther@suse.de>
34227 PR tree-optimization/79683
34228 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34229 vector types for data-refs.
34231 2017-02-23 Martin Liska <mliska@suse.cz>
34233 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34235 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34237 PR middle-end/79665
34238 * internal-fn.c (get_range_pos_neg): Moved to ...
34239 * tree.c (get_range_pos_neg): ... here. No longer static.
34240 * tree.h (get_range_pos_neg): New prototype.
34241 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34242 are known to be in between 0 and signed maximum inclusive, try to
34243 expand both unsigned and signed divmod and use the cheaper one from
34246 2017-02-22 Jeff Law <law@redhat.com>
34248 PR tree-optimization/79578
34249 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34250 to compare base operands.
34252 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34255 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34256 gpc_reg_operand instead of fpr_reg_operand.
34258 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34260 * config/mips/mips.c (mips_return_in_memory): Force FP
34261 vector types to be returned in memory for o32 ABI.
34263 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34265 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34266 instead of DW_TAG_member for static data member declarations and don't
34267 set no_linkage_name for static inline data members.
34268 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34271 2017-02-22 Martin Liska <mliska@suse.cz>
34273 * doc/invoke.texi: Replace inequality signs with square brackets
34276 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34279 * lra-constraints.c (simplify_operand_subreg): Handle
34280 WORD_REGISTER_OPERATIONS targets.
34282 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34285 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34286 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34287 elimination by swapping fld*.
34289 2017-02-22 Richard Biener <rguenther@suse.de>
34291 PR tree-optimization/79673
34292 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34293 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34294 irrelevant address-space qualifiers and avoiding a
34295 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34297 2017-02-22 Richard Biener <rguenther@suse.de>
34299 PR tree-optimization/79666
34300 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34301 to not symbolically negate if that may introduce undefined
34304 2017-02-22 Martin Liska <mliska@suse.cz>
34307 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34308 * data-streamer-out.c (streamer_write_gcov_count_stream):
34310 * value-prof.c (stream_out_histogram_value): Make assert more
34311 precise based on type of counter.
34313 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34316 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34317 Use nonimmediate_operand instead of memory_operand for operand 1.
34318 (float-extend standard_x87sse_constant_load splitter): Ditto.
34320 2017-02-21 Jeff Law <law@redhat.com>
34322 PR tree-optimization/79621
34323 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34324 blocks with edges to themselves.
34326 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34329 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34330 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34331 Use gimple_call_builtin_p.
34334 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34335 on temporarily removed DEBUG_INSNs.
34337 PR tree-optimization/79649
34338 * tree-loop-distribution.c (classify_partition): Give up on
34339 non-generic address space loads/stores.
34341 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34343 * doc/loop.texi (Loop manipulation): Remove nonexistent
34344 tree_ssa_loop_version from the documentation.
34345 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34347 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34350 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34351 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34352 * config/rs6000/rs6000.c: Include except.h.
34353 (rs6000_expand_split_stack_prologue): Call
34354 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34356 2017-02-21 Martin Jambor <mjambor@suse.cz>
34359 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34360 have been analyzed.
34362 2017-02-21 Martin Jambor <mjambor@suse.cz>
34364 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34365 for backward compatibility only.
34366 * doc/invoke.texi (Option Summary): Remove all references to
34367 -fipa-cp-alignment.
34369 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34373 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34375 * lra-constraints.c (curr_insn_transform): Handle
34376 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34378 2017-02-21 Martin Liska <mliska@suse.cz>
34380 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34382 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34384 PR translation/79638
34385 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34387 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34390 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34391 (arm_function_ok_for_sibcall): Return false for an indirect call by
34392 descriptor if all the argument registers are used.
34393 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34394 alignment of the function.
34396 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34398 PR tree-optimization/61441
34399 * simplify-rtx.c (simplify_const_unary_operation): For
34400 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34401 the sNaN unmodified.
34403 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34405 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34406 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34407 instead of SYSTEM_HEADER_DIR.
34409 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34410 Martin Liška <mliska@suse.cz>
34412 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34413 Fix typos and grammar, use active voice, and clarify.
34415 2017-02-20 Marek Polacek <polacek@redhat.com>
34417 PR middle-end/79537
34418 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34421 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34423 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34426 * config/i386/i386.c (ix86_expand_builtin): Handle
34427 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34428 ix86_builtins_isa[fcode].isa as a requirement of those
34429 flags and any other flag in the bitmask.
34430 (ix86_init_mmx_sse_builtins): Use 0 instead of
34431 ~OPTION_MASK_ISA_64BIT as mask.
34432 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34433 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34434 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34435 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34437 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34440 * lra-constraints.c (split_reg): Check requested split mode
34441 is supported by the register.
34443 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34445 * lra-constraints.c (simplify_operand_subreg): Remove early
34448 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34451 * lra-constraints.c (curr_insn_transform): Tighten condition
34452 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34454 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34457 * lra-constraints.c (curr_insn_transform): Handle
34458 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34460 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34463 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34465 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34467 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34470 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34473 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34475 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34478 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34480 * final.c (last_columnnum, override_columnnum): New variables.
34481 (final_start_function): Set last_columnnum, pass it to begin_prologue
34482 hook and pass 0 to dwarf2out_begin_prologue.
34483 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34484 to source_line debug hook.
34485 (notice_source_line): Compute last_columnnum and for debug_column_info
34486 return true on column changes.
34487 * debug.h (struct gcc_debug_hooks): Add column argument to
34488 source_line and begin_prologue hooks.
34489 (debug_nothing_int_charstar_int_bool): Remove prototype.
34490 (debug_nothing_int_int_charstar,
34491 debug_nothing_int_int_charstar_int_bool): New prototypes.
34492 (dwarf2out_begin_prologue): Add column argument.
34493 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34494 begin_prologue hooks.
34495 (debug_nothing_int_charstar_int_bool): Remove.
34496 (debug_nothing_int_int_charstar,
34497 debug_nothing_int_int_charstar_int_bool): New functions.
34498 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34499 through to dwarf2out_source_line.
34500 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34501 (dwarf2out_source_line): Add column argument, emit it if requested.
34502 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34504 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34505 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34506 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34507 through to dwarf2out_begin_prologue.
34508 (vmsdbgout_source_line): Add column argument, pass it through to
34509 dwarf2out_source_line.
34510 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34511 dbxout_source_line caller.
34512 (dbxout_source_line): Add column argument.
34514 * common.opt (gno-column-info, gcolumn-info): New options.
34515 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34516 (check_die): Also test for multiple DW_AT_decl_column attributes.
34517 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34518 DW_AT_decl_column if requested.
34519 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34521 (gen_variable_die): Likewise.
34522 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34523 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34526 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34527 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34528 (ix86_handle_option): Handle OPT_m3dnowa.
34529 * doc/invoke.texi (-m3dnowa): Document.
34530 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34531 -m3dnowa instead of -m3dnow -march=athlon.
34534 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34535 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34537 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34540 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34541 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34542 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34543 generator for vsx_xxpermdi_<mode>_be.
34544 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34545 force big-endian semantics.
34546 (vsx_xxpermdi_<mode>_be): New define_expand with same
34547 implementation as previous version of vsx_xxpermdi_<mode>.
34549 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34551 PR tree-optimization/79327
34552 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34553 variable, its initialization and use.
34555 2017-02-17 Julia Koval <julia.koval@intel.com>
34557 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34558 (OPTION_MASK_ISA_PKU_UNSET): New.
34559 (ix86_handle_option): Handle -mrdpid.
34560 * config/i386/cpuid.h (bit_RDPID): New.
34561 * config/i386/driver-i386.c (host_detect_local_cpu):
34562 Detect RDPID feature.
34563 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34564 * config/i386/i386-c.c (ix86_target_macros_internal):
34566 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34567 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34568 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34569 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34570 * config/i386/i386.md (define_insn "rdpid"): New.
34571 * config/i386/i386.opt Add -mrdpid.
34572 * config/i386/immintrin.h (_rdpid_u32): New.
34574 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34576 PR rtl-optimization/79541
34577 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34578 instead of transforming it into USE.
34580 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34582 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34583 If HONOR_SNANS (SFmode) force the input to a register.
34584 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34585 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34586 an frsp or similar insn.
34588 2017-02-17 Martin Liska <mliska@suse.cz>
34590 PR rtl-optimization/79577
34591 * params.def (selsched-max-sched-times): Increase minimum to 1.
34593 2017-02-17 Martin Liska <mliska@suse.cz>
34595 PR rtl-optimization/79574
34596 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34598 2017-02-17 Martin Liska <mliska@suse.cz>
34600 PR tree-optimization/79529
34601 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34602 ssa_defined_default_def_p to handle cases which are implicitly
34604 * tree-ssa.c (ssa_defined_default_def_p): New function.
34605 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34606 which are implicitly defined.
34607 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34609 2017-02-17 Richard Biener <rguenther@suse.de>
34611 PR middle-end/79576
34612 * params.def (max-ssa-name-query-depth): Limit to 10.
34614 2017-02-17 Richard Biener <rguenther@suse.de>
34616 PR tree-optimization/79552
34617 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34620 2017-02-17 Richard Biener <rguenther@suse.de>
34623 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34625 2017-02-17 Marek Polacek <polacek@redhat.com>
34627 PR middle-end/79536
34628 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34629 (fold_negate_expr): New wrapper.
34631 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34633 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34634 Correct terminology and de-emphasize pre-standard behavior.
34636 2017-02-16 Alan Modra <amodra@gmail.com>
34638 PR rtl-optimization/79286
34639 * ira.c (def_dominates_uses): New function.
34640 (update_equiv_regs): Don't create an equivalence for insns that
34641 may trap where the register def does not dominate the use.
34643 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34645 PR rtl-optimization/78127
34646 * lra.c (lra): Call lra_eliminate before finish the loop after
34649 2017-02-16 Richard Biener <rguenther@suse.de>
34651 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34653 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34654 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34655 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34656 (isl_val_int_from_wi): New function.
34657 (extract_affine_gmp): Rename to ...
34658 (extract_affine_wi): ... this, take a widest_int.
34659 (extract_affine_int): Just wrap extract_affine_wi.
34660 (add_param_constraints): Use isl_val_int_from_wi.
34661 (add_loop_constraints): Likewise, and extract_affine_wi.
34663 2017-02-15 Jeff Law <law@redhat.com>
34665 PR middle-end/79521
34666 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34667 ira_init_register_move_cost_if_necessary.
34669 2017-02-15 Martin Sebor <msebor@redhat.com>
34671 PR middle-end/32003
34672 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34673 removed in a prior commit.
34675 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34677 PR tree-optimization/79347
34678 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34679 counters during peeling.
34681 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34683 * Makefile.in (site.exp): Remove "set ISLVER".
34685 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34688 * real.c (real_from_integer): Call real_convert even for decimal.
34690 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34693 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34695 2017-02-14 Andrew Pinski <apinski@cavium.com>
34697 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34698 cores and change the partno/implementer to be correct.
34699 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34700 the 'B" as the implementer.
34701 * config/aarch64/aarch64-tune.md: Regenerate.
34703 2017-02-14 Carl Love <cel@us.ibm.com>
34705 * config/rs6000/rs6000.c: Add case statement entry to make the
34706 xvcvuxdsp built-in argument unsigned.
34707 * config/rs6000/vsx.md: Fix the source and return operand types so they
34708 match the instruction definitions from the ISA document. Fix typo
34709 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34712 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34715 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34716 member early_clobber_alts.
34717 * lra-lives.c (reg_early_clobber_p): New.
34718 (process_bb_lives): Use it.
34719 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34720 (debug_operand_data): Initialize early_clobber_alts.
34721 (setup_operand_alternative): Set up early_clobber_alts.
34722 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34723 alternatives to new_insn_reg.
34724 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34726 (lra_update_insn_regno_info): Pass the new arg.
34728 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34730 PR middle-end/79505
34731 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34732 (new_oacc_loop_raw): Don't clear already cleared fields.
34735 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34736 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34737 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34739 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34742 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34744 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34747 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34748 the extra instruction to the right place to store 128-bit constant
34751 2017-02-14 Martin Sebor <msebor@redhat.com>
34753 PR middle-end/79448
34754 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34755 warning for strings of unknown length.
34757 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34759 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34761 2017-02-14 Jeff Law <law@redhat.com>
34764 * ira-costs.c (scan_one_insn): Initialize register move costs
34765 for pseudos seen in USE/CLOBBER insns.
34767 PR tree-optimization/79095
34768 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34769 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34770 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34771 if the operands are known to be not equal, then the resulting range
34773 (intersect_ranges): If the new range is ~[0,0] and the old range is
34774 wide, then prefer ~[0,0].
34775 * tree-vrp.c (overflow_comparison_p_1): New function.
34776 (overflow_comparison_p): New function.
34777 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34778 if NAME is used in an overflow test.
34779 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34780 overflow check that can be expressed as an equality test, then adjust
34781 ops to be that equality test.
34783 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34785 * config/s390/s390-builtin-types.def: Remove flags argument.
34786 * config/s390/s390.c (s390_init_builtins): Likewise.
34788 2017-02-14 Martin Liska <mliska@suse.cz>
34790 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34791 vector. Fix trailing white spaces.
34793 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34795 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34798 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34800 PR rtl-optimization/68664
34801 * config/arm/arm.c (arm_sched_can_speculate_insn):
34802 New function. Declare prototype.
34803 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34805 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34807 PR rtl-optimization/68664
34808 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34810 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34812 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34814 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34815 max skip bytes for function, loop and jump.
34817 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34819 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34820 ABS_EXPR for gimple dump.
34822 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34825 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34827 PR tree-optimization/79408
34828 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34829 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34830 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34831 also if rhs1 is INTEGER_CST.
34833 2017-02-14 Richard Biener <rguenther@suse.de>
34835 PR middle-end/79432
34836 * tree-into-ssa.c (insert_phi_nodes): When the function can
34837 have abnormal edges rewrite SSA names with broken use-def
34838 dominance out of SSA and register them for PHI insertion.
34840 2017-02-13 Martin Sebor <msebor@redhat.com>
34842 PR middle-end/79496
34843 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34844 clearing info.nowrite flag when snprintf size argument is a range.
34846 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34848 * cprop.c (cprop_jump): Add missing space in string literal.
34849 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34850 (get_constraint_for_component_ref): Likewise.
34851 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34852 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34853 * lra-constraints.c (process_alt_operands): Likewise.
34854 * ipa-inline.c (inline_small_functions): Likewise.
34855 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34856 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34857 * trans-mem.c (diagnose_tm_1_op): Likewise.
34858 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34859 (grid_parallel_clauses_gridifiable): Likewise.
34861 * config/nvptx/mkoffload.c (process): Add space in between
34864 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34865 "MOD4_SSE_REGS" and "ALL_REGS".
34867 * spellcheck.c (test_data): Add , in between "foo" and "food".
34869 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34872 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34873 boundary crossing check and subsequent code generation agree.
34875 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34877 * config/aarch64/aarch64.c (has_memory_op): Delete.
34878 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34881 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34883 PR rtl-optimization/79388
34884 PR rtl-optimization/79450
34885 * combine.c (distribute_notes): When removing TEM_INSN for which
34886 corresponding dest has last value recorded, invalidate that last
34889 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34891 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34892 of explicit '@'. Add missing assembly comment marker on branch costs
34895 2017-02-13 Nathan Sidwell <nathan@acm.org>
34897 * gengtype-lex.l (<in_struct>): Add '/'.
34899 2017-02-13 Martin Liska <mliska@suse.cz>
34902 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34904 2017-02-13 Richard Biener <rguenther@suse.de>
34906 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34908 * configure: Re-generate.
34909 * config.in: Likewise.
34910 * graphite-dependences.c: Simplify as if
34911 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34912 * graphite-isl-ast-to-gimple.c: Likewise.
34913 * graphite-optimize-isl.c: Likewise.
34914 * graphite-poly.c: Likewise.
34915 * graphite-sese-to-poly.c: Likewise.
34916 * graphite.h: Likewise.
34917 * toplev.c: Include isl/version.h and use isl_version () for
34918 printing the ISL version.
34919 * doc/install.texi: Update ISL requirement.
34921 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34923 * doc/standards.texi (Standards): Update reference to
34926 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34928 * doc/extend.texi (Named Address Spaces): sourceware.org now
34930 * doc/install.texi (Binaries): Ditto.
34933 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34935 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34936 standard terminology "stringize"/"stringizing" throughout.
34937 * doc/cppinternals.texi: Likewise.
34939 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34941 * doc/extend.texi: Fix some spelling mistakes and typos.
34942 * doc/invoke.texi: Likewise.
34944 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34947 * params.def (inline-min-speedup) Change from 10 to 8.
34949 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34951 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34954 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34957 * ipa-inline-analysis.c (get_minimal_bb): New function.
34958 (record_modified): Use it.
34959 (remap_edge_change_prob): Handle also ancestor functions.
34961 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34963 * doc/contrib.texi (Contributors): Remove broken link into
34964 the Mauve CVS repository.
34966 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34968 PR middle-end/79454
34969 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34970 result computation whenever lhs doesn't have vector mode, not
34971 just when it has BLKmode.
34973 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34975 * doc/makefile.texi (profiledbootstrap): Refer to the
34976 installation instructions only in textual form.
34978 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34981 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34983 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34985 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34986 (Specific): Update mingw-w64 reference.
34988 (Specific): Remove broken link to Renesas RX processor.
34990 2017-02-10 Richard Biener <rguenther@suse.de>
34992 * toplev.c (process_options): Do not mention obsolete graphite
34993 options when printing sorry message about missing graphite support.
34994 Mention -floop-nest-optimize.
34996 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34998 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34999 (vtst_p16): Likewise.
35000 (vtstq_p8): Likewise.
35001 (vtstq_p16): Likewise.
35003 (vtstq_p64): Likewise.
35004 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35005 (vset_lane_p64): New.
35006 (vsetq_lane_p64): New.
35008 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35010 PR tree-optimization/79411
35011 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35012 stmt operands are SSA_NAMEs used in abnormal phis.
35013 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35016 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35019 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35022 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35024 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35026 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35028 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35032 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35033 not arbitrary TREE_CONSTANT.
35036 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35037 "omp declare target link" attribute unless is_global_var.
35038 * omp-offload.c (find_link_var_op): Likewise.
35040 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35041 Chung-Lin Tang <cltang@codesourcery.com>
35043 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35045 (gimplify_adjust_omp_clauses): Don't delete TILE.
35046 (gimplify_omp_for): Deal with TILE.
35047 * internal-fn.c (expand_GOACC_TILE): New function.
35048 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35050 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35051 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35053 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35054 avoid DIV for outermost collapse var.
35055 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35056 Remove out of date comments, fix whitespace.
35057 * omp-general.c (omp_extract_for_data): Deal with tiling.
35058 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35059 adjust OLF_DIM_BASE value.
35060 (struct omp_for_data): Add tiling field.
35061 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35062 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35063 for auto loops. Remove default auto determining, moved to
35064 oacc_loop_fixed_partitions.
35065 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35066 stmts, add e_mask field.
35067 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35068 (oacc_thread_numbers): Use oacc_dim_call.
35069 (oacc_xform_tile): New.
35070 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35071 (finish_oacc_loop): Adjust for ifns vector.
35072 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35073 add case for GOACC_TILE fns.
35074 (oacc_loop_xform_loop): Delete.
35075 (oacc_loop_process): Iterate over call list directly, and add
35076 handling for GOACC_TILE fns.
35077 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35079 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35080 vector partitioning to outer loops. Assign 2 partitions to loops
35081 when available. Add TILE handling.
35082 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35083 (execite_oacc_device_lower): Process GOACC_TILE fns,
35084 ignore unknown specs.
35085 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35086 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35087 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35089 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35091 * configure.ac (ACX_BUGURL): Update.
35092 * configure: Regenerate.
35094 2017-02-09 Richard Biener <rguenther@suse.de>
35096 PR tree-optimization/69823
35097 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35098 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35100 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35102 * config/arc/arc-c.def: Add __NPS400__ definition.
35103 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35104 (TARGET_NPS400): Define.
35106 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35108 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35110 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35111 pointer, arch_info.
35112 (arc_cpu_types): Fill the arch_info field with a pointer into the
35113 arc_arch_types table.
35114 (arc_selected_cpu): Declare.
35115 * config/arc/arc.c (arc_selected_cpu): Make global.
35116 (arc_selected_arch): Delete.
35117 (arc_base_cpu): Delete.
35118 (arc_override_options): Remove references to deleted variables,
35119 update access to arch information.
35120 (ARC_OPT): Update access to arch information.
35121 (ARC_OPTX): Likewise.
35122 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35123 (TARGET_ARC600): Update access to arch information.
35124 (TARGET_ARC601): Likewise.
35125 (TARGET_ARC700): Likewise.
35126 (TARGET_EM): Likewise.
35127 (TARGET_HS): Likewise.
35128 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35131 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35134 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35135 condition/operands for integer GE/LE/GEU/LEU operations.
35137 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35139 PR translation/79397
35140 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35143 2017-02-08 Martin Jambor <mjambor@suse.cz>
35146 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35147 whether allocation happened.
35148 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35149 nothing was allocated.
35151 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35153 PR tree-optimization/79408
35154 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35155 constant, but SSA_NAME with a known integer range, use the minimum
35156 of that range instead of op1 to determine if modulo can be replaced
35157 with its first operand.
35159 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35161 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35163 2017-02-08 Richard Biener <rguenther@suse.de>
35165 PR tree-optimization/71824
35166 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35167 Check all loops contained in the merged region.
35169 2017-02-07 Andrew Pinski <apinski@cavium.com>
35171 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35173 2017-02-07 Andrew Pinski <apinski@cavium.com>
35175 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35176 (thunderxt88): Likewise.
35177 (thunderxt81): Disable LSE and change v8.1 to v8.
35178 (thunderxt83): Likewise.
35180 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35181 Richard Biener <rguenther@suse.de>
35183 PR middle-end/79399
35184 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35185 type from int to size_t.
35186 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35188 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35190 PR rtl-optimization/79386
35191 * cprop.c (bypass_conditional_jumps): Initialize
35192 bypass_last_basic_block already before splitting bbs after
35193 unconditional traps...
35194 (bypass_conditional_jumps): ... rather than here.
35197 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35198 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35199 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35200 fix -masm=intel patterns.
35202 2017-02-07 Richard Biener <rguenther@suse.de>
35204 PR tree-optimization/79256
35205 PR middle-end/79278
35206 * builtins.c (get_object_alignment_2): Use min_align_of_type
35207 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35208 and ADJUST_FIELD_ALIGN.
35210 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35212 * doc/tm.texi: Regenerate.
35213 * stor-layout.c (layout_decl): Adjust.
35214 (update_alignment_for_field): Likewise.
35215 (place_field): Likewise.
35216 (min_align_of_type): Likewise.
35217 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35218 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35219 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35220 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35221 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35222 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35223 * config/i386/i386.c (x86_field_alignment): Likewise.
35224 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35225 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35226 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35227 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35228 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35229 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35233 2017-01-30 Richard Biener <rguenther@suse.de>
35235 PR tree-optimization/79256
35236 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35237 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35240 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35242 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35243 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35244 builtins to SImode and emit a zero-extend, if necessary.
35246 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35248 * docs/invoke.texi (RISC-V Options): Alphabetize.
35250 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35252 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35255 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35257 * config/riscv/riscv.c: New file.
35258 * common/config/riscv/riscv-common.c: Likewise.
35259 * config.gcc: Likewise.
35260 * config/riscv/constraints.md: Likewise.
35261 * config/riscv/elf.h: Likewise.
35262 * config/riscv/generic.md: Likewise.
35263 * config/riscv/linux.h: Likewise.
35264 * config/riscv/multilib-generator: Likewise.
35265 * config/riscv/peephole.md: Likewise.
35266 * config/riscv/pic.md: Likewise.
35267 * config/riscv/predicates.md: Likewise.
35268 * config/riscv/riscv-builtins.c: Likewise.
35269 * config/riscv/riscv-c.c: Likewise.
35270 * config/riscv/riscv-ftypes.def: Likewise.
35271 * config/riscv/riscv-modes.def: Likewise.
35272 * config/riscv/riscv-opts.h: Likewise.
35273 * config/riscv/riscv-protos.h: Likewise.
35274 * config/riscv/riscv.h: Likewise.
35275 * config/riscv/riscv.md: Likewise.
35276 * config/riscv/riscv.opt: Likewise.
35277 * config/riscv/sync.md: Likewise.
35278 * config/riscv/t-elf-multilib: Likewise.
35279 * config/riscv/t-linux: Likewise.
35280 * config/riscv/t-linux-multilib: Likewise.
35281 * config/riscv/t-riscv: Likewise.
35282 * configure.ac: Likewise.
35283 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35284 Waterman as RISC-V maintainers.
35285 * doc/install.texi: Add RISC-V entries.
35286 * doc/invoke.texi: Add RISC-V options section.
35287 * doc/md.texi: Add RISC-V constraints section.
35288 * configure: Regenerated.
35290 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35293 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35294 false values to be constant vectors with all 0 or all 1 bits set.
35295 (vcondu<mode><mode>): Likewise.
35296 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35298 (fpmask_comparison_operator): Update comment.
35299 (vecint_comparison_operator): New predicate.
35300 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35301 vector conditionals when the true and false values are constant
35302 vectors with all 0 bits or all 1 bits set.
35304 2017-02-06 Martin Sebor <msebor@redhat.com>
35306 PR tree-optimization/79376
35307 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35309 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35311 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35312 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35313 to simplify split condition.
35315 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35317 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35318 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35321 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35323 PR rtl-optimization/68664
35324 * target.def (can_speculate_insn): New hook.
35325 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35326 * doc/tm.texi: Regenerate.
35327 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35328 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35329 (rs6000_sched_can_speculate_insn): New function.
35331 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35333 PR tree-optimization/79284
35334 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35335 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35336 vectorizable_mask_load_store, vectorizable_operation,
35337 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35338 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35339 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35340 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35341 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35342 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35343 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35344 is_gimple_assign (stmt). Replace another such test with
35345 is_gimple_assign (stmt).
35347 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35350 * config/avr/avr.c (rtl-iter.h): Include it.
35351 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35352 (avr_legitimate_combined_insn): ...and implementation.
35354 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35356 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35357 * config/s390/s390.c (s390_const_operand_ok)
35358 (s390_canonicalize_comparison, s390_extract_part)
35359 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35360 (s390_contiguous_bitmask_p, s390_rtx_costs)
35361 (legitimize_pic_address): Likewise.
35362 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35363 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35364 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35365 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35366 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35368 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35370 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35371 REGNO($0) == REGNO($1).
35373 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35375 * config/s390/linux.h(SIZE_TYPE): Add comment.
35377 2017-02-06 Julian Brown <julian@codesourcery.com>
35378 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35379 Virendra Pathak <virendra.pathak@broadcom.com>
35381 * config/aarch64/aarch64-cores.def: Change the scheduler
35383 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35384 * config/aarch64/thunderx2t99.md: New file.
35386 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35388 * doc/standards.texi (Go Language): Update link to language
35391 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35393 * tree-eh.c (lower_resx): Sanitize profile.
35394 (cleanup_empty_eh_move_lp): Likewise.
35396 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35399 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35401 * cfgloopmanip.h (loop_version): Update prototype.
35402 * modulo-sched.c (sms_schedule): Update call of loop_version.
35403 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35404 * tree-parloops.c (gen_parallel_loop): Likewise.
35405 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35406 * tree-ssa-loop-split.c (split_loop): Likewise.
35407 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35408 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35410 2017-02-05 Martin Liska <mliska@suse.cz>
35413 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35415 (print_operand_address): Initialize a struct to zero.
35417 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35419 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35420 garbage collector only in textual form.
35422 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35424 * doc/extend.texi (x86 specific memory model extensions for
35425 transactional memory): Simplify a phrase.
35427 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35430 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35431 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35432 (atomic_storedi_1): Likewise.
35434 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35436 PR tree-optimization/79338
35437 * tree-parloops.c (gather_scalar_reductions): Don't call
35438 vect_analyze_loop_form for loop->inner before destroying loop's
35441 2017-02-03 Martin Sebor <msebor@redhat.com>
35443 PR tree-optimization/79327
35444 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35445 when precision has resulted in leading zeros.
35446 (format_integer): Adjust the likely counter to assume an unknown
35447 argument that may be zero is non-zero.
35449 2017-02-03 Jason Merrill <jason@redhat.com>
35452 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35453 avoid copying non-taken branch.
35455 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35457 PR tree-optimization/79340
35458 * tree-vect-loop.c (vectorizable_reduction): Release
35459 vec_defs elements after safe_splicing them into other vectors.
35462 PR tree-optimization/79327
35463 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35464 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35466 (format_integer): Use wide_int_to_tree instead of build_int_cst
35467 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35468 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35469 of shortest and longest sequence.
35471 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35473 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35474 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35476 2017-02-03 Walter Lee <walt@tilera.com>
35479 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35480 after initial stackframe link reg save.
35481 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35483 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35486 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35487 wu for stxssp alternative.
35489 2017-02-03 Martin Sebor <msebor@redhat.com>
35491 PR tree-optimization/79352
35492 * gimple-fold.c (get_range_strlen): Add argument.
35493 (get_range_strlen): Change return type to bool.
35494 (get_maxval_strlen): Pass in a dummy argument.
35495 * gimple-fold.h (get_range_strlen): Change return type to bool.
35496 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35497 * tree.h (array_at_struct_end_p): Add argument.
35498 * tree.c (array_at_struct_end_p): Handle it.
35500 2017-02-03 Martin Liska <mliska@suse.cz>
35503 * multiple_target.c (create_dispatcher_calls): Redirect edge
35504 from a caller of a dispatcher.
35505 (expand_target_clones): Make the clones local.
35506 (ipa_target_clone): Do both target clones and resolvers.
35507 (ipa_dispatcher_calls): Remove the pass.
35508 (pass_dispatcher_calls::gate): Likewise.
35509 (make_pass_dispatcher_calls): Likewise.
35510 * passes.def (pass_target_clone): Put as very first IPA early
35513 2017-02-03 Martin Liska <mliska@suse.cz>
35515 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35516 in case of a function with ifunc attribute.
35518 2017-02-03 Martin Liska <mliska@suse.cz>
35520 * cgraph.c (cgraph_node::dump): Dump function version info.
35521 * symtab.c (symtab_node::dump_base): Add missing new line.
35523 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35525 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35526 (ifcombine_ifandif): Use it.
35528 2017-02-03 Martin Liska <mliska@suse.cz>
35530 * doc/invoke.texi: Document default value for
35531 use-after-scope-direct-emission-threshold.
35533 2017-02-03 Martin Liska <mliska@suse.cz>
35535 PR tree-optimization/79339
35536 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35537 (format_floating): Likewise.
35539 2017-02-03 Martin Liska <mliska@suse.cz>
35542 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35544 (ipa_node_params_t::remove): Likewise.
35545 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35546 initialization from removed ipa_node_params_t::insert.
35547 (ipa_node_params::~ipa_node_params): Move from removed
35548 ipa_node_params_t::release.
35549 * symbol-summary.h (symbol_summary::m_released): New member.
35550 Do not release a summary twice. Do not allow to call finalizer
35551 for types of a summary that live in GGC memory.
35553 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35555 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35558 2017-02-02 Martin Sebor <msebor@redhat.com>
35560 PR middle-end/79275
35561 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35562 (format_string): Tighten up the range of output for non-constant
35563 strings and correct the expected range for wide non-constant strings.
35565 2017-02-02 Martin Sebor <msebor@redhat.com>
35567 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35569 PR middle-end/32003
35570 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35572 (-fdump-tree-@var): Add to index and document how to come up
35573 with pass-specific option and dump file names.
35574 (-fdump-passes): Clarify where to look for output.
35576 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35578 PR middle-end/77445
35579 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35580 statistics of the analyzed path; allow threading for speed when
35581 any of BBs along the path are optimized for speed.
35583 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35585 PR middle-end/78468
35586 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35587 settings of the virtual registers.
35590 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35592 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35593 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35596 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35598 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35599 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35601 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35603 * config/s390/s390.md: Add missing comments with the expanded
35605 * config/s390/vector.md: Likewise.
35606 * config/s390/vx-builtins.md: Likewise.
35608 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35611 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35612 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35613 conditions on a single line.
35615 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35617 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35618 __S390_VX__ to __VX__.
35620 2017-02-01 Andrew Pinski <apinski@cavium.com>
35622 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35623 stmt_info to record_stmt_cost.
35624 (vect_get_known_peeling_cost): Pass stmt_info if known to
35626 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35627 cpu_vector_cost field into
35628 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35629 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35630 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35631 splitting of scalar_stmt_cost and vec_stmt_cost.
35632 (thunderx_vector_cost): Likewise.
35633 (cortexa57_vector_cost): LIkewise.
35634 (exynosm1_vector_cost): Likewise.
35635 (xgene1_vector_cost): Likewise.
35636 (thunderx2t99_vector_cost): Improve after the splitting of the two
35638 (aarch64_builtin_vectorization_cost): Update for the splitting of
35639 scalar_stmt_cost and vec_stmt_cost.
35641 2017-02-01 Torvald Riegel <triegel@redhat.com>
35642 Richard Henderson <rth@redhat.com>
35644 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35645 conditional on existance of a fast atomic load.
35646 * optabs-query.c (can_atomic_load_p): New function.
35647 * optabs-query.h (can_atomic_load_p): Declare it.
35648 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35649 no fast atomic load is available for the particular size of access.
35650 (expand_atomic_compare_and_swap): Likewise.
35651 (expand_atomic_load): Likewise.
35652 (expand_atomic_store): Likewise.
35653 (expand_atomic_fetch_op): Likewise.
35654 * testsuite/lib/target-supports.exp
35655 (check_effective_target_sync_int_128): Remove x86 because it provides
35656 no fast atomic load.
35657 (check_effective_target_sync_int_128_runtime): Likewise.
35659 2017-02-01 Richard Biener <rguenther@suse.de>
35661 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35662 (graphite_transform_loops): Provide opt-info for optimized nests.
35663 * tree-parloop.c (parallelize_loops): Provide opt-info for
35664 parallelized loops.
35666 2017-02-01 Richard Biener <rguenther@suse.de>
35668 PR middle-end/79315
35669 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35670 was not set before.
35672 2017-02-01 Richard Biener <rguenther@suse.de>
35674 PR tree-optimization/71824
35675 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35676 Verify the loops are valid in the merged SESE region.
35677 (scop_detection::can_represent_loop_1): Check analyzing the
35678 evolution of the number of iterations in the region succeeds.
35680 2017-01-31 Ian Lance Taylor <iant@golang.org>
35682 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35683 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35685 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35687 PR preprocessor/79210
35688 * input.c (get_substring_ranges_for_loc): Replace line_width
35689 assertion with error-handling.
35691 2017-01-31 Richard Biener <rguenther@suse.de>
35693 PR tree-optimization/77318
35694 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35695 (create_pw_aff_from_tree): Take loop parameter.
35696 (add_condition_to_pbb): Pass loop of the condition to
35697 create_pw_aff_from_tree.
35699 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35701 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35702 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35704 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35708 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35710 (convert_int_to_float128): Likewise.
35711 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35712 (convert_int_to_float128): Likewise.
35713 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35714 (UNSPEC_IEEE128_CONVERT): Likewise.
35715 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35716 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35717 Use local variables for IBM extended format.
35718 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35719 (fix_trunc<mode>si2_fprs): Likewise.
35720 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35721 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35722 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35723 to know that we can now have integers of all sizes in vector
35725 (fix<uns>_<mode>di2_hw): Likewise.
35726 (float<uns>_<mode>si2_hw): Likewise.
35727 (fix_<mode>si2_hw): Likewise.
35728 (fixuns_<mode>si2_hw): Likewise.
35729 (float<uns>_<mode>di2_hw): Likewise.
35730 (float_<mode>di2_hw): Likewise.
35731 (float_<mode>si2_hw): Likewise.
35732 (floatuns_<mode>di2_hw): Likewise.
35733 (floatuns_<mode>si2_hw): Likewise.
35734 (xscvqp<su>wz_<mode>): Delete, no longer used.
35735 (xscvqp<su>dz_<mode>): Likewise.
35736 (xscv<su>dqp_<mode>): Likewise.
35737 (ieee128_mfvsrd_64bit): Likewise.
35738 (ieee128_mfvsrd_32bit): Likewise.
35739 (ieee128_mfvsrwz): Likewise.
35740 (ieee128_mtvsrw): Likewise.
35741 (ieee128_mtvsrd_64bit): Likewise.
35742 (ieee128_mtvsrd_32bit): Likewise.
35744 2017-01-31 Martin Liska <mliska@suse.cz>
35747 * ipa-prop.c (ipa_free_all_node_params): Call release method
35748 instead of ~sumbol_summary to not to trigger double times
35751 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35753 PR tree-optimization/71691
35754 * bitmap.h (class auto_bitmap): New.
35755 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35756 is_maybe_undefined instead of ssa_undefined_value_p.
35758 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35760 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35761 __S390_ARCH_LEVEL__ to __ARCH__.
35763 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35765 PR tree-optimization/79267
35766 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35767 if should_remove_lhs_p is true.
35769 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35772 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35773 (add_alignment_attribute): New.
35774 (base_type_die): Add alignment attribute.
35775 (subrange_type_die): Likewise.
35776 (modified_type_die): Likewise.
35777 (gen_array_type_die): Likewise.
35778 (gen_descr_array_type_die: Likewise.
35779 (gen_enumeration_type_die): Likewise.
35780 (gen_subprogram_die): Likewise.
35781 (gen_variable_die): Likewise.
35782 (gen_field_die): Likewise.
35783 (gen_ptr_to_mbr_type_die): Likewise.
35784 (gen_struct_or_union_type_die): Likewise.
35785 (gen_subroutine_type_die): Likewise.
35786 (gen_typedef_die): Likewise.
35787 (base_type_cmp): Compare alignment attribute.
35789 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35792 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35793 (setb_unsigned) New pattern for setb with CCUNS.
35794 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35795 subfc./subfe sequence to avoid overflow problems. Generate a
35796 shorter sequence with cmpld/setb for power9.
35797 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35798 for generating subfc. instruction.
35799 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35800 now uses this instruction.
35802 2017-01-30 Ian Lance Taylor <iant@google.com>
35805 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35806 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35808 2017-01-30 Martin Sebor <msebor@redhat.com>
35810 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35811 Move constant to the right of a relational operator.
35812 (get_mpfr_format_length, format_character, format_string): Ditto.
35813 (should_warn_p, maybe_warn): Same.
35815 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35817 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35820 * asan.c (get_translation_unit_decl): Remove function.
35821 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35823 2017-01-30 Martin Liska <mliska@suse.cz>
35825 PR gcov-profile/79259
35826 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35827 -fprofile-generate.
35829 2017-01-30 Martin Liska <mliska@suse.cz>
35832 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35833 Initialize variables with NULL value.
35835 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35838 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35840 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35842 2017-01-30 Richard Biener <rguenther@suse.de>
35844 PR tree-optimization/79276
35845 * tree-vrp.c (process_assert_insertions): Properly adjust common
35846 when removing a duplicate.
35848 2017-01-30 Richard Biener <rguenther@suse.de>
35850 PR tree-optimization/79256
35851 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35852 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35854 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35856 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35859 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35860 ("*r<noxa>sbg_<mode>_sll_bitmask")
35861 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35862 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35863 Use contiguous_bitmask_nowrap_operand.
35865 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35868 * config/rs6000/altivec.h (vec_xl): Revise #define.
35869 (vec_xst): Likewise.
35871 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35873 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35875 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35877 PR rtl-optimization/79194
35878 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35879 traps before call to bypass_conditional_jumps.
35881 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35883 PR tree-optimization/71374
35884 * lra-constraints.c (check_conflict_input_operands): New.
35885 (match_reload): Use it.
35887 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35890 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35891 account to calculate conflict_set.
35893 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35895 PR rtl-optimization/78559
35896 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35897 other_insn in combine.
35899 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35901 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35902 uint16_type_node for BT_UINT16.
35904 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35906 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35907 "RTL Tests" to menu.
35908 (GIMPLE Tests): New node.
35909 (RTL Tests): New node.
35911 2017-01-27 Richard Biener <rguenther@suse.de>
35913 PR tree-optimization/79245
35914 * tree-loop-distribution.c (distribute_loop): Apply cost
35915 modeling also to detected patterns.
35917 2017-01-27 Richard Biener <rguenther@suse.de>
35919 PR tree-optimization/71433
35920 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35921 (compare_assert_loc): New function.
35922 (process_assert_insertions): Sort and optimize assert locations
35923 to remove duplicates and push down identical assertions on
35924 edges to their destination block.
35926 2017-01-27 Richard Biener <rguenther@suse.de>
35928 PR tree-optimization/79244
35929 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35930 out SSA names even if abnormal.
35932 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35934 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35935 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35936 instead of MPFR_RNDN.
35938 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35941 * arm.c (arm_option_override): Don't call build_target_option_node
35942 until after doing all option overrides.
35943 (arm_valid_target_attribute_tree): Likewise.
35945 2017-01-27 Martin Liska <mliska@suse.cz>
35947 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35948 for {cd}tors and C++ {cd}tors.
35950 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35952 * config/s390/s390.md ("*setmem_long_and")
35953 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35955 2017-01-26 Martin Sebor <msebor@redhat.com>
35957 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35960 2017-01-26 Martin Sebor <msebor@redhat.com>
35962 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35963 HAVE_DFmode before using XFmode or DFmode.
35964 (parse_directive): Avoid using the z length modifier to avoid
35965 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35967 PR middle-end/78703
35968 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35969 to accept adjustment as an array.
35970 (get_int_range): New function.
35971 (struct directive): Make width and prec arrays.
35972 (directive::set_width, directive::set_precision): Call get_int_range.
35973 (format_integer, format_floating): Handle width and precision ranges.
35974 (format_string, parse_directive): Same.
35976 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35979 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35980 comdat_type_p set, just clone them, but keep the children in the
35984 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35985 which have direct callers with -fvar-tracking-assignments enabled
35987 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35988 inside of type units.
35990 2017-01-26 Martin Sebor <msebor@redhat.com>
35992 PR middle-end/78703
35993 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35995 (struct format_result): Replace number_chars, number_chars_min,
35996 and number_chars_max with a single member of struct result_range.
35998 (format_result::operator+=): Adjust.
35999 (struct fmtresult): Remove bounded. Handle likely and unlikely
36001 (fmtresult::adjust_for_width_or_precision): New function.
36002 (fmtresult:type_max_digits): New function.
36003 (bytes_remaining): Handle likely and unlikely counters.
36004 (min_bytes_remaining): Remove.
36005 (format_percent): Simplify.
36006 (format_integer, format_floating): Set likely and unlikely counters.
36007 (get_string_length, format_character, format_string): Same.
36008 (format_plain, should_warn_p): New function.
36009 (maybe_warn): Call should_warn_p. Update diagnostic messages
36010 and handle those for all directives, including plain strings.
36011 (format_directive): Handle likely and unlikely counters.
36012 Remove unnecessary quoting from diagnostics. Add an informational
36014 (add_bytes): Remove.
36015 (pass_sprintf_length::compute_format_length): Simplify.
36016 (try_substitute_return_value): Handle likely and unlikely counters.
36018 2017-01-26 Carl Love <cel@us.ibm.com>
36020 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36021 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36023 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36026 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36027 endianess for subregs into account.
36028 * lra-constraints.c (lra_constraints): Do risky transformations
36029 always on the first iteration.
36030 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36031 last_call_used_reg_set.
36032 (process_bb_lives): Define and use last_call_used_reg_set.
36033 * lra.c (lra): Always continue after lra_constraints on the first
36036 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36038 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36040 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36042 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36044 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36045 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36046 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36047 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36048 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36049 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36050 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36051 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36052 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36054 2017-01-26 Marek Polacek <polacek@redhat.com>
36057 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36058 for the third operand.
36060 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36062 PR middle-end/79236
36063 * omp-low.c (struct omp_context): Add simt_stmt field.
36064 (scan_omp_for): Return omp_context *.
36065 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36066 context to the _simt_ SIMD stmt.
36067 (lower_omp_for): For combined SIMD with sibling _simt_
36068 SIMD, make sure to use the same decls in _looptemp_
36069 clauses as in the sibling.
36071 2017-01-26 David Sherwood <david.sherwood@arm.com>
36073 PR middle-end/79212
36074 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36077 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36080 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36081 emit fld b; fld a; if possible.
36083 * brig-builtins.def: Update copyright years.
36084 * config/arm/arm_acle_builtins.def: Update copyright years.
36086 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36089 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36090 constraint instead of o for the stxsd instruction.
36092 2017-01-25 Carl Love <cel@us.ibm.com>
36094 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36095 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36097 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36099 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36101 2017-01-25 Richard Biener <rguenther@suse.de>
36103 PR tree-optimization/69264
36104 * target.def (vector_alignment_reachable): Improve documentation.
36105 * doc/tm.texi: Regenerate.
36106 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36108 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36109 earlier changes with respect to TYPE_USER_ALIGN.
36110 (vector_alignment_reachable_p): Likewise. Improve dumping.
36112 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36115 * config/arm/arm.md (xordi3): Force constant operand into a register
36118 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36120 * doc/invoke.texi (-fstore-merging): Correct default optimization
36121 levels at which it is enabled.
36122 (-O): Move -fstore-merging from list to...
36125 2017-01-25 Richard Biener <rguenther@suse.de>
36128 * omp-expand.c: Include debug.h.
36129 (expand_omp_taskreg): Make sure to generate early debug before
36130 outlining anything from a function.
36131 (expand_omp_target): Likewise.
36132 (grid_expand_target_grid_body): Likewise.
36134 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36137 * asan.c (get_translation_unit_decl): New function.
36138 (asan_add_global): Extract modules file name from globals
36139 TRANSLATION_UNIT_DECL name.
36141 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36144 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36145 for long calls with APCS frame and VFP.
36147 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36149 * cfg.c (original_copy_tables_initialized_p): New function.
36150 * cfg.h (original_copy_tables_initialized_p): New decl.
36151 * cfgrtl.c (relink_block_chain): Guard the call to
36152 free_original_copy_tables with a call to
36153 original_copy_tables_initialized_p.
36154 * cgraph.h (symtab_node::native_rtl_p): New decl.
36155 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36156 (symtab_node::needed_p): Don't assert for early assembly output
36157 for __RTL functions.
36158 (cgraph_node::finalize_function): Set "force_output" for __RTL
36160 (cgraph_node::analyze): Bail out early for __RTL functions.
36161 (analyze_functions): Update assertion to support __RTL functions.
36162 (cgraph_node::expand): Bail out early for __RTL functions.
36163 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36165 * function.h (struct function): Update comment for field
36167 * gimple-expr.c: Include "tree-pass.h".
36168 (gimple_has_body_p): Return false for __RTL functions.
36169 * Makefile.in (OBJS): Add run-rtl-passes.o.
36170 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36172 (gcc::pass_manager::get_clean_slate): New accessor.
36173 * passes.c: Include "insn-addr.h".
36174 (should_skip_pass_p): Add logging. Update logic for running
36175 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36176 property-provider override so it is only done for gimple passes.
36178 (skip_pass): New function.
36179 (execute_one_pass): Call skip_pass when skipping passes.
36180 * read-md.c (md_reader::read_char): Support filtering
36181 the input to a subset of line numbers.
36182 (md_reader::md_reader): Initialize fields
36183 m_first_line and m_last_line.
36184 (md_reader::read_file_fragment): New function.
36185 * read-md.h (md_reader::read_file_fragment): New decl.
36186 (md_reader::m_first_line): New field.
36187 (md_reader::m_last_line): New field.
36188 * read-rtl-function.c (function_reader::create_function): Only
36189 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36190 curr_properties. Set DECL_INITIAL to a dummy block.
36191 (read_rtl_function_body_from_file_range): New function.
36192 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36194 * run-rtl-passes.c: New file.
36195 * run-rtl-passes.h: New file.
36197 2017-01-24 Jeff Law <law@redhat.com>
36199 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36202 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36204 PR tree-optimization/79159
36205 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36206 (record_nonwrapping_iv): Improve boundary using above function if no
36207 value range information.
36209 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36210 Martin Jambor <mjambor@suse.cz>
36212 * brig-builtins.def: New file.
36213 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36214 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36215 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36216 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36217 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36218 * builtin-types.def (BT_INT8): New.
36219 (BT_INT16): Likewise.
36220 (BT_UINT8): Likewise.
36221 (BT_UINT16): Likewise.
36222 (BT_FN_ULONG): Likewise.
36223 (BT_FN_UINT_INT): Likewise.
36224 (BT_FN_UINT_ULONG): Likewise.
36225 (BT_FN_UINT_LONG): Likewise.
36226 (BT_FN_UINT_PTR): Likewise.
36227 (BT_FN_ULONG_PTR): Likewise.
36228 (BT_FN_INT8_FLOAT): Likewise.
36229 (BT_FN_INT16_FLOAT): Likewise.
36230 (BT_FN_UINT32_FLOAT): Likewise.
36231 (BT_FN_UINT16_FLOAT): Likewise.
36232 (BT_FN_UINT8_FLOAT): Likewise.
36233 (BT_FN_UINT64_FLOAT): Likewise.
36234 (BT_FN_UINT16_UINT32): Likewise.
36235 (BT_FN_UINT32_UINT16): Likewise.
36236 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36237 (BT_FN_INT_PTR_INT): Likewise.
36238 (BT_FN_UINT_PTR_UINT): Likewise.
36239 (BT_FN_LONG_PTR_LONG): Likewise.
36240 (BT_FN_ULONG_PTR_ULONG): Likewise.
36241 (BT_FN_VOID_UINT64_UINT64): Likewise.
36242 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36243 (BT_FN_INT8_INT8_INT8): Likewise.
36244 (BT_FN_INT16_INT16_INT16): Likewise.
36245 (BT_FN_INT_INT_INT): Likewise.
36246 (BT_FN_UINT_FLOAT_UINT): Likewise.
36247 (BT_FN_FLOAT_UINT_UINT): Likewise.
36248 (BT_FN_ULONG_UINT_UINT): Likewise.
36249 (BT_FN_ULONG_UINT_PTR): Likewise.
36250 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36251 (BT_FN_UINT_UINT_UINT): Likewise.
36252 (BT_FN_VOID_UINT_PTR): Likewise.
36253 (BT_FN_UINT_UINT_PTR: Likewise.
36254 (BT_FN_UINT32_UINT64_PTR): Likewise.
36255 (BT_FN_INT_INT_UINT_UINT): Likewise.
36256 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36257 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36258 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36259 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36260 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36261 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36262 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36263 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36264 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36265 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36266 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36267 * doc/frontends.texi: List BRIG FE.
36268 * doc/install.texi (Testing): Add BRIG tesring requirements.
36269 * doc/invoke.texi (Overall Options): Mention BRIG.
36270 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36272 2017-01-24 Richard Biener <rguenther@suse.de>
36274 PR translation/79208
36275 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36277 2017-01-24 Martin Jambor <mjambor@suse.cz>
36280 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36281 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36282 and known_contexts.
36284 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36286 PR middle-end/79123
36287 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36288 casts from signed to unsigned really don't have a range.
36290 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36292 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36293 GMP_RNDx for compatiblity.
36295 2017-01-24 Martin Liska <mliska@suse.cz>
36298 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36299 that would prevent us to call alloca with -1 as argument.
36301 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36303 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36304 Avoid -Wformat-security warning.
36306 2017-01-23 Andrew Pinski <apinski@cavium.com>
36308 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36311 2017-01-23 Martin Sebor <msebor@redhat.com>
36313 PR middle-end/78703
36314 * gimple-ssa-sprintf.c (warn_level): New global.
36315 (format_integer): Use it here and throughout the rest of the file.
36316 Use the same switch to compute sign as base.
36317 (maybe_warn): New function.
36318 (format_directive): Factor out warnings into maybe_warn.
36319 Add debugging output. Use warn_level.
36320 (add_bytes): Use warn_level.
36321 (pass_sprintf_length::compute_format_length): Add debugging output.
36322 (try_substitute_return_value): Same.
36323 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36325 PR middle-end/78703
36326 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36327 (struct fmtresult, format_integer, format_floating): Adjust.
36328 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36329 (get_string_length, format_string,format_directive): Same.
36330 (pass_sprintf_length::compute_format_length): Same.
36331 (try_substitute_return_value): Simplify slightly.
36333 PR middle-end/78703
36334 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36335 (fmtresult::operator+=): Outlined.
36336 (struct fmtresult): Add ctors.
36337 (struct conversion_spec): Rename...
36338 (struct directive): ...to this. Add and remove data members.
36339 (directive::set_width, directive::set_precision): New functions.
36340 (format_percent): Use fmtresult ctor.
36341 (get_width_and_precision): Remove.
36342 (format_integer): Make naming changes. Avoid computing width and
36344 (format_floating): Same. Adjust indentation.
36345 (format_character, format_none): New functions.
36346 (format_string): Moved character handling to format_character.
36347 (format_directive): Remove arguments, change return type.
36348 (parse_directive): New function.
36349 (pass_sprintf_length::compute_format_length): Move directive
36350 parsing to parse_directive.
36352 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36354 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36355 (assign_assembler_name_if_needed): ... this.
36356 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36357 (assign_assembler_name_if_needed): ... this.
36358 (free_lang_data_in_cgraph): Adjust callers.
36359 * cgraphunit.c (cgraph_node::analyze): Likewise.
36360 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36363 2017-01-23 Richard Biener <rguenther@suse.de>
36365 PR tree-optimization/79088
36366 PR tree-optimization/79188
36367 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36368 resetting loop bounds after last path deletion. Reset loop
36369 bounds of the target loop, make code match the comments.
36370 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36371 Make sure loops need no fixups.
36373 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36375 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36376 exponent support with double type for first argument.
36377 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36378 type returned by __builtin_vec_extract_sig,
36379 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36380 functions from "vector int" to "vector unsigned int" or from
36381 "vector long long int" to "vector unsigned long long int".
36382 Changed type returned by __builtin_vec_extract_exp,
36383 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36384 functions from "vector int" to "vector unsigned int" or from
36385 "vector long long int" to "vector unsigned long long int".
36386 Changed return type of __builtin_vec_test_data_class,
36387 __builtin_vec_test_data_class_sp, and
36388 __builtin_vec_test_data_class_dp from "vector int" to
36389 "vector bool int" or from "vector long long int" to "vector bool
36390 long long int" and changed second argument type from "unsigned
36391 int" to "int". Added new overloaded function forms "vector float
36392 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36393 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36394 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36395 double, vector unsigned long long int)" and "vector double
36396 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36397 long int)". Changed return type of
36398 __builtin_scalar_test_data_class and
36399 __builtin_scalar_test_data_class_sp and
36400 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36401 int" and changed second argument from "unsigned int" to "int".
36402 Changed type returned by __builtin_scalar_test_neg,
36403 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36404 from "int" to "bool int". Added new overloaded function form
36405 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36406 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36407 exponent double-precision with floating point first argument.
36408 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36409 documentation of scalar_test_data_class, scalar_test_neg,
36410 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36411 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36412 vec_test_data_class built-in functions to reflect refinements in
36413 their type signatures.
36415 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36417 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36419 (aarch64_elf_asm_destructor): Likewise.
36421 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36423 PR rtl-optimization/78634
36424 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36425 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36426 * ifcvt.c (noce_try_cmove): Add missing cost check.
36428 PR rtl-optimization/71724
36429 * combine.c (if_then_else_cond): Look for situations where it is
36430 beneficial to undo the work of one of the recursive calls.
36432 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36434 PR tree-optimization/70754
36435 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36436 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36437 combined stmt before it if not NULL.
36438 (combine_chains): Process refs reversely and compute dominance point
36441 2017-01-23 Martin Liska <mliska@suse.cz>
36443 PR tree-optimization/79196
36444 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36445 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36447 (strlen_optimize_stmt): Call the renamed function.
36449 2017-01-23 Michael Matz <matz@suse.de>
36451 PR tree-optimization/78384
36452 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36454 2017-01-23 Richard Biener <rguenther@suse.de>
36456 PR tree-optimization/79186
36457 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36458 both incoming edges before moving an assert.
36460 2017-01-23 Martin Jambor <mjambor@suse.cz>
36462 * ipa-prop.c (load_from_param_1): Removed.
36463 (load_from_unmodified_param): Bits from load_from_param_1 put back
36465 (load_from_param): Removed.
36466 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36467 with stmt. Reverted back to use of load_from_unmodified_param.
36469 2017-01-23 Martin Jambor <mjambor@suse.cz>
36472 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36473 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36474 field a pointer to garbage collected vector, mark lattices and
36475 ipcp_orig_node with GTY((skip)).
36476 (ipa_get_param_count): Adjust to descriptors being a pointer.
36477 (ipa_get_param): Likewise.
36478 (ipa_get_type): Likewise.
36479 (ipa_get_param_move_cost): Likewise.
36480 (ipa_set_param_used): Likewise.
36481 (ipa_get_controlled_uses): Likewise.
36482 (ipa_set_controlled_uses): Likewise.
36483 (ipa_is_param_used): Likewise.
36484 (ipa_node_params_t): Move into garbage collector. New methods insert
36486 (ipa_node_params_sum): Annotate wth GTY(()).
36487 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36489 (ipa_load_from_parm_agg): Adjust declaration.
36490 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36491 * ipa-profile.c (ipa_profile): Likewise.
36492 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36493 (ipa_populate_param_decls): Make descriptors parameter garbage
36495 (ipa_dump_param): Adjust to descriptors being a pointer.
36496 (ipa_alloc_node_params): Likewise.
36497 (ipa_initialize_node_params): Likewise.
36498 (load_from_param_1): Make descriptors parameter garbage collected.
36499 (load_from_unmodified_param): Likewise.
36500 (load_from_param): Likewise.
36501 (ipa_load_from_parm_agg): Likewise.
36502 (ipa_node_params::~ipa_node_params): Removed.
36503 (ipa_free_all_node_params): Remove call to delete operator.
36504 (ipa_node_params_t::insert): New.
36505 (ipa_node_params_t::remove): Likewise.
36506 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36507 copy known_csts and known_contexts vectors.
36508 (ipa_read_node_info): Adjust to descriptors being a pointer.
36509 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36511 (ipcp_transform_function): Make descriptors variable garbage
36514 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36516 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36517 * config/i386/avx512dqintrin.h: Ditto.
36518 * config/i386/avx512fintrin.h: Ditto.
36519 * config/i386/i386.c: Handle new builtins.
36520 * config/i386/i386-builtin.def: Add new builtins.
36521 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36522 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36524 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36525 Martin Liska <mliska@suse.cz>
36527 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36528 * asan.c (asan_expand_poison_ifn): Support stores and use
36529 appropriate ASAN report function.
36530 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36531 * internal-fn.def (ASAN_POISON_USE): Declare.
36532 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36533 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36534 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36535 ASAN_POISON calls w/o LHS.
36536 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36537 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36538 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36539 * gimplify.c (asan_poison_variables): Add attribute
36540 use_after_scope_memory to variables that really needs to live
36542 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36543 having the attribute.
36545 2017-01-23 Martin Liska <mliska@suse.cz>
36547 * asan.c (create_asan_shadow_var): New function.
36548 (asan_expand_poison_ifn): Likewise.
36549 * asan.h (asan_expand_poison_ifn): New declaration.
36550 * internal-fn.c (expand_ASAN_POISON): Likewise.
36551 * internal-fn.def (ASAN_POISON): New builtin.
36552 * sanopt.c (pass_sanopt::execute): Expand
36553 asan_expand_poison_ifn.
36554 * tree-inline.c (copy_decl_for_dup_finish): Make function
36556 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36557 * tree-ssa.c (is_asan_mark_p): New function.
36558 (execute_update_addresses_taken): Rewrite local variables
36559 (identified just by use-after-scope as addressable) into SSA.
36561 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36563 * doc/install.texi (Specific): opensource.apple.com uses https
36564 now. Remove trailing slash.
36566 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36568 * README.Portability: Remove note on an Irix compatibility issue.
36570 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36572 * gcov.c (INCLUDE_ALGORITHM): Define.
36573 (INCLUDE_VECTOR): Define.
36574 No longer include <vector> and <algorithm> directly.
36576 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36578 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36580 * doc/invoke.texi (Code Gen Options): Ditto.
36582 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36585 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36587 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36589 rtl-optimization/79125
36590 * cprop.c (local_cprop_pass): Handle cases where we make an
36591 unconditional trap.
36593 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36597 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36598 read from, for big endian.
36600 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36602 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36603 register pauth builtins for LP64 only.
36605 2017-01-20 Marek Polacek <polacek@redhat.com>
36608 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36611 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36613 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36615 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36616 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36617 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36619 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36622 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36623 in big-endian mode when they are not a single duplicated value.
36625 2017-01-20 Richard Biener <rguenther@suse.de>
36627 * BASE-VER: Bump to 7.0.1.
36629 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36631 * omp-low.c (omplow_simd_context): New struct. Use it...
36632 (lower_rec_simd_input_clauses): ...here and...
36633 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36634 references to idx, lane, max_vf, is_simt.
36636 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36638 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36641 2017-01-20 Martin Jambor <mjambor@suse.cz>
36643 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36644 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36646 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36647 (GTFILES): Rename hsa.c to hsa-common.c.
36648 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36649 * hsa-dump.c: Likewise.
36650 * hsa-gen.c: Likewise.
36651 * hsa-regalloc.c: Likewise.
36652 * ipa-hsa.c: Likewise.
36653 * omp-expand.c: Likewise.
36654 * omp-low.c: Likewise.
36655 * toplev.c: Likewise.
36657 2017-01-20 Marek Polacek <polacek@redhat.com>
36660 * doc/invoke.texi: Document -Wduplicated-branches.
36661 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36662 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36663 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36664 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36665 return 0 only when not OEP_LEXICOGRAPHIC.
36666 (fold_build_cleanup_point_expr): Use the expression
36667 location when building CLEANUP_POINT_EXPR.
36668 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36669 * tree.c (add_expr): Handle error_mark_node.
36671 2017-01-20 Martin Liska <mliska@suse.cz>
36674 * tree-profile.c (init_ic_make_global_vars): Do not call
36676 (gimple_init_gcov_profiler): Likewise.
36678 2017-01-20 Martin Liska <mliska@suse.cz>
36681 * cgraph.h (maybe_create_reference): Remove argument and
36683 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36685 * ipa-cp.c (create_specialized_node): Likewise.
36686 * symtab.c (symtab_node::maybe_create_reference): Handle
36687 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36689 2017-01-20 Martin Liska <mliska@suse.cz>
36691 * read-rtl-function.c (function_reader::create_function): Use
36692 build_decl instread of build_decl_stat.
36694 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36696 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36697 * config/i386/avx512dqintrin.h: Ditto.
36698 * config/i386/avx512fintrin.h: Ditto.
36699 * config/i386/i386-builtin-types.def: Add new types.
36700 * config/i386/i386.c: Handle new types.
36701 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36702 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36703 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36704 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36705 (__builtin_ia32_kshiftridi): New.
36706 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36708 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36712 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36713 define to rs6000_init_stack_protect_guard.
36714 (rs6000_init_stack_protect_guard): New function.
36716 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36717 Yunqiang Su <yunqiang.su@imgtec.com>
36719 * config.gcc (supported_defaults): Add madd4.
36720 (with_madd4): Add validation.
36721 (all_defaults): Add madd4.
36722 * config/mips/mips.opt (mmadd4): New option.
36723 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36725 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36727 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36728 (ISA_HAS_FUSED_MADD4): Likewise.
36729 * doc/invoke.texi (-mmadd4): Document the new option.
36730 * doc/install.texi (--with-madd4): Document the new option.
36732 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36734 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36735 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36736 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36737 (aarch64_init_pauth_hint_builtins): New.
36738 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36739 (aarch64_expand_builtin): Expand new builtins.
36741 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36743 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36744 * combine-stack-adj.c (no_unhandled_cfa): Handle
36745 REG_CFA_TOGGLE_RA_MANGLE.
36746 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36747 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36748 info for return address signing.
36749 (aarch64_expand_epilogue): Likewise.
36751 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36753 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36754 * config/aarch64/aarch64-protos.h
36755 (aarch64_return_address_signing_enabled): New declaration.
36756 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36758 (aarch64_expand_prologue): Sign return address before it's pushed onto
36760 (aarch64_expand_epilogue): Authenticate return address fetched from
36762 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36763 (aarch64_attributes): New function attributes for "sign-return-address".
36764 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36765 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36766 ("*do_return"): Generate combined instructions according to key index.
36767 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36768 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36770 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36771 * config/aarch64/aarch64.opt (msign-return-address=): New.
36772 * doc/extend.texi (AArch64 Function Attributes): Documents
36773 "sign-return-address=".
36774 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36776 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36778 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36779 overall option summary.
36781 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36783 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36784 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36785 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36786 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36788 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36790 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36791 -mpower9-minmax by default for -mcpu=power9.
36792 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36793 128-bit floating point.
36795 2017-01-20 Alan Modra <amodra@gmail.com>
36797 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36798 optimizing for size.
36800 2017-01-20 Alan Modra <amodra@gmail.com>
36803 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36804 for strcmp and strncmp from corresponding builtin decl.
36806 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36808 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36809 instead of i386/rtems-64.h.
36810 * config/i386/rtems-64.h: Remove.
36812 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36816 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36818 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36820 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36822 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36823 Change int to HOST_WIDE_INT.
36824 * config/aarch64/aarch64-protos.h
36825 (aarch64_simd_gen_const_vector_dup): Likewise.
36826 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36828 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36830 * langhooks-def.h (lhd_type_for_size): New decl.
36831 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36832 * langhooks.c (lhd_type_for_size): New function, taken from
36835 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36837 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36838 define_bypass for CR latency.
36839 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36840 (power9-alu2): Add define_bypass for CR latency.
36842 (power9-mul): Update insn latency.
36843 (power9-mul-compare): Update insn latency, bypass latency and remove
36846 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36848 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36850 * config/aarch64/aarch64.md
36851 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36852 aarch64_nopcrelative_literal_loads.
36853 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36855 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36857 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36858 TARGET_LOONGSON_3A.
36859 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36861 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36864 * config.gcc (supported_defaults): Add lxc1-sxc1.
36865 (with_lxc1_sxc1): Add validation.
36866 (all_defaults): Add lxc1-sxc1.
36867 * config/mips/mips.opt (mlxc1-sxc1): New option.
36868 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36870 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36871 __mips_no_lxc1_sxc1.
36872 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36873 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36874 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36876 2017-01-19 Richard Biener <rguenther@suse.de>
36878 PR tree-optimization/72488
36879 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36880 sure to restore SSA info.
36881 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36883 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36885 PR rtl-optimization/79121
36886 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36887 of the inner type when shifting an extended value.
36889 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36892 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36893 interposable aliases.
36895 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36898 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36899 Use the evmergelohi instruction.
36900 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36901 (mov_sitf_e500_subreg8_2_be): Likewise.
36902 (mov_sitf_e500_subreg12_2_le): Likewise.
36903 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36904 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36905 (mov_sitf_e500_subreg8_2_le): Likewise.
36906 (mov_sitf_e500_subreg12_2_be): Likewise.
36908 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36910 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36911 attribute from vecsimple to vecperm.
36912 (altivec_vbpermq2): Likewise.
36914 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36917 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36919 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36920 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36921 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36922 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36923 case where N arg is SIZE_MAX.
36924 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36925 (cmpstrsi): Add pattern.
36927 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36929 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36930 __builtin_vec_revb builtins.
36931 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36932 built-in functions to support generation of the ISA 3.0 XXBR<x>
36933 vector byte reverse instructions.
36934 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36935 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36936 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36937 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36938 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36939 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36940 (P9V_BUILTIN_VEC_REVB): Likewise.
36941 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36942 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36943 (p9_xxbrq_v16qi): Likewise.
36944 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36945 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36946 (p9_xxbrh_v8hi): Likewise.
36947 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36948 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36949 vec_revb built-in functions.
36951 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36953 PR rtl-optimization/78952
36954 * config/i386/i386.md (any_extract): New code iterator.
36955 (*insvqi_2): Use any_extract for source operand.
36956 (*insvqi_3): Use any_shiftrt for source operand.
36958 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36960 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36962 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36964 2017-01-18 Matthias Klose <doko@ubuntu.com>
36966 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36968 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36970 * config/rs6000/altivec.h (vec_bperm): Change #define.
36971 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36972 (altivec_vbpermq2): New define_insn.
36973 (altivec_vbpermd): Likewise.
36974 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36975 function interface.
36976 (VBPERMD): Likewise.
36977 (VBPERM): New polymorphic function interface.
36978 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36979 Add entries for P9V_BUILTIN_VEC_VBPERM.
36980 * doc/extend.texi: Add interfaces for vec_bperm.
36982 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36984 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36985 first letter of error messages.
36986 (s390_resolve_overloaded_builtin): Likewise.
36987 * config/s390/s390.c (s390_expand_builtin): Likewise.
36988 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36989 (s390_valid_target_attribute_inner_p): Likewise.
36990 * config/s390/s390.md ("tabort"): Likewise.
36992 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36994 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36995 (ISA_AVOID_DIV_HILO): New macro.
36996 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36997 (ISA_HAS_DDIV): Likewise.
36999 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37001 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37003 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37005 * doc/invoke.texi (fabi-version): Spelling fix.
37007 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37010 * doc/invoke.texi (fabi-version): Mention mangling fix for
37013 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37016 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37018 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37021 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37022 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37024 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37026 (stack_protect_test): Ditto.
37027 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37028 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37030 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37031 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37032 -mstack-protector-guard-offset=.
37033 (RS/6000 and PowerPC Options): Ditto.
37035 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37037 * config/i386/i386.h (MASK_CLASS_P): New define.
37038 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37039 there are no registers from different register sets also when
37040 mask registers are used. Update function comment.
37041 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37042 to (*k/*r) and (*k/*km) alternatives.
37044 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37046 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37047 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37048 (EH_RETURN_HANDLER_RTX): New define.
37049 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37050 Force frame pointer in EH return functions.
37051 (aarch64_expand_epilogue): Add barrier for eh_return.
37052 (aarch64_final_eh_return_addr): Remove.
37053 (aarch64_eh_return_handler_rtx): New function.
37054 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37056 (aarch64_eh_return_handler_rtx): New prototype.
37058 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37060 * config/rs6000/altivec.h (vec_rlmi): New #define.
37061 (vec_vrlnm): Likewise.
37062 (vec_rlnm): Likewise.
37063 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37064 (UNSPEC_VRLNM): Likewise.
37065 (VIlong): New mode iterator.
37066 (altivec_vrl<VI_char>mi): New define_insn.
37067 (altivec_vrl<VI_char>nm): Likewise.
37068 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37070 (VRLDNM): Likewise.
37071 (RLNM): New polymorphic function entry.
37072 (VRLWMI): New monomorphic function entry.
37073 (VRLDMI): Likewise.
37074 (RLMI): New polymorphic function entry.
37075 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37076 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37077 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37080 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37083 * dwarf2out.c (field_byte_offset): Restore the
37084 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37085 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37086 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37089 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37092 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37094 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37097 * dwarf2out.c (add_data_member_location_attribute): For constant
37098 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37099 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37100 DW_AT_byte_size attributes.
37102 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37104 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37105 after forcing to constant memory when the code model is medium.
37107 2017-01-17 Julia Koval <julia.koval@intel.com>
37110 * config/i386/avx512fintrin.h
37111 (_mm512_i32gather_ps): Change __addr type to void const*.
37112 (_mm512_mask_i32gather_ps): Ditto.
37113 (_mm512_i32gather_pd): Ditto.
37114 (_mm512_mask_i32gather_pd): Ditto.
37115 (_mm512_i64gather_ps): Ditto.
37116 (_mm512_mask_i64gather_ps): Ditto.
37117 (_mm512_i64gather_pd): Ditto.
37118 (_mm512_mask_i64gather_pd): Ditto.
37119 (_mm512_i32gather_epi32): Ditto.
37120 (_mm512_mask_i32gather_epi32): Ditto.
37121 (_mm512_i32gather_epi64): Ditto.
37122 (_mm512_mask_i32gather_epi64): Ditto.
37123 (_mm512_i64gather_epi32): Ditto.
37124 (_mm512_mask_i64gather_epi32): Ditto.
37125 (_mm512_i64gather_epi64): Ditto.
37126 (_mm512_mask_i64gather_epi64): Ditto.
37127 (_mm512_i32scatter_ps): Change __addr type to void*.
37128 (_mm512_mask_i32scatter_ps): Ditto.
37129 (_mm512_i32scatter_pd): Ditto.
37130 (_mm512_mask_i32scatter_pd): Ditto.
37131 (_mm512_i64scatter_ps): Ditto.
37132 (_mm512_mask_i64scatter_ps): Ditto.
37133 (_mm512_i64scatter_pd): Ditto.
37134 (_mm512_mask_i64scatter_pd): Ditto.
37135 (_mm512_i32scatter_epi32): Ditto.
37136 (_mm512_mask_i32scatter_epi32): Ditto.
37137 (_mm512_i32scatter_epi64): Ditto.
37138 (_mm512_mask_i32scatter_epi64): Ditto.
37139 (_mm512_i64scatter_epi32): Ditto.
37140 (_mm512_mask_i64scatter_epi32): Ditto.
37141 (_mm512_i64scatter_epi64): Ditto.
37142 (_mm512_mask_i64scatter_epi64): Ditto.
37143 * config/i386/avx512pfintrin.h
37144 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37145 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37146 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37147 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37148 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37149 (_mm512_prefetch_i32scatter_ps): Ditto.
37150 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37151 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37152 (_mm512_prefetch_i64scatter_pd): Ditto.
37153 (_mm512_prefetch_i64scatter_ps): Ditto.
37154 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37155 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37156 * config/i386/avx512vlintrin.h
37157 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37158 (_mm_mmask_i32gather_ps): Ditto.
37159 (_mm256_mmask_i32gather_pd): Ditto.
37160 (_mm_mmask_i32gather_pd): Ditto.
37161 (_mm256_mmask_i64gather_ps): Ditto.
37162 (_mm_mmask_i64gather_ps): Ditto.
37163 (_mm256_mmask_i64gather_pd): Ditto.
37164 (_mm_mmask_i64gather_pd): Ditto.
37165 (_mm256_mmask_i32gather_epi32): Ditto.
37166 (_mm_mmask_i32gather_epi32): Ditto.
37167 (_mm256_mmask_i32gather_epi64): Ditto.
37168 (_mm_mmask_i32gather_epi64): Ditto.
37169 (_mm256_mmask_i64gather_epi32): Ditto.
37170 (_mm_mmask_i64gather_epi32): Ditto.
37171 (_mm256_mmask_i64gather_epi64): Ditto.
37172 (_mm_mmask_i64gather_epi64): Ditto.
37173 (_mm256_i32scatter_ps): Change __addr type to void*.
37174 (_mm256_mask_i32scatter_ps): Ditto.
37175 (_mm_i32scatter_ps): Ditto.
37176 (_mm_mask_i32scatter_ps): Ditto.
37177 (_mm256_i32scatter_pd): Ditto.
37178 (_mm256_mask_i32scatter_pd): Ditto.
37179 (_mm_i32scatter_pd): Ditto.
37180 (_mm_mask_i32scatter_pd): Ditto.
37181 (_mm256_i64scatter_ps): Ditto.
37182 (_mm256_mask_i64scatter_ps): Ditto.
37183 (_mm_i64scatter_ps): Ditto.
37184 (_mm_mask_i64scatter_ps): Ditto.
37185 (_mm256_i64scatter_pd): Ditto.
37186 (_mm256_mask_i64scatter_pd): Ditto.
37187 (_mm_i64scatter_pd): Ditto.
37188 (_mm_mask_i64scatter_pd): Ditto.
37189 (_mm256_i32scatter_epi32): Ditto.
37190 (_mm256_mask_i32scatter_epi32): Ditto.
37191 (_mm_i32scatter_epi32): Ditto.
37192 (_mm_mask_i32scatter_epi32): Ditto.
37193 (_mm256_i32scatter_epi64): Ditto.
37194 (_mm256_mask_i32scatter_epi64): Ditto.
37195 (_mm_i32scatter_epi64): Ditto.
37196 (_mm_mask_i32scatter_epi64): Ditto.
37197 (_mm256_i64scatter_epi32): Ditto.
37198 (_mm256_mask_i64scatter_epi32): Ditto.
37199 (_mm_i64scatter_epi32): Ditto.
37200 (_mm_mask_i64scatter_epi32): Ditto.
37201 (_mm256_i64scatter_epi64): Ditto.
37202 (_mm256_mask_i64scatter_epi64): Ditto.
37203 (_mm_i64scatter_epi64): Ditto.
37204 (_mm_mask_i64scatter_epi64): Ditto.
37205 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37206 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37207 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37208 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37209 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37210 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37211 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37212 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37213 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37214 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37215 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37216 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37217 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37218 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37219 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37220 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37221 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37222 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37223 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37224 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37225 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37226 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37227 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37228 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37229 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37230 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37231 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37232 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37233 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37234 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37235 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37236 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37237 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37238 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37239 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37240 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37241 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37242 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37243 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37244 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37245 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37246 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37247 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37248 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37249 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37250 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37251 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37252 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37253 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37254 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37255 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37256 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37257 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37258 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37259 definitions accordingly.
37261 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37262 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37265 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37268 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37271 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37272 hard regs for inner regno.
37274 2017-01-17 Martin Liska <mliska@suse.cz>
37277 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37278 assumption and add comment.
37280 2017-01-17 Nathan Sidwell <nathan@acm.org>
37282 * ipa-visibility.c (localize_node): New function, broken out of ...
37283 (function_and_variable_visibility): ... here. Call it.
37285 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37287 PR middle-end/77445
37288 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37289 correctly set frequency of oudgoing edge.
37290 (duplicate_thread_path): Fix profile updating.
37292 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37295 * configure.ac: Add GCC_BASE_VER.
37296 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37297 version from BASE-VER file.
37298 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37299 (gcc.o): Depend on $(BASEVER).
37300 * common.opt (dumpfullversion): New option.
37301 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37302 * doc/invoke.texi: Document -dumpfullversion.
37303 * doc/install.texi: Document --with-gcc-major-version-only.
37304 * configure: Regenerated.
37306 2017-01-17 Richard Biener <rguenther@suse.de>
37308 PR tree-optimization/71433
37309 * tree-vrp.c (register_new_assert_for): Merge same asserts
37310 on all incoming edges.
37311 (process_assert_insertions_for): Handle insertions at the
37314 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37316 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37317 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37319 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37322 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37325 2017-01-17 Alan Modra <amodra@gmail.com>
37328 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37329 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37330 symbolic stack limit when pic.
37332 2017-01-16 Martin Sebor <msebor@redhat.com>
37334 PR tree-optimization/78608
37335 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37337 2017-01-16 Jeff Law <law@redhat.com>
37340 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37341 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37342 for several include directories that may be relative to sysroot.
37343 * config/i386/x-mingw32 (gplus_includedir): Define.
37344 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37345 (native_system_includedir): Likewise.
37346 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37347 override if TARGET_SYSTEM_ROOT is defined.
37348 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37350 PR tree-optimization/79090
37351 PR tree-optimization/33562
37352 PR tree-optimization/61912
37353 PR tree-optimization/77485
37354 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37355 and computed trims into the dump file.
37357 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37359 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37361 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37364 * gimplify.c (gimplify_init_constructor): If want_value and
37365 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37369 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37370 sequence. Formatting fixes.
37371 (doloop_optimize): Formatting fixes.
37374 * gcc.c (debug_level_greater_than_spec_func): New function.
37375 (static_spec_functions): Add debug-level-gt spec function.
37376 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37378 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37379 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37380 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37381 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37382 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37383 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37385 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37387 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37388 QImode fixups to general and mask registers only.
37390 2017-01-16 Carl Love <cel@us.ibm.com>
37392 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37393 for built-in functions
37394 vector signed char vec_nabs (vector signed char)
37395 vector signed short vec_nabs (vector signed short)
37396 vector signed int vec_nabs (vector signed int)
37397 vector signed long long vec_nabs (vector signed long long)
37398 vector float vec_nabs (vector float)
37399 vector double vec_nabs (vector double)
37400 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37402 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37403 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37404 * doc/extend.texi: Update the documentation file for the new built-in
37407 2017-01-16 Martin Sebor <msebor@redhat.com>
37409 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37412 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37414 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37415 UNSPEC_VSX__XXSPLTD to require special splat handling.
37417 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37420 * system.h: Poison strndup.
37422 2017-01-16 Alan Modra <amodra@gmail.com>
37425 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37428 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37430 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37432 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37434 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37435 call recog here. Assert that INSN_CODE (insn) is non-negative.
37437 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37440 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37442 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37443 in the currently scheduled RTL fragment.
37445 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37447 PR rtl-optimization/78751
37448 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37451 2017-01-14 Jeff Law <law@redhat.com>
37453 PR tree-optimization/79090
37454 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37455 variable length stores.
37456 (compute_trims): Delete dead assignment to *trim_tail.
37457 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37460 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37462 PR rtl-optimization/78626
37463 PR rtl-optimization/78727
37464 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37465 of a block, and split such blocks after everything else is finished.
37467 2017-01-14 Alan Modra <amodra@gmail.com>
37470 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37471 target legitimate_combined_insn.
37472 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37473 (rs6000_legitimate_combined_insn): New function.
37474 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37476 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37477 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37478 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37480 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37482 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37484 2017-01-13 Jeff Law <law@redhat.com>
37486 PR tree-optimization/33562
37487 PR tree-optimization/61912
37488 PR tree-optimization/77485
37489 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37491 (delete_dead_assignment): Likewise.
37492 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37493 statement to delete_dead_call and delete_dead_assignment.
37495 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37498 * substring-locations.c (format_warning_va): Strengthen case 1 so
37499 that both endpoints of the substring must be within the format
37500 range for just the substring to be printed.
37502 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37504 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37505 * config/i386/i386.c (ix86_target_string): Add missing options
37506 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37507 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37508 flags_other and ix86_target_other to flags2_other. Display unknown
37510 (ix86_valid_target_attribute_inner_p): Add missing options and
37511 reorder options by implied ISAs, as in ix86_target_string.
37513 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37515 * hash-table.h (hash_table::too_empty_p): New function.
37516 (hash_table::expand): Use it.
37517 (hash_table::traverse): Likewise.
37518 (hash_table::empty_slot): Use sizeof (value_type) instead of
37519 sizeof (PTR) to convert bytes to elements. Shrink the table
37520 if the current size is excessive for the current number of
37523 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37525 * ira-costs.c (record_reg_classes): Break from the inner loop
37526 early once alt_fail is known to be true. Update outer loop
37527 handling accordingly.
37529 2017-01-13 Jeff Law <law@redhat.com>
37531 * tree-ssa-dse.c (decrement_count): New function.
37532 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37533 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37534 when we know the partially dead statement is a mem* function.
37536 PR tree-optimization/61912
37537 PR tree-optimization/77485
37538 * tree-ssa-dse.c: Include expr.h.
37539 (maybe_trim_constructor_store): New function.
37540 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37542 PR tree-optimization/33562
37543 PR tree-optimization/61912
37544 PR tree-optimization/77485
37545 * doc/invoke.texi: Document new dse-max-object-size param.
37546 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37547 * tree-ssa-dse.c: Include params.h.
37548 (dse_store_status): New enum.
37549 (initialize_ao_ref_for_dse): New, partially extracted from
37551 (valid_ao_ref_for_dse, normalize_ref): New.
37552 (setup_live_bytes_from_ref, compute_trims): Likewise.
37553 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37554 (maybe_trim_partially_dead_store): Likewise.
37555 (maybe_trim_complex_store): Likewise.
37556 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37557 Track what bytes live from the original store. Return tri-state
37558 for dead, partially dead or live.
37559 (dse_dom_walker): Add constructor, destructor and new private members.
37560 (delete_dead_call, delete_dead_assignment): New extracted from
37562 (dse_optimize_stmt): Make a member of dse_dom_walker.
37563 Use initialize_ao_ref_for_dse.
37565 PR tree-optimization/33562
37566 PR tree-optimization/61912
37567 PR tree-optimization/77485
37568 * sbitmap.h (bitmap_count_bits): Prototype.
37569 (bitmap_clear_range, bitmap_set_range): Likewise.
37570 * sbitmap.c (bitmap_clear_range): New function.
37571 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37573 2017-01-13 Martin Liska <mliska@suse.cz>
37576 * function.c (set_cfun): Add new argument force.
37577 * function.h (set_cfun): Likewise.
37578 * ipa-inline-transform.c (inline_call): Use the function when
37579 strict alising from is dropped for function we inline to.
37581 2017-01-13 Richard Biener <rguenther@suse.de>
37583 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37584 for dumping GIMPLE INTEGER_CSTs.
37586 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37588 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37589 to 201112L since C++17.
37591 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37594 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37595 if -fsanitize=kernel-address is present.
37597 2017-01-13 Richard Biener <rguenther@suse.de>
37599 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37600 as _Literal ( type ) number in case usual suffixes do not
37601 preserve all information.
37603 2017-01-13 Richard Biener <rguenther@suse.de>
37605 PR tree-optimization/77283
37606 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37607 and ssa-iterators.h.
37608 (is_feasible_trace): Implement a cost model based on joiner
37611 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37614 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37615 char or short to __float128/_Float128 directly.
37617 2017-01-12 Martin Sebor <msebor@redhat.com>
37619 to -Wformat-overflow.
37620 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37621 (min_bytes_remaining): Same.
37622 (get_string_length): Same.
37623 (format_string): Same.
37624 (format_directive): Same.
37626 (pass_sprintf_length::handle_gimple_call): Same.
37628 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37630 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37631 info.nowrite calls with no lhs that can't throw. Return bool
37632 whether gsi_remove has been called or not.
37633 (pass_sprintf_length::handle_gimple_call): Return bool whether
37634 try_substitute_return_value called gsi_remove. Formatting fix.
37635 (pass_sprintf_length::execute): Don't use gsi_remove if
37636 handle_gimple_call returned true.
37639 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37640 be removed due to side-effects, don't remove following barrier nor
37641 turn the successor edge into fallthru edge.
37643 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37646 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37647 element-reversing loads and stores as not swappable.
37649 2017-01-12 Nathan Sidwell <nathan@acm.org>
37650 Nicolai Stange <nicstange@gmail.com>
37652 * combine.c (try_combine): Don't ignore result of overlap checking
37653 loop. Combine overlap & asm check into single loop.
37655 2017-01-12 Richard Biener <rguenther@suse.de>
37657 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37658 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37660 2017-01-12 Richard Biener <rguenther@suse.de>
37662 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37663 and TS_TARGET_OPTION directly derive from TS_BASE.
37664 * tree-core.h (tree_optimization_option): Derive from tree_base.
37665 (tree_target_option): Likewise.
37667 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37669 * config/i386/i386.c (memory_address_length): Increase len
37670 only when rip_relative_addr_p returns false.
37672 2017-01-11 Julia Koval <julia.koval@intel.com>
37674 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37675 (OPTION_MASK_ISA_SGX_SET): New.
37676 (ix86_handle_option): Handle OPT_msgx.
37677 * config.gcc: Added sgxintrin.h.
37678 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37679 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37680 * config/i386/i386.c (ix86_target_string): Add -msgx.
37682 (ix86_option_override_internal): Handle new options.
37683 (ix86_valid_target_attribute_inner_p): Add sgx.
37684 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37685 * config/i386/i386.opt: Add msgx.
37686 * config/i386/sgxintrin.h: New file.
37687 * config/i386/x86intrin.h: Add sgxintrin.h.
37689 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37692 * fold-const.c (maybe_nonzero_address): Return 1 for function
37694 (tree_single_nonzero_warnv_p): Don't handle function local objects
37698 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37701 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37704 * opts.c: Include "spellcheck.h"
37705 (struct string_fragment): New struct.
37706 (struct edit_distance_traits<const string_fragment &>): New
37708 (get_closest_sanitizer_option): New function.
37709 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37711 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37713 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37715 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37716 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37717 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37718 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37719 for initial die_offset if dwarf_split_debug_info.
37720 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37721 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37723 (output_skeleton_debug_sections): Formatting fix. Use
37724 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37725 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37727 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37729 * config/arm/cortex-a53.md: Add bypasses for
37730 cortex_a53_r2f_cvt.
37731 (cortex_a53_r2f): Only use for transfers.
37732 (cortex_a53_f2r): Likewise.
37733 (cortex_a53_r2f_cvt): Add reservation for conversions.
37734 (cortex_a53_f2r_cvt): Likewise.
37736 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37738 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37739 to all inlined functions, change static to extern.
37741 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37744 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37746 (arm_assemble_integer): Likewise.
37748 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37750 * config.gcc: Use new awk script to check CPU, FPU and architecture
37751 parameters for --with-... options.
37752 * config/arm/parsecpu.awk: New file
37753 * config/arm/arm-cpus.in: New file.
37754 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37756 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37758 * config/arm/t-arm: Update dependency rules.
37759 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37760 of processing .def files.
37761 * config/arm/genopt.sh: Deleted.
37762 * config/arm/gentune.sh: Deleted.
37763 * config/arm/arm-cores.def: Deleted.
37764 * config/arm/arm-arches.def: Deleted.
37765 * config/arm/arm-fpus.def: Deleted.
37766 * config/arm/arm-tune.md: Regenerated.
37767 * config/arm/arm-tables.opt: Regenerated.
37768 * config/arm/arm-cpu.h: New generated file.
37769 * config/arm/arm-cpu-data.h: New generated file.
37770 * config/arm/arm-cpu-cdata.h: New generated file.
37772 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37775 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37777 (input_varpool_node): Unpack dynamically_initialized bit.
37779 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37781 PR rtl-optimization/79032
37782 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37783 the alignment of the adjusted memory reference against that of MODE,
37784 instead of the alignment of the original memory reference.
37786 2017-01-11 Martin Jambor <mjambor@suse.cz>
37788 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37790 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37791 decorated functions.
37793 2017-01-11 Richard Biener <rguenther@suse.de>
37795 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37796 set range/nonnull info for PHI results. Do not set it on
37797 stmts marked for removal.
37799 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37801 * expr.c (store_field): In the bitfield case, fetch the return value
37802 from the registers before applying a single big-endian adjustment.
37803 Always do a final load for a BLKmode value not larger than a word.
37805 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37808 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37809 that we correctly handle column numbers greater than
37810 LINE_MAP_MAX_COLUMN_NUMBER.
37812 2017-01-10 Martin Sebor <msebor@redhat.com>
37814 PR middle-end/78245
37815 * gimple-ssa-sprintf.c (get_destination_size): Call
37816 {init,fini}object_sizes.
37817 * tree-object-size.c (addr_object_size): Adjust.
37818 (pass_through_call): Adjust.
37819 (pass_object_sizes::execute): Adjust.
37820 * tree-object-size.h (fini_object_sizes): Declare.
37822 2017-01-10 Martin Sebor <msebor@redhat.com>
37824 PR tree-optimization/78775
37825 * builtins.c (get_size_range): Move...
37826 * calls.c: ...to here.
37827 (alloc_max_size): Accept zero argument.
37828 (operand_signed_p): Remove.
37829 (maybe_warn_alloc_args_overflow): Call get_size_range.
37830 * calls.h (get_size_range): Declare.
37832 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37834 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37835 from TI's devices.csv file as of September 2016.
37836 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37838 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37840 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37841 * doc/invoke.texi: Likewise.
37842 * doc/md.texi: Likewise.
37843 * doc/objc.texi: Likewise.
37845 2017-01-10 Joshua Conner <joshconner@google.com>
37847 * config/arm/fuchsia-elf.h: New file.
37848 * config/fuchsia.h: New file.
37849 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37850 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37852 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37854 2016-01-10 Richard Biener <rguenther@suse.de>
37856 PR tree-optimization/79034
37857 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37858 Propagate out degenerate PHIs in the joiner.
37860 2017-01-10 Martin Liska <mliska@suse.cz>
37862 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37863 (sort_congruence_classes_by_decl_uid): Likewise.
37864 (sort_congruence_class_groups_by_decl_uid): Likewise.
37865 (sem_item_optimizer::merge_classes): Sort class, groups in these
37866 classes and members in the groups by DECL_UID of declarations.
37867 This would make merge operations stable.
37869 2017-01-10 Martin Liska <mliska@suse.cz>
37871 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37872 usage of m_classes_vec.
37873 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37874 (sem_item_optimizer::get_group_by_hash): Likewise.
37875 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37876 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37877 (sem_item_optimizer::verify_classes): Likewise.
37878 (sem_item_optimizer::process_cong_reduction): Likewise.
37879 (sem_item_optimizer::dump_cong_classes): Likewise.
37880 (sem_item_optimizer::merge_classes): Likewise.
37881 * ipa-icf.h (congruence_class_hash): Rename from
37882 congruence_class_group_hash. Remove declaration of m_classes_vec.
37884 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37886 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37887 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37888 * config.gcc: Add avx512vpopcntdqintrin.h.
37889 * config/i386/avx512vpopcntdqintrin.h: New.
37890 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37891 * config/i386/i386-builtin-types.def: Add new types.
37892 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37893 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37894 __builtin_ia32_vpopcountq_v8di_mask): New.
37895 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37896 __AVX512VPOPCNTDQ__.
37897 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37898 (PTA_AVX512VPOPCNTDQ): Define.
37899 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37900 TARGET_AVX512VPOPCNTDQ_P): Define.
37901 * config/i386/i386.opt: Add mavx512vpopcntdq.
37902 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37903 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37905 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37907 PR middle-end/77484
37908 * predict.def (PRED_CALL): Set to 67.
37910 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37912 * expr.c (store_field): In the bitfield case, if the value comes from
37913 a function call and is of an aggregate type returned in registers, do
37914 not modify the field mode; extract the value in all cases if the mode
37915 is BLKmode and the size is not larger than a word.
37917 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37920 * config/i386/cpuid.h: Fix undefined behavior.
37922 2017-01-04 Jeff Law <law@redhat.com>
37924 PR tree-optimization/79007
37925 PR tree-optimization/67955
37926 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37927 conservative for pt.null when flag_non_call_exceptions is on.
37929 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37931 PR translation/79019
37932 PR translation/79020
37933 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37934 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37935 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37937 * config/avr/avr.opt (maccumulate-args): Likewise.
37938 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37939 * common.opt (freport-bug): Likewise.
37940 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37941 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37942 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37943 translatable string.
37944 * config/i386/i386.c (function_value_32): Likewise.
37945 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37946 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37948 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37949 * common/config/msp430/msp430-common.c (msp430_handle_option):
37951 * symtab.c (symtab_node::verify_base): Likewise.
37952 * opts.c (set_debug_level): Likewise.
37953 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37954 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37955 missing whitespace to translatable strings.
37956 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37957 * config/sh/superh.h: Likewise.
37958 * config/i386/xopintrin.h: Likewise.
37959 * config/i386/znver1.md: Likewise.
37960 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37961 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37962 * double-int.h (struct double_int): Likewise.
37963 * double-int.c (div_and_round_double): Likewise.
37964 * wide-int.cc: Likewise.
37965 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37966 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37967 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37968 (crossjumps_occurred): ... this.
37969 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37972 PR tree-optimization/78899
37973 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37974 returning bool return struct loop *, NULL for failure and the new
37976 (versionable_outer_loop_p): Don't version outer loop if it has
37977 dont_vectorized bit set.
37978 (tree_if_conversion): When versioning outer loop, ensure
37979 tree_if_conversion is performed also on the inner loop of the
37980 non-vectorizable outer loop copy.
37981 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37982 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37983 prevent vectorization of it.
37984 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37985 the outer loop vectorization of the non-scalar version is attempted
37986 before vectorization of the inner loop in scalar version. If
37987 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37988 vectorization of its inner loop.
37989 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37990 has 2 inner loops, rename also on edges from bb whose single pred
37991 is outer_loop->header. Fix typo in function comment.
37993 2017-01-09 Martin Sebor <msebor@redhat.com>
37996 * asan.c (asan_emit_stack_protection): Increase local buffer size
37997 to avoid snprintf truncation warning.
37999 2017-01-09 Andrew Pinski <apinski@cavium.com>
38001 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38002 to reference thunderx2t99 for the tuning structure
38003 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38005 (thunderx2t99_extra_costs): This.
38006 * config/aarch64/aarch64-tune.md: Regenerate.
38007 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38008 (vulcan_addrcost_table): This.
38009 (vulcan_regmove_cost): Rename to ...
38010 (thunderx2t99_regmove_cost): This.
38011 (vulcan_vector_cost): Rename to ...
38012 (thunderx2t99_vector_cost): this.
38013 (vulcan_branch_cost): Rename to ...
38014 (thunderx2t99_branch_cost): This.
38015 (vulcan_tunings): Rename to ...
38016 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38017 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38019 2017-01-09 Martin Jambor <mjambor@suse.cz>
38023 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38024 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38025 (propagate_vr_accross_jump_function): Use the above function for all
38026 value range computations for pass-through jump functions and type
38027 converasion from explicit value range values.
38028 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38029 parameters from TYPE_ARG_TYPES.
38030 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38031 (ipa_write_node_info): Stream type of the actual argument.
38032 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38034 2017-01-09 Martin Liska <mliska@suse.cz>
38037 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38038 (lookup_compiler): Do not show error message with have_E.
38040 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38042 PR tree-optimization/78938
38043 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38044 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38045 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38046 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38049 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38051 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38054 2017-01-09 Richard Biener <rguenther@suse.de>
38056 PR tree-optimization/78997
38057 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38058 name condition properly.
38060 2017-01-09 Richard Biener <rguenther@suse.de>
38063 * dwarf2out.c (is_cxx): New overload with context.
38064 (is_naming_typedef_decl): Use it.
38066 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38068 * invoke.texi (Option Summary): Correct spacing in option lists
38069 and add line breaks to fix over-long lines.
38071 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38073 PR middle-end/17660
38075 * extend.texi (Common Variable Attributes): Add xref to GCC
38076 Internals manual to explain mode attribute keywords.
38078 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38081 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38082 and Preprocessor Options.
38083 (Options for Linking): Document -pthread here....
38084 (RS/6000 and PowerPC Options): ...not here.
38085 (Solaris 2 Options): ...or here.
38086 * doc/cppopts.texi: Document -pthread.
38088 2017-01-08 Martin Sebor <msebor@redhat.com>
38090 PR middle-end/77708
38091 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38092 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38093 New member functions.
38094 (format_directive): Used them.
38096 (pass_sprintf_length::handle_gimple_call): Same.
38097 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38098 to avoid truncation for any argument.
38099 (extract_affine_mul): Same.
38100 * tree.c (get_file_function_name): Same.
38102 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38104 PR middle-end/77484
38105 * predict.def (PRED_INDIR_CALL): Set to 86.
38107 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38109 PR preprocessor/54124
38110 * doc/cppopts.texi: Reformat -d subtable to list the full name
38111 of the options. Add cross-reference to the docs for the general
38112 compiler -d options.
38113 * doc/invoke.texi (Developer Options): Add cross-reference to the
38114 preprocessor-specific -d option documentation.
38116 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38118 PR preprocessor/13498
38119 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38120 redudant material, and reflect new command-line options.
38121 (System Headers): Likewise.
38123 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38125 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38126 -isystem, and -idirafter. Copy-edit.
38127 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38128 default for -ftrack-macro-expansion. Delete obsolete and
38129 badly-formatted implementation details about -fdebug-cpp output.
38130 * doc/cppwarnopts.texi: Copy-edit.
38132 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38135 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38136 that the transition from a max line width >= 1<<10 to narrower
38137 lines works correctly.
38139 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38141 * doc/options.texi (PerFunction): New.
38142 * opt-functions.awk (switch_flags): Map both Optimization and
38143 PerFunction to CL_OPTIMIZATION.
38144 * opth-gen.awk: Test for PerFunction flag along with
38146 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38147 it only when the latter is present. Skip those that don't in
38148 the hash function generator.
38149 * common.opt (fvar-tracking): Mark as PerFunction instead of
38151 (fvar-tracking-assignments): Likewise.
38152 (fvar-tracking-assignments-toggle): Likewise.
38153 (fvar-tracking-uninit): Likewise.
38155 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38157 PR translation/79018
38158 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38161 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38164 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38165 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38166 TARGET_LONG_JUMP_TABLE_OFFSETS.
38167 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38168 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38169 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38170 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38171 * config/m68k/m68k.md (tablejump expander): Likewise.
38172 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38173 TARGET_LONG_JUMP_TABLE_OFFSETS.
38174 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38175 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38177 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38178 David Holsgrove <david.holsgrove@xilinx.com>
38180 * common/config/microblaze/microblaze-common.c
38181 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38182 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38184 * config/microblaze/microblaze.c (microblaze_must_save_register)
38185 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38187 (microblaze_eh_return): New function.
38188 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38189 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38190 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38191 * config/microblaze/microblaze.md (eh_return): New pattern.
38193 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38195 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38196 GCC_DIAGNOSTIC_STRINGIFY): Define.
38198 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38200 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38202 * config/arm/arm.md (<mcrr>): New.
38204 * config/arm/arm.c (arm_arch5te): New.
38205 (arm_option_override): Set arm_arch5te.
38206 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38208 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38209 (arm_mcrr_qualifiers): ... this. New.
38210 (MRRC_QUALIFIERS): Define to...
38211 (arm_mrrc_qualifiers): ... this. New.
38212 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38214 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38215 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38216 (MRRCI, mrrc, MRRC): New.
38217 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38218 VUNSPEC_MRRC2): New.
38220 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38222 * config/arm/arm.md (<mcr>): New.
38224 * config/arm/arm.c (arm_coproc_builtin_available): Add
38225 support for mcr, mrc, mcr2 and mrc2.
38226 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38227 (arm_mcr_qualifiers): ... this. New.
38228 (MRC_QUALIFIERS): Define to ...
38229 (arm_mrc_qualifiers): ... this. New.
38230 (MCR_QUALIFIERS): Define to ...
38231 (arm_mcr_qualifiers): ... this. New.
38232 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38234 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38235 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38236 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38237 VUNSPEC_MRC2): New.
38239 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38241 * config/arm/arm.md (*ldc): New.
38245 * config/arm/arm.c (arm_coproc_builtin_available): Add
38246 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38247 (arm_coproc_ldc_stc_legitimate_address): New.
38248 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38249 'qualifier_const_pointer'.
38250 (LDC_QUALIFIERS): Define to...
38251 (arm_ldc_qualifiers): ... this. New.
38252 (STC_QUALIFIERS): Define to...
38253 (arm_stc_qualifiers): ... this. New.
38254 * config/arm/arm-protos.h
38255 (arm_coproc_ldc_stc_legitimate_address): New.
38256 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38257 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38258 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38259 stc2, stcl, stc2l): New.
38260 * config/arm/constraints.md (Uz): New.
38261 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38262 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38263 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38264 VUNSPEC_STC2L): New.
38266 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38268 * config/arm/arm.md (<cdp>): New.
38269 * config/arm/arm.c (neon_const_bounds): Rename this ...
38270 (arm_const_bounds): ... this.
38271 (arm_coproc_builtin_available): New.
38272 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38273 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38274 (CDP_QUALIFIERS): Define to...
38275 (arm_cdp_qualifiers): ... this. New.
38277 (arm_expand_builtin_args): Add case for 6 arguments.
38278 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38279 (arm_const_bounds): ... this.
38280 (arm_coproc_builtin_available): New.
38281 * config/arm/arm_acle.h (__arm_cdp): New.
38283 * config/arm/arm_acle_builtins.def (cdp): New.
38285 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38286 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38287 'arm_const_bounds'.
38288 * config/arm/types.md (coproc): New.
38289 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38290 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38291 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38292 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38294 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38296 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38297 (UBINOP_QUALIFIERS): New.
38299 (acle_builtin_data): New. Change comment.
38300 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38301 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38302 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38303 arm_acle_builtins.def.
38304 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38305 (arm_init_acle_builtins): New.
38306 (CRC32_BUILTIN): Remove.
38307 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38308 crc32cb, crc32ch and crc32cw.
38309 (arm_init_crc32_builtins): Remove.
38310 (arm_init_builtins): Use arm_init_acle_builtins rather
38311 than arm_init_crc32_builtins.
38312 (arm_expand_acle_builtin): New.
38313 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38314 * config/arm/arm_acle_builtins.def: New.
38316 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38318 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38319 (arm_builtin_datum): ... this.
38320 (arm_init_neon_builtin): Rename to ...
38321 (arm_init_builtin): ... this. Add a new parameters PREFIX
38322 and USE_SIG_IN_NAME.
38323 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38324 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38325 'arm_builtin_datum'.
38326 (arm_init_vfp_builtins): Likewise.
38327 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38328 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38329 (arm_expand_neon_args): Rename to ...
38330 (arm_expand_builtin_args): ... this. Rename builtin_arg
38331 enum values and differentiate between ARG_BUILTIN_MEMORY
38332 and ARG_BUILTIN_NEON_MEMORY.
38333 (arm_expand_neon_builtin_1): Rename to ...
38334 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38335 values, arm_expand_builtin_args and add bool parameter NEON.
38336 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38337 (arm_expand_vfp_builtin): Likewise.
38338 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38340 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38342 PR middle-end/77484
38343 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38344 * predict.c (tree_estimate_probability_bb): Reverse direction of
38345 polymorphic call predictor.
38347 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38349 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38350 (determine_pass_name_match): ...this new function and...
38351 (should_skip_pass_p): ...this new function.
38353 2017-01-06 Nathan Sidwell <nathan@acm.org>
38355 * ipa-visibility.c (function_and_variable_visibility): Reformat
38356 comments and long lines. Remove extrneous if.
38357 * symtab.c (symtab_node::make_decl_local): Fix code format.
38358 (symtab_node::set_section_for_node): Fix comment typo.
38360 2017-01-06 Martin Liska <mliska@suse.cz>
38363 * lra-constraints.c: Rename invariant to lra_invariant.
38364 * predict.c (set_even_probabilities): Initialize e to NULL.
38366 2017-01-05 Martin Sebor <msebor@redhat.com>
38368 PR tree-optimization/78910
38369 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38370 (format_integer): Correct off-by-one error in the handling
38371 of precision with negative numbers in signed conversions..
38373 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38375 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38377 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38379 PR tree-optimization/71016
38380 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38381 factor_out_conditional_conversion. Formatting fix.
38382 (factor_out_conditional_conversion): Add cond_stmt argument.
38383 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38384 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38387 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38389 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38390 read-rtl-function.o, and selftest-rtl.o.
38391 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38392 (selftest::aarch64_test_loading_full_dump): New function.
38393 (selftest::aarch64_run_selftests): New function.
38394 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38395 selftest::aarch64_run_selftests.
38396 * config/i386/i386.c
38397 (selftest::ix86_test_loading_dump_fragment_1): New function.
38398 (selftest::ix86_test_loading_call_insn): New function.
38399 (selftest::ix86_test_loading_full_dump): New function.
38400 (selftest::ix86_test_loading_unspec): New function.
38401 (selftest::ix86_run_selftests): Call the new functions.
38402 * emit-rtl.c (maybe_set_max_label_num): New function.
38403 * emit-rtl.h (maybe_set_max_label_num): New decl.
38404 * function.c (instantiate_decls): Guard call to
38405 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38406 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38408 * gensupport.c (gen_reader::gen_reader): Pass "false"
38409 for new "compact" param of rtx_reader.
38410 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38411 rather than an empty string for NULL strings.
38412 * read-md.c: Potentially include config.h rather than bconfig.h.
38413 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38414 (have_error): New global, copied from errors.c.
38415 (md_reader::read_name): Rename to...
38416 (md_reader::read_name_1): ...this, adding "out_loc" param,
38417 and converting "missing name or number" to returning false, rather
38419 (md_reader::read_name): Reimplement in terms of read_name_1.
38420 (md_reader::read_name_or_nil): New function.
38421 (md_reader::read_string): Handle "(nil)" by returning NULL.
38422 (md_reader::md_reader): Add new param "compact".
38423 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38424 (md_reader::read_file): New method.
38425 * read-md.h (md_reader::md_reader): Add new param "compact".
38426 (md_reader::read_file): New method.
38427 (md_reader::is_compact): New accessor.
38428 (md_reader::read_name): Convert return type from void to file_location.
38429 (md_reader::read_name_or_nil): New decl.
38430 (md_reader::read_name_1): New decl.
38431 (md_reader::m_compact): New field.
38432 (noop_reader::noop_reader): Pass "false" for new "compact" param
38434 (rtx_reader::rtx_reader): Add new "compact" param.
38435 (rtx_reader::read_rtx_operand): Make virtual and convert return
38436 type from void to rtx.
38437 (rtx_reader::read_until): New decl.
38438 (rtx_reader::handle_any_trailing_information): New virtual function.
38439 (rtx_reader::postprocess): New virtual function.
38440 (rtx_reader::finalize_string): New virtual function.
38441 (rtx_reader::m_in_call_function_usage): New field.
38442 (rtx_reader::m_reuse_rtx_by_id): New field.
38443 * read-rtl-function.c: New file.
38444 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38445 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38446 (selftest::verify_three_block_rtl_cfg): New decl.
38447 * read-rtl-function.h: New file.
38448 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38449 For host, include function.h, memmodel.h, and emit-rtl.h.
38450 (one_time_initialization): New function.
38451 (struct compact_insn_name): New struct.
38452 (compact_insn_names): New array.
38453 (find_code): Handle insn codes in compact dumps.
38454 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38455 (bind_subst_iter_and_attr): Likewise.
38456 (add_condition_to_string): Likewise.
38457 (add_condition_to_rtx): Likewise.
38458 (apply_attribute_uses): Likewise.
38459 (add_current_iterators): Likewise.
38460 (apply_iterators): Likewise.
38461 (initialize_iterators): Guard usage of apply_subst_iterator with
38462 #ifdef GENERATOR_FILE.
38463 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38464 (md_reader::read_mapping): Likewise.
38465 (add_define_attr_for_define_subst): Likewise.
38466 (add_define_subst_attr): Likewise.
38467 (read_subst_mapping): Likewise.
38468 (check_code_iterator): Likewise.
38469 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38471 (one_time_initialization): New function.
38472 (rtx_reader::read_until): New method.
38473 (read_flags): New function.
38474 (parse_reg_note_name): New function.
38475 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38476 Handle reuse_rtx ids.
38477 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38478 Add parsing support for RTL dumps, mirroring the special-cases in
38479 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38480 values, and calling handle_any_trailing_information.
38481 (rtx_reader::read_rtx_operand): Convert return type from void
38482 to rtx, returning return_rtx. Handle case 'e'. Call
38483 finalize_string on XSTR and XTMPL fields.
38484 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38485 "(nil)" values were omitted. Call the postprocess vfunc on the
38487 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38488 class ctor. Initialize m_in_call_function_usage. Call
38489 one_time_initialization.
38490 * rtl-tests.c (selftest::test_uncond_jump): Call
38491 set_new_first_and_last_insn.
38492 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38493 * selftest-rtl.c: New file.
38494 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38495 (selftest::get_insn_by_uid): New decl.
38496 * selftest-run-tests.c (selftest::run_tests): Call
38497 read_rtl_function_c_tests.
38498 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38499 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38502 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38504 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38505 operands in a special way. Assert that pos+len <= mode precision.
38507 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38509 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38510 3 argument Alias with unlimited for the negative form.
38511 (fno-vect-cost-model): Removed.
38513 2017-01-05 Martin Liska <mliska@suse.cz>
38515 * hsa-gen.c (gen_hsa_divmod): New function.
38516 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38518 2017-01-05 Martin Liska <mliska@suse.cz>
38521 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38524 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38526 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38527 small constant length operands.
38529 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38531 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38532 between loop iterations.
38534 2017-01-05 Martin Liska <mliska@suse.cz>
38537 * gimplify.c (gimplify_decl_expr): Compare to
38538 asan_poisoned_variables instread of checking flags.
38539 (gimplify_target_expr): Likewise.
38540 (gimplify_expr): Likewise.
38541 (gimplify_function_tree): Conditionally initialize
38542 asan_poisoned_variables.
38544 2017-01-04 Jeff Law <law@redhat.com>
38546 PR tree-optimizatin/78812
38547 * rtl.h (contains_mem_rtx_p): Prototype.
38548 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38549 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38550 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38551 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38552 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38554 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38556 * input.c (assert_char_at_range): Default-initialize actual_range.
38558 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38560 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38561 to match the caller.
38563 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38565 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38566 insns after final jump in test to emit dummy move.
38568 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38570 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38571 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38573 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38575 * multiple_target.c (create_dispatcher_calls): Init e_next.
38576 * tree-ssa-loop-split.c (split_loop): Init border.
38577 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38580 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38585 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38586 (altivec_register_operand): Do not return true if the operand
38587 contains a SUBREG mixing SImode and SFmode.
38588 (vsx_register_operand): Likewise.
38589 (vsx_reg_sfsubreg_ok): New predicate.
38590 (vfloat_operand): Do not return true if the operand contains a
38591 SUBREG mixing SImode and SFmode.
38592 (vint_operand): Likewise.
38593 (vlogical_operand): Likewise.
38594 (gpc_reg_operand): Likewise.
38595 (int_reg_operand): Likewise.
38596 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38597 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38598 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38600 (rs6000_emit_move_si_sf_subreg): New helper function.
38601 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38602 fixup SUBREGs involving SImode and SFmode.
38603 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38604 numbers for the new peephole2 optimization.
38605 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38606 the GLIBC math library that do AND/IOR/XOR operations on single
38607 precision floating point.
38608 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38609 target macros to say whether we need to avoid SUBREGs mixing
38611 (TARGET_ALLOW_SF_SUBREG): Likewise.
38612 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38613 (UNSPEC_SI_FROM_SF): Likewise.
38614 (iorxor): Change spacing.
38615 (and_ior_xor): New iterator for AND, IOR, and XOR.
38616 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38617 (movdi_from_sf_zero_ext): Likewise.
38618 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38619 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38620 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38621 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38622 (fms<mode>4): Likewise.
38623 (fnma<mode>4): Likewise.
38624 (fnms<mode>4): Likewise.
38625 (nfma<mode>4): Likewise.
38626 (nfms<mode>4): Likewise.
38628 2017-01-04 Marek Polacek <polacek@redhat.com>
38631 * doc/invoke.texi: Document -Wpointer-compare.
38633 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38635 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38638 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38639 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38642 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38645 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38646 documentation of the powerpc_popcntb_ok attribute.
38647 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38648 code to issue warning messages if a requested CPU configuration is
38649 not supported by the binary (assembler and loader) toolchain.
38650 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38651 made to define a built-in function that has been disabled.
38652 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38653 made to define a built-in function that has been disabled.
38654 (altivec_init_builtins): Add comment explaining why definition
38655 of the DST built-in functions is not preceded by an assertion
38656 check. Add assertions to prevent ICE if attempts are made to
38657 define an altivec predicate or an abs* built-in function that has
38659 (htm_init_builtins): Add comment explaining why definition of the
38660 htm built-in functions is not preceded by an assertion check.
38662 2017-01-04 Jeff Law <law@redhat.com>
38664 PR tree-optimizatin/67955
38665 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38666 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38667 the points-to solution does not include pt_null. Use DECL_PT_UID
38670 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38672 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38673 Use gen_int_mode instead of gen_lopwart for const_int operands.
38675 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38677 PR tree-optimization/71563
38678 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38679 out of range value - has low bits known to be zero.
38681 2017-01-04 Alan Modra <amodra@gmail.com>
38683 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38684 * configure: Regenerate.
38685 * config.in: Regenerate.
38687 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38690 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38691 a substring of the message, but strcmp with the whole message. Ifdef
38692 ENABLE_NLS, translate the message first using dgettext.
38694 2017-01-03 Jeff Law <law@redhat.com>
38696 PR tree-optimizatin/78856
38697 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38698 (mark_threaded_blocks): Remove code to truncate thread paths that
38699 cross multiple loop headers. Instead invalidate the cached loop
38700 iteration information and handle case of a thread path walking
38701 into an irreducible region.
38703 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38706 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38707 assertions. Add support for doing the signbit if the IEEE 128-bit
38708 floating point value is in a GPR.
38709 * config/rs6000/rs6000.md (Fsignbit): Delete.
38710 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38711 Update the length attribute if the value is in a GPR.
38712 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38713 the sign or zero extension instruction, since the value is always 0/1.
38714 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38717 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38718 extracting SImode to a GPR register so that we can generate a
38719 store, limit the vector to be in a traditional Altivec register
38720 for the vextuwrx instruction.
38722 2017-01-03 Ian Lance Taylor <iant@google.com>
38724 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38726 2017-01-03 Martin Sebor <msebor@redhat.com>
38728 PR tree-optimization/78696
38729 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38730 precision. Use MPFR for %f for greater fidelity. Correct handling
38732 (pass_sprintf_length::compute_format_length): Set width and precision
38733 specified by asrerisk to void_node for vararg functions.
38734 (try_substitute_return_value): Adjust dump output.
38736 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38738 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38740 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38742 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38743 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38744 -mlra/-mno-lra was passed to the compiler.
38746 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38748 PR rtl-optimization/65618
38749 * emit-rtl.c (try_split): Move initialization of "before" and
38750 "after" to just before the call to emit_insn_after_setloc.
38752 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38754 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38756 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38758 * dwarf2out.c (gen_enumeration_type_die): When
38759 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38761 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38763 PR tree-optimization/78965
38764 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38765 Change first argument from const call_info & to call_info &. For %n
38766 set info.nowrite to false.
38768 PR middle-end/78901
38769 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38770 possibly throwing calls.
38772 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38773 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38774 and fns handling, rather than in a separate case SSA_NAME.
38776 2017-01-02 Jeff Law <law@redhat.com>
38778 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38779 fixes for first_period and second_period variables.
38781 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38784 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38785 (*insvqi_1): New insn pattern.
38786 (*insvqi_1_mem_rex64): Ditto.
38787 (*insvqi_2): Ditto.
38788 (*insvqi_3): Rename from *insvqi.
38790 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38792 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38794 * doc/cfg.texi (Edges): Remove reference to Java.
38795 (Maintaining the CFG): Ditto.
38797 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38799 PR middle-end/77674
38800 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38801 transparent aliases.
38803 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38805 PR middle-end/77484
38806 * predict.def (PRED_CALL): Update hitrate.
38807 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38808 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38809 into direct/indirect/polymorphic variants.
38811 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38813 Update copyright years.
38815 * gcc.c (process_command): Update copyright notice dates.
38816 * gcov-dump.c (print_version): Ditto.
38817 * gcov.c (print_version): Ditto.
38818 * gcov-tool.c (print_version): Ditto.
38819 * gengtype.c (create_file): Ditto.
38820 * doc/cpp.texi: Bump @copying's copyright year.
38821 * doc/cppinternals.texi: Ditto.
38822 * doc/gcc.texi: Ditto.
38823 * doc/gccint.texi: Ditto.
38824 * doc/gcov.texi: Ditto.
38825 * doc/install.texi: Ditto.
38826 * doc/invoke.texi: Ditto.
38828 Copyright (C) 2017 Free Software Foundation, Inc.
38830 Copying and distribution of this file, with or without modification,
38831 are permitted in any medium without royalty provided the copyright
38832 notice and this notice are preserved.