1 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * poly-int.h: New file.
6 * poly-int-types.h: Likewise.
7 * coretypes.h: Include them.
8 (POLY_INT_CONVERSION): Define.
9 * target.def (estimated_poly_value): New hook.
10 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
11 * doc/tm.texi: Regenerate.
12 * doc/poly-int.texi: New file.
13 * doc/gccint.texi: Include it.
14 * doc/rtl.texi: Describe restrictions on subreg modes.
15 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
16 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
17 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
18 * targhooks.h (default_estimated_poly_value): Declare.
19 * targhooks.c (default_estimated_poly_value): New function.
20 * target.h (estimated_poly_value): Likewise.
21 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
22 (wi::unary_traits): Delete.
23 (wi::binary_traits::signed_shift_result_type): Define for
24 offset_int << HOST_WIDE_INT, etc.
25 (generic_wide_int::operator <<=): Define for all types and use
26 wi::lshift instead of <<.
27 (wi::hwi_with_prec): Add a default constructor.
28 (wi::ints_for): New class.
29 (operator <<): Define for all wide-int types.
30 (operator /): New function.
31 (operator %): Likewise.
32 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
33 (ASSERT_MAYBE_NE_AT): New macros.
35 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
36 Dominik Vogt <vogt@linux.vnet.ibm.com>
39 * emit-rtl.c (init_emit): Remove ??? comment.
40 * explow.c (get_dynamic_stack_size): Take known alignment of stack
41 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
42 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
43 alignment of 3 virtual registers to BITS_PER_WORD.
45 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
47 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
49 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
50 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
52 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
55 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
57 2017-12-13 Jakub Jelinek <jakub@redhat.com>
59 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
60 with the exception of debug begin stmt markers appear before
64 * final.c (rest_of_handle_final): Call variable_tracking_main only
65 if !flag_var_tracking.
67 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
68 Jakub Jelinek <jakub@redhat.com>
72 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
73 labels debug stmts that can only appear after labels.
75 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
77 PR rtl-optimization/82398
78 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
79 EXPR_USEFULNESS in priority comparison.
81 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
83 PR rtl-optimization/83393
84 * combine.c (move_deaths): If reg_stat points to a too new insn in
85 last_death, do not use it: find the proper insn instead.
87 2017-12-12 Jeff Law <law@redhat.com>
89 PR tree-optimization/83298
90 PR tree-optimization/83362
91 PR tree-optimization/83383
92 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
93 push_value_range a public interface. Add new argument to
94 record_ranges_from_stmt.
95 * gimple-ssa-evrp-analyze.c
96 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
97 Update comments. Handle recording temporary equivalences.
98 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
99 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
100 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
101 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
102 gimple-ssa-evrp-analyze.h.
103 (record_temporary_equivalences_from_phis): Add new argument. When
104 the PHI arg is an SSA_NAME, set the result's range to the range
106 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
108 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
109 Pass it down to children as needed.
110 (thread_outgoing_edges): Likewise.
111 (thread_across_edge): Likewise. Push/pop range state as needed.
112 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
114 2017-12-12 Julia Koval <julia.koval@intel.com>
116 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
117 (PTA_CANNONLAKE): Remove PTA_CLWB.
119 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
122 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
123 (vcondv2div2df): Likewise.
124 (vconduv2dfv2di): Likewise.
126 2017-12-12 Jakub Jelinek <jakub@redhat.com>
128 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
129 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
130 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
131 ATTR_MATHFN_FPROUNDING_ERRNO.
133 2017-12-12 Richard Biener <rguenther@suse.de>
135 PR tree-optimization/83385
136 * graphite-scop-detection.c (get_order, order): Remove.
137 (bb_to_rpo): New global.
139 (build_scops): Sort pbbs in RPO order.
141 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
143 * combine.c (simplify_set): Do not transform subregs to zero_extends
144 if the destination mode is a vector mode.
146 2017-12-12 Jakub Jelinek <jakub@redhat.com>
148 PR tree-optimization/83359
149 * tree-cfg.h (fold_loop_internal_call): Declare.
150 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
151 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
152 (find_loop_dist_alias): New function.
153 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
154 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
157 PR tree-optimization/80631
158 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
159 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
160 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
161 hardcoding zero as the value if COND_EXPR is never true. For
162 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
163 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
164 hardcoding MAX_EXPR as the reduction operation.
165 (is_nonwrapping_integer_induction): Allow negative step.
166 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
167 vect_create_epilog_for_reduction, if no value is suitable, don't
168 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
170 2017-12-12 Richard Biener <rguenther@suse.de>
172 PR tree-optimization/81889
173 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
174 range info from the non-wrapping IV instead of just the range
177 2017-12-12 Julia Koval <julia.koval@intel.com>
179 * config.gcc: Add vaesintrin.h.
180 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
182 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
183 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
185 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
186 * config/i386/immintrin.h: Include vaesintrin.h.
187 * config/i386/sse.md (vaesdec_<mode>): New pattern.
188 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
189 _mm_aesdec_epi128): New intrinsics.
191 2017-12-12 Julia Koval <julia.koval@intel.com>
193 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
194 OPTION_MASK_ISA_VAES_UNSET): New.
195 (ix86_handle_option): Handle -mvaes.
196 * config/i386/cpuid.h: Define bit_VAES.
197 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
198 * config/i386/i386-c.c (__VAES__): New.
199 * config/i386/i386.c (ix86_target_string): Add -mvaes.
200 (ix86_valid_target_attribute_inner_p): Ditto.
201 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
202 * config/i386/i386.opt: Add -mvaes.
203 * doc/invoke.texi: Ditto.
205 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
207 * debug.h (gcc_debug_hooks): Add inline_entry.
208 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
209 * debug.c (do_nothing_debug_hooks): Likewise.
210 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
211 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
212 (dwarf2_lineno_debug_hooks): Likewise.
214 * common.opt (gstatement-frontiers): New, setting
215 debug_nonbind_markers_p.
216 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
217 * toplev.c (process_options): Autodetect value for debug statement
219 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
220 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
222 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
223 markers. Integrate source bind into debug stmt expand loop.
224 (pass_expand::execute): Check debug marker limit. Avoid deep
225 TER and expand debug locations for debug bind insns only.
226 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
228 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
229 * final.c (reemit_insn_block_notes): Take current block from
230 nonbind markers. Declare note where it's first set.
231 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
232 begin stmt markers if enabled.
233 (notice_source_line): Handle nonbind markers. Fail if their
234 location is unknown or that of builtins.
235 (rest_of_handle_final): Convert begin stmt markers to notes if
236 var-tracking didn't run.
237 (rest_of_clean_state): Skip begin stmt markers.
238 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
240 * function.c (allocate_struct_function): Set begin_stmt_markers.
241 * function.h (struct function): Add debug_marker_count counter
242 and debug_nonbind_markers flag.
243 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
244 * gimple-low.c (lower_function_body): Adjust
245 debug_nonbind_markers.
246 (lower_stmt): Drop or skip gimple debug stmts.
247 (lower_try_catch): Skip debug stmts.
248 * gimple.c (gimple_build_debug_begin_stmt): New.
249 (gimple_copy): Increment debug_marker_count if copying one.
250 * gimple.h (gimple_build_debug_begin_stmt): Declare.
251 * gimplify.c (rexpr_location): New.
252 (rexpr_has_location): New.
253 (warn_switch_unreachable_r): Handle gimple debug stmts.
254 (shortcut_cond_r): Call expr_location.
256 (find_goto_label): New.
257 (shortcut_cond_expr): Call expr_has_location, expr_location, and
259 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
261 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
262 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
263 (LANG_HOOKS_INITIALIZER): ... this.
264 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
265 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
267 (update_ebb_live_info): Skip debug insn markers.
268 * lra.c (debug_insn_static_data): Rename to...
269 (debug_bind_static_data): ... this.
270 (debug_marker_static_data): New.
271 (lra_set_insn_recog_data): Select one of the above depending
273 (lra_update_isn_regno_info): Don't assume debug insns have
275 (push_insns): Skip debug insns.
276 * lto-streamer-in.c (input_function): Drop debug stmts
277 depending on active options. Adjust debug_nonbind_markers.
278 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
279 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
280 begin stmt marker notes.
281 (print_insn): Likewise.
282 * recog.c (extract_insn): Recognize rtl for debug markers.
283 * rtl.def (DEBUG_MARKER): New.
284 * tree-inline.c: Include params.h.
285 (remap_gimple_stmt): Handle nonbind markers.
286 (maybe_move_debug_stmts_to_successors): Likewise.
287 (copy_debug_stmt): Likewise.
288 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
289 markers regardless of no side effects.
290 (tsi_link_before): Don't update container's side effects when adding
292 (tsi_link_after): Likewise.
293 (expr_first): Skip begin stmt markers.
294 (expr_last): Likewise.
295 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
296 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
297 Disregard nonbind markers.
298 * tree.c (make_node_stat): Don't set side effects for begin stmt
300 (build1_stat): Likewise.
301 * tree.def (DEBUG_BEGIN_STMT): New.
302 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
303 * var-tracking.c (delete_debug_insns): Renamed to...
304 (delete_vta_debug_insns): ... this.
305 (reemit_marker_as_note): New.
306 (vt_initialize): Reemit markers.
307 (delete_vta_debug_insns): Likewise.
308 (vt_debug_insns_local): Reemit or delete markers.
309 (variable_tracking_main_1): Likewise.
310 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
311 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
312 (gimple_debug_nonbind_marker_p): New.
313 (gimple_build_debug_bind): Adjust.
314 (gimple_build_debug_begin_stmt): New.
315 * doc/invoke.texi (max-debug-marker-count): New param.
316 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
317 (debug_parameter_ref, debug_marker): New.
318 (NOTE_INSN_BEGIN_STMT): New.
319 (DEBUG_INSN): Describe begin stmt markers.
321 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
322 without debug insns, we wouldn't, but clean up debug insns
323 after a control flow insn nevertheless.
325 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
326 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
327 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
328 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
329 (rtl_tidy_fallthru_edge): Likewise.
330 (rtl_verify_fallthru): Likewise.
331 (rtl_verify_bb_layout): Likewise.
332 (skip_insns_after_block): Likewise.
333 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
334 * dwarf2out.c: Include print-rtl.h.
335 (dwarf2out_next_real_insn): New.
336 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
337 Dump debug binds in asm comments.
338 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
339 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
340 callers to use gsi_start_nondebug_bb instead.
341 (gsi_after_labels): Skip gimple debug stmts.
342 (gsi_start_nondebug): New.
343 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
344 (proper_loop_form_for_interchange): Adjust.
345 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
346 * gimple.h (gimple_seq_last_nondebug_stmt): New.
347 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
348 (collect_fallthrough_labels): Likewise.
349 (should_warn_for_implicit_fallthrough): Likewise.
350 (warn_implicit_fallthrough_r): Likewise.
351 (expand_FALLTHROUGH_r): Likewise.
352 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
353 (graphite_copy_stmts_from_block): Skip nonbind markers.
354 * haifa-sched.c (sched_extend_bb): Skip debug insns.
355 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
356 * jump.c (clean_barriers): Skip debug insns.
357 * omp-expand.c (expand_parallel_call): Skip debug insns.
358 (expand_task_call): Likewise.
359 (remove_exit_barrier): Likewise.
360 (expand_omp_taskreg): Likewise.
361 (expand_omp_for_init_counts): Likewise.
362 (expand_omp_for_generic): Likewise.
363 (expand_omp_for_static_nochunk): Likewise.
364 (expand_omp_for_static_chunk): Likewise.
365 (expand_omp_simd): Likewise.
366 (expand_omp_taskloop_for_outer): Likewise.
367 (expand_omp_taskloop_for_inner): Likewise.
368 (expand_oacc_for): Likewise.
369 (expand_omp_sections): Likewise.
370 (expand_omp_single): Likewise.
371 (expand_omp_synch): Likewise.
372 (expand_omp_atomic_load): Likewise.
373 (expand_omp_atomic_store): Likewise.
374 (expand_omp_atomic_fetch_op): Likewise.
375 (expand_omp_atomic_pipeline): Likewise.
376 (expand_omp_atomic_mutex): Likewise.
377 (expand_omp_target): Likewise.
378 (grid_expand_omp_for_loop): Likewise.
379 (grid_expand_target_grid_body): Likewise.
380 (build_omp_regions_1): Likewise.
381 * omp-low.c (check_combined_parallel): Skip debug stmts.
382 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
383 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
384 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
386 * tree-cfg.c (make_blobs_1): Skip debug stmts.
387 (make_edges): Likewise.
388 (cleanup_dead_labels): Likewise.
389 (gimple_can_merge_blocks_p): Likewise.
390 (stmt_starts_bb_p): Likewise.
391 (gimple_block_label): Likewise.
392 (gimple_redirect_edge_and_branch): Likewise.
393 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
395 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
397 * tree-pretty-print (print_declaration): Omit initializer in slim
399 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
401 (eliminate_unnecessary_stmts): Stabilize block removal order.
402 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
403 * var-tracking.c (get_first_insn): New.
404 (vt_emit_notes): Call it.
405 (vt_initialize): Walk any insns before the first BB.
406 (delete_debug_insns): Likewise.
408 * gimple.h (enum gimple_debug_subcode): Add
409 GIMPLE_DEBUG_BEGIN_STMT.
410 (gimple_debug_begin_stmt_p): New.
411 (gimple_debug_nonbind_marker_p): New.
412 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
413 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
414 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
415 * insn-notes.def (BEGIN_STMT): New.
416 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
417 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
418 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
419 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
420 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
421 (INSN_DEBUG_MARKER_KIND): New.
422 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
423 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
424 (INSN_VAR_LOCATION_PTR): New.
425 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
427 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
428 deps only if debug bind insns are enabled.
429 (pass_expand::execute): Avoid deep TER and expand
430 debug locations for debug bind insns only.
431 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
432 debug stmts special handling down to debug bind stmts.
433 * combine.c (try_combine): Narrow debug insns special handling
434 down to debug bind insns.
435 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
436 Narrow debug insns preexisting special handling down to debug
438 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
439 handling down to debug bind insns.
440 * function.c (instantiate_virtual_regs): Skip debug markers,
441 adjust handling of debug binds.
442 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
443 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
444 * haifa-sched.c (schedule_insn): Narrow special handling of debug
445 insns to debug bind insns.
446 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
447 special handling of debug stmts to debug bind stmts.
448 * ipa-split.c (split_function): Likewise.
449 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
450 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
452 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
453 * regrename.c (build_def_use): Likewise.
454 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
455 (pass_cprop_hardreg): Narrow special casing of debug insns to
457 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
458 * reload1.c (reload): Likewise.
459 * sese.c (sese_insert_phis_for_liveouts): Narrow special
460 casing of debug stmts to debug bind stmts.
461 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
462 * ssa-iterators.h (num_imm_uses): Likewise.
463 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
464 debug stmts to debug bind stmts.
465 * tree-inline.c (tree_function_versioning): Narrow special casing
466 of debug stmts to debug bind stmts.
467 * tree-loop-distribution.c (generate_loops_for_partition):
468 Narrow special casing of debug stmts to debug bind stmts.
469 * tree-sra.c (analyze_access_subtree): Narrow special casing
470 of debug stmts to debug bind stmts.
471 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
472 stmts to debug bind stmts.
473 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
474 casing of debug stmts to debug bind stmts.
475 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
476 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
477 casing of debug stmts to debug bind stmts.
478 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
480 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
481 debug stmts to debug bind stmts.
482 (gimple_replace_ssa_lhs): Likewise.
483 (insert_debug_temp_for_var_def): Likewise.
484 (insert_debug_temps_for_defs): Likewise.
485 (reset_debug_uses): Likewise.
486 * tree-ssanames.c (release_ssa_name_fn): Likewise.
487 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
488 (adjust_debug_stmts): Likewise.
489 (adjust_phi_and_debug_stmts): Likewise.
490 (vect_do_peeling): Likewise.
491 * tree-vect-loop.c (vect_transform_loop): Likewise.
492 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
493 * var-tracking.c (adjust_mems): Narrow special casing of debug
494 insns to debug bind insns.
495 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
496 (compute_bb_dataflow, vt_find_locations): Likewise.
497 (vt_expand_loc, emit_notes_for_changes): Likewise.
498 (vt_init_cfa_base): Likewise.
499 (vt_emit_notes): Likewise.
500 (vt_initialize): Likewise.
501 (vt_finalize): Likewise.
503 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
504 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
505 (next_nonnote_nondebug_insn_bb): New.
506 (prev_nonnote_nondebug_insn_bb): New.
507 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
508 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
509 (prev_nonnote_nondebug_insn_bb): Declare.
510 (next_nonnote_nondebug_insn_bb): Declare.
511 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
512 * cfgrtl.c (get_last_bb_insn): Likewise.
513 * lra.c (push_insns): Likewise.
515 2017-12-11 David Malcolm <dmalcolm@redhat.com>
518 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
520 * selftest.c (selftest::test_runner::test_runner): New ctor.
521 (selftest::test_runner::~test_runner): New dtor.
522 * selftest.h (class selftest::test_runner): New class.
524 2017-12-11 Carl Love <cel@us.ibm.com>
526 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
527 vec_extract_fp32_from_shortl]): Add #defines.
528 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
529 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
530 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
531 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
532 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
533 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
534 * doc/extend.texi: Add documentation for the added builtins.
536 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
538 PR rtl-optimization/80693
539 PR rtl-optimization/81019
540 PR rtl-optimization/81020
541 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
542 are not mentioned in i3. Place the REG_UNUSED note on i2,
543 possibly modified to REG_DEAD, if it did not originate in i3.
545 2017-12-11 Jakub Jelinek <jakub@redhat.com>
547 * recog.c (store_data_bypass_p_1): New function.
548 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
549 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
551 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
553 PR rtl-optimization/83361
554 * ifcvt.c (if_convert): Call fixup_partitions.
556 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
558 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
559 early folding of splat_u{8,16,32}.
561 2017-12-11 Jakub Jelinek <jakub@redhat.com>
563 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
564 output_operand_lossage first argument with capital letter.
565 (aarch64_override_options): Don't start error and sorry first argument
568 2017-12-11 Andi Kleen <ak@linux.intel.com>
570 PR gcov-profile/83355
571 * auto-profile.c (string_table::get_index_by_decl): Don't
572 recurse when abstract origin points to itself.
574 2017-12-11 Bin Cheng <bin.cheng@arm.com>
576 PR tree-optimization/83320
577 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
578 (prune_datarefs_not_in_loop): Ditto.
580 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
582 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
584 2017-12-10 Jakub Jelinek <jakub@redhat.com>
586 PR tree-optimization/83337
587 * gimple-loop-interchange.cc (compute_access_stride): Handle
588 bitfield DRs properly.
590 2017-12-09 Jakub Jelinek <jakub@redhat.com>
592 PR tree-optimization/83338
593 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
594 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
595 vector of unsigned integers to vector of signed integers.
597 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
599 PR rtl-optimization/83317
600 * lra-constraints.c (process_address_1): Add insn code check.
602 2017-12-08 Michael Matz <matz@suse.de>
604 Fix PR tree-optimization/83323
605 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
606 head-controlled loops and loop BBs.
607 * common.opt (funroll-and-jam): Remove, instead ...
608 (floop-unroll-and-jam): ... reuse this option.
609 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
610 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
611 (-floop-unroll-and-jam): ... this option.
613 2017-12-08 Jakub Jelinek <jakub@redhat.com>
615 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
616 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
617 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
619 2017-12-08 Richard Biener <rguenther@suse.de>
622 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
623 handle accesses outside of zero-sized vars.
625 2017-12-08 Martin Jambor <mjambor@suse.cz>
627 PR tree-optimization/83141
628 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
629 test for MEM_REFs implicitely changing types with padding. Remove
631 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
633 2017-12-08 Jakub Jelinek <jakub@redhat.com>
635 * config/arc/arc.c (arc_attribute_table): Add exclusions to
637 * config/avr/avr.c (avr_attribute_table): Likewise.
638 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
639 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
640 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
641 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
642 of struct attribute_spec.
643 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
645 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
648 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
650 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
653 * profile-count.c (profile_count::from_gcov_type): Move from
654 profile-count.h; handle overflow.
655 * profile-count.h (profile_count::from_gcov_type): Move offline.
657 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
659 PR rtl-optimization/83304
660 * combine.c (move_deaths): If we do not know where a register died,
663 2017-12-08 Richard Biener <rguenther@suse.de>
665 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
666 Provide -fopt-info-loop feedback when we interchange in a nest.
668 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
670 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
671 for armv6 ARM CPU IDs.
673 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
675 * common/config/arm/arm-common.c: Include <algorithm>.
676 (INCLUDE_VECTOR): Define.
677 (compare_opt_names): New function.
678 (arm_rewrite_selected_arch): Only strip out extensions that can be
679 expressed through -mfpu. Sort the remaining extensions
682 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
684 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
685 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
686 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
687 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
689 2017-06-08 Tristan Gingold <gindold@adacore.com>
692 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
693 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
695 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
697 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
698 of struct attribute_spec.
700 2017-12-08 Julia Koval <julia.koval@intel.com>
702 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
703 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
704 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
705 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
706 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
707 _mm_maskz_dpwssds_epi32): Ditto.
709 2017-12-08 Richard Biener <rguenther@suse.de>
711 PR tree-optimization/81303
712 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
713 conditions try to create a comparison vector type matching
714 the data vector type.
715 (vectorizable_condition): Adjust.
716 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
717 Leave invariant conditions alone in case we can vectorize those.
719 2017-12-08 Julia Koval <julia.koval@intel.com>
721 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
722 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
723 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
724 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
725 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
727 2017-12-08 Julia Koval <julia.koval@intel.com>
729 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
730 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
731 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
732 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
733 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
734 _mm_maskz_dpbusds_epi32): New intrinsics.
736 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
738 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
740 (ld<bh>io_signed): Likewise.
741 (st<bhw>io): Likewise.
742 * config/nios2/predicates.md (ldstio_memory_operand): Allow
743 SMALL_INT12 constant integer operand.
745 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
747 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
749 (nios2_large_constant_allowed): ...this. Adjust uses.
750 (nios2_plus_symbolic_constant_p): Rename to...
751 (nios2_plus_large_constant_p): ...this. Adjust uses.
752 (nios2_legitimate_address_p): Correct CONST_INT handling.
753 (nios2_symbolic_memory_operand_p): Rename to...
754 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
755 (nios2_large_constant_p): Check for large constant integers too.
756 (nios2_split_large_constant): Handle constant integers.
757 (nios2_split_symbolic_memory_operand): Rename to...
758 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
759 (nios2_legitimize_constant_address): Handle constant integers.
760 (r0rel_constant_p): Handle small constant integers.
761 (nios2_print_operand_address): Handle r0-relative integer addresses.
762 * config/nios2/nios2-protos.h: Adjust for renamed functions.
763 * config/nios2/nios2.md: Adjust for renamed functions.
765 2017-12-07 Andrew Waterman <andrew@sifive.com>
767 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
768 (TARGET_HAVE_SRODATA_SECTION): New define.
769 (riscv_select_section): New function.
771 2017-12-08 Joseph Myers <joseph@codesourcery.com>
772 Alexander Monakov <amonakov@ispras.ru>
773 Jakub Jelinek <jakub@redhat.com>
776 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
778 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
780 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
781 a bool success value. Don't call output_operand_lossage here.
782 (aarch64_print_ldpstp_address): Return a bool success value.
783 (aarch64_print_operand_address): Call output_addr_const if
784 aarch64_print_address_internal fails.
785 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
786 'y'; call output_operand_lossage instead. Call output_operand_lossage
787 if aarch64_print_ldpstp_address fails.
789 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
791 * tree-vector-builder.h
792 (tree_vector_builder::binary_encoded_nelts): Declare.
793 * tree-vector-builder.c
794 (tree_vector_builder::binary_encoded_nelts): New function.
795 * fold-const.c (negate_expr_p): Likewise.
796 (operand_equal_p, fold_checksum_tree): Likewise.
797 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
798 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
799 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
800 (uniform_vector_p): Likewise.
801 * varasm.c (const_hash_1, compare_constant): Likewise.
802 * tree-ssa-ccp.c: Include tree-vector-builder.h.
803 (valid_lattice_transition): Operate directly on the VECTOR_CST
805 * ipa-icf.c: Include tree-vector-builder.h.
806 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
807 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
809 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
811 * tree.c (build_vector): Delete.
812 * tree.h (build_vector): Make static and move into the self-testing
815 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
817 * vector-builder.h (vector_builder::derived): New const overload.
818 (vector_builder::elt): New function.
819 * tree-vector-builder.h (tree_vector_builder::type): New function.
820 (tree_vector_builder::apply_step): Declare.
821 * tree-vector-builder.c (tree_vector_builder::apply_step): New
823 * gimple-fold.h (tree_vector_builder): Declare.
824 (gimple_build_vector): Take a tree_vector_builder instead of a
825 type and vector of elements.
826 * gimple-fold.c (gimple_build_vector): Likewise.
827 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
829 (get_initial_defs_for_reduction): Likewise.
830 (vectorizable_induction): Likewise.
832 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
834 * tree-vector-builder.h
835 (tree_vector_builder::new_binary_operation): Declare.
836 * tree-vector-builder.c
837 (tree_vector_builder::new_binary_operation): New function.
838 * fold-const.c (fold_relational_const): Use it.
839 (const_binop): Likewise. Check that both input vectors have
840 the same number of elements, thus excluding things like WIDEN_SUM.
841 Check whether it is possible to operate directly on the encodings
844 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
846 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
847 new_unary_operation, operating only on the encoded elements.
848 (const_unop): Likewise.
849 (exact_inverse): Likewise.
850 (distributes_over_addition_p): New function.
851 (const_binop): Use tree_vector_builder and new_unary_operation
852 for combinations of VECTOR_CST and INTEGER_CST. Operate only
853 on the encoded elements unless the encoding is strided and the
854 operation does not distribute over addition.
855 (fold_convert_const): Use tree_vector_builder and
856 new_unary_operation. Operate only on the encoded elements
857 for truncating integer conversions, or for non-stepped encodings.
859 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
861 * config/sparc/sparc.c: Include tree-vector-builder.h.
862 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
863 * expmed.c: Include tree-vector-builder.h.
864 (make_tree): Use tree_vector_builder instead of build_vector.
865 * fold-const.c: Include tree-vector-builder.h.
866 (const_binop): Use tree_vector_builder instead of build_vector.
867 (const_unop): Likewise.
868 (native_interpret_vector): Likewise.
869 (fold_vec_perm): Likewise.
870 (fold_ternary_loc): Likewise.
871 * gimple-fold.c: Include tree-vector-builder.h.
872 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
874 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
875 (simplify_vector_constructor): Use tree_vector_builder instead
877 * tree-vect-generic.c: Include tree-vector-builder.h.
878 (add_rshift): Use tree_vector_builder instead of build_vector.
879 (expand_vector_divmod): Likewise.
880 (optimize_vector_constructor): Likewise.
881 * tree-vect-loop.c: Include tree-vector-builder.h.
882 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
883 of build_vector. Explicitly use a stepped encoding for
885 * tree-vect-slp.c: Include tree-vector-builder.h.
886 (vect_get_constant_vectors): Use tree_vector_builder instead
888 (vect_transform_slp_perm_load): Likewise.
889 (vect_schedule_slp_instance): Likewise.
890 * tree-vect-stmts.c: Include tree-vector-builder.h.
891 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
892 (vect_gen_perm_mask_any): Likewise.
893 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
894 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
896 (build_vector_from_val): Likewise. Explicitly use a duplicate
899 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
901 * doc/generic.texi (VECTOR_CST): Describe new representation of
903 * vector-builder.h: New file.
904 * tree-vector-builder.h: Likewise.
905 * tree-vector-builder.c: Likewise.
906 * Makefile.in (OBJS): Add tree-vector-builder.o.
907 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
908 * tree-core.h (tree_base): Add a vector_cst field to the u union.
909 (tree_vector): Change the number of elements to
910 vector_cst_encoded_nelts.
911 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
912 (VECTOR_CST_ELTS): Delete.
913 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
914 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
915 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
916 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
917 (VECTOR_CST_ENCODED_ELT): Likewise.
918 (vector_cst_encoded_nelts): New function.
919 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
920 VECTOR_CST_NELTS_PER_PATTERN as arguments.
921 (vector_cst_int_elt, vector_cst_elt): Declare.
922 * tree.c: Include tree-vector-builder.h.
923 (tree_code_size): Abort if passed VECTOR_CST.
924 (tree_size): Update for new VECTOR_CST layout.
925 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
926 VECTOR_CST_NELTS_PER_PATTERN as arguments.
927 (build_vector): Use tree_vector_builder.
928 (vector_cst_int_elt, vector_cst_elt): New functions.
929 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
930 encoded elements and then create the vector in the canonical form.
931 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
932 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
933 (tree_c_tests): Call test_vector_cst_patterns.
934 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
936 (hash_tree): Likewise.
937 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
938 (streamer_write_tree_header): Likewise.
939 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
940 (streamer_alloc_tree): Likewise. Update call to make_vector.
941 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
943 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
945 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
946 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
947 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
949 2017-12-07 Bin Cheng <bin.cheng@arm.com>
950 Richard Biener <rguenther@suse.de>
952 PR tree-optimization/81303
953 * Makefile.in (gimple-loop-interchange.o): New object file.
954 * common.opt (floop-interchange): Reuse the option from graphite.
955 * doc/invoke.texi (-floop-interchange): Ditto. New document for
956 -floop-interchange and mention it for -O3.
957 * opts.c (default_options_table): Enable -floop-interchange at -O3.
958 * gimple-loop-interchange.cc: New file.
959 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
960 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
961 * passes.def (pass_linterchange): New pass.
962 * timevar.def (TV_LINTERCHANGE): New time var.
963 * tree-pass.h (make_pass_linterchange): New declaration.
964 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
965 interchange. Record IV before/after increment in new parameters.
966 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
967 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
969 (check_reduction_path): ...New function here.
970 * tree-vectorizer.h (check_reduction_path): New declaration.
972 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
975 PR rtl-optimization/80818
976 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
977 always early clobbered.
978 * lra-lives.c (process_bb_lives): Check input hard regs for early
979 clobbered non-operand hard reg.
981 2017-12-07 Jakub Jelinek <jakub@redhat.com>
984 * tree-cfg.c (verify_gimple_assign_binary): Don't require
985 types_compatible_p, just that TYPE_MODE is the same.
987 2017-12-07 Martin Sebor <msebor@redhat.com>
990 * attribs.c (empty_attribute_table): Initialize new member of
991 struct attribute_spec.
992 (decl_attributes): Add argument. Handle mutually exclusive
993 combinations of attributes.
994 (selftests::test_attribute_exclusions): New function.
995 (selftests::attribute_c_tests): Ditto.
996 * attribs.h (decl_attributes): Add default argument.
997 * selftest.h (attribute_c_tests): Declare.
998 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
999 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1001 * doc/extend.texi (Common Function Attributes): Update const and pure.
1002 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1003 of struct attribute_spec.
1004 * config/arc/arc.c (arc_attribute_table): Same.
1005 * config/arm/arm.c (arm_attribute_table): Same.
1006 * config/avr/avr.c ( avr_attribute_table): Same.
1007 * config/bfin/bfin.c (bfin_attribute_table): Same.
1008 * config/cr16/cr16.c (cr16_attribute_table): Same.
1009 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1010 * config/h8300/h8300.c (h8300_attribute_table): Same.
1011 * config/i386/i386.c (ix86_attribute_table): Same.
1012 * config/ia64/ia64.c (ia64_attribute_table): Same.
1013 * config/m32c/m32c.c (m32c_attribute_table): Same.
1014 * config/m32r/m32r.c (m32r_attribute_table): Same.
1015 * config/m68k/m68k.c (m68k_attribute_table): Same.
1016 * config/mcore/mcore.c (mcore_attribute_table): Same.
1017 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1018 * config/mips/mips.c (mips_attribute_table): Same.
1019 * config/msp430/msp430.c (msp430_attribute_table): Same.
1020 * config/nds32/nds32.c (nds32_attribute_table): Same.
1021 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1022 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1023 * config/rl78/rl78.c (rl78__attribute_table): Same.
1024 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1025 * onfig/rx/rx.c (rx_attribute_table): Same.
1026 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1027 * config/sh/sh.c (sh_attribute_table): Same.
1028 * config/sparc/sparc.c (sparc_attribute_table): Same.
1029 * config/spu/spu.c (spu_attribute_table): Same.
1030 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1031 * config/v850/v850.c (v850_attribute_table): Same.
1032 * config/visium/visium.c (visium_attribute_table): Same.
1034 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1037 * config/arm/arm.c (INCLUDE_STRING): Define.
1038 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1039 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1042 2017-12-07 Michael Matz <matz@suse.de>
1044 Add unroll and jam pass
1046 * gimple-loop-jam.c: New file.
1047 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1048 * common.opt (funroll-and-jam): New option.
1049 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1050 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1051 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1052 * passes.def: Add pass_loop_jam.
1053 * timevar.def (TV_LOOP_JAM): Add.
1054 * tree-pass.h (make_pass_loop_jam): Declare.
1055 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1056 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1057 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1058 to flow_loop_tree_node_add.
1059 (duplicate_subloops, copy_loops_to): Append to sibling list.
1060 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1061 * doc/invoke.texi (-funroll-and-jam): Document new option.
1062 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1064 2017-12-07 Richard Biener <rguenther@suse.de>
1066 PR tree-optimization/83296
1067 PR tree-optimization/67769
1068 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1069 flow sensitive info in an unrelated BB.
1070 (value_replacement): Use reset_flow_sensitive_info.
1071 (minmax_replacement): Reset flow sensitive info on the def
1072 we move. Do not reset flow sensitive info in the whole BB
1073 we move the stmt to.
1074 (abs_replacement): Likewise.
1076 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1079 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1080 rs6000_cpu to the given -mcpu=, or to the default processor.
1082 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1084 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1085 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1086 instead of rs6000_cpu_attr.
1087 (rs6000_adjust_cost): Ditto.
1088 (is_microcoded_insn): Ditto.
1089 (rs6000_adjust_priority): Ditto.
1090 (rs6000_issue_rate): Ditto.
1091 (rs6000_use_sched_lookahead): Ditto.
1092 (rs6000_use_sched_lookahead_guard): Ditto.
1093 (rs6000_sched_reorder): Ditto.
1094 (force_new_group): Ditto.
1095 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1096 (group_ending_nop): Ditto.
1098 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1100 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1101 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1102 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1103 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1105 (rs6000_reassociation_width): Ditto.
1106 (rs6000_emit_epilogue): Ditto.
1107 (rs6000_adjust_cost): Ditto.
1108 (is_microcoded_insn): Ditto.
1109 (is_cracked_insn): Ditto.
1110 (rs6000_adjust_priority): Ditto.
1111 (rs6000_sched_reorder): Ditto.
1112 (rs6000_sched_reorder2): Ditto.
1113 (insn_must_be_first_in_group): Ditto.
1114 (insn_must_be_last_in_group): Ditto.
1115 (rs6000_register_move_cost): Ditto.
1116 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1119 2017-12-07 Julia Koval <julia.koval@intel.com>
1121 * config.gcc: Add vaesintrin.h.
1122 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1124 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1125 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1127 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1128 * config/i386/immintrin.h: Include vaesintrin.h.
1129 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1130 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1131 _mm_aesdec_epi128): New intrinsics.
1133 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1135 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1136 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1137 c-family/c-spellcheck.cc.
1138 (best_macro_match::best_macro_match): Likewise.
1140 (struct edit_distance_traits<cpp_hashnode *>): Move to
1141 c-family/c-spellcheck.h.
1142 (class best_macro_match): Likewise.
1144 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1146 PR tree-optimization/83293
1147 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1148 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1149 might insert into empty bb.
1152 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1153 simplify for plus with :c added, and pointer_plus without that.
1154 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1155 with undefined overflow and the conversion is not widening,
1156 perform negation in utype and only convert to type afterwards.
1157 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1158 simplify for plus with :c added, and pointer_plus without that.
1159 If type is integral with undefined overflow and the conversion is
1160 not widening, perform minus in utype and only convert to type
1161 afterwards. Move the last pointer_diff_expr simplify into the
1164 2017-12-06 Martin Sebor <msebor@redhat.com>
1166 PR tree-optimization/82646
1167 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1168 strncpy, not maxlen.
1170 2017-12-06 Martin Sebor <msebor@redhat.com>
1172 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1175 PR tree-optimization/83075
1176 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1177 strncat/strncpy don't change length of source string.
1179 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1182 2017-11-29 Martin Aberg <maberg@gaisler.com>
1184 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1185 to prevent b2bst errata sequence.
1186 (sqrtdf2_fix): Likewise.
1188 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1190 PR tree-optimization/81945
1191 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1192 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1193 to dest_cfun has orig_loop_num set, either remap it to the new
1194 loop number if the loop got moved too, or clear it.
1196 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1198 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1200 (thunderx2t99_nothing): New insn reservation.
1201 (thunderx2t99_mrs): New insn reservation.
1202 (thunderx2t99_multiple): New insn reservation.
1203 (thunderx2t99_alu_basi): Add bfx to reservation.
1204 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1206 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1209 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1211 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1213 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1214 tree-ssa-pre.c::remove_dead_inserted_code.
1215 * tree-ssa-dce.h: New file.
1216 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1217 (remove_dead_inserted_code): Move and rename to function
1218 tree-ssa-dce.c::simple_dce_from_worklist.
1219 (pass_pre::execute): Update use.
1221 2017-12-05 Richard Biener <rguenther@suse.de>
1223 PR tree-optimization/83277
1224 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1225 to code-gen liveout vars.
1227 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1229 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1230 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1231 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1234 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1236 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1238 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1239 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1242 2017-12-05 Richard Biener <rguenther@suse.de>
1244 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1245 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1246 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1247 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1249 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1251 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1253 2017-12-05 Martin Liska <mliska@suse.cz>
1254 Jakub Jelinek <jakub@redhat.com>
1256 * doc/invoke.texi: Document the options.
1257 * flag-types.h (enum sanitize_code): Add
1258 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1259 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1260 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1261 * opts.c: Define new sanitizer options.
1262 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1263 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1265 2017-12-05 Julia Koval <julia.koval@intel.com>
1267 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1268 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1269 (ix86_handle_option): Handle -mavx512vnni.
1270 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1271 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1272 * config/i386/i386-c (__AVX512VNNI__): New.
1273 * config/i386/i386.c (ix86_target_string): Handle new option.
1274 (ix86_valid_target_attribute_inner_p): Handle new option.
1275 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1276 * config/i386/i386.opt (mavx512vnni): New option.
1278 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1281 * athlon.md: Disable for generic.
1282 * haswell.md: Enable for generic.
1283 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1284 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1286 (ix86_adjust_cost): Move generic to haswell path.
1288 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1290 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1291 instead of MEM_P in a couple more places. Fix formatting issues.
1293 2017-12-04 Jim Wilson <jimw@sifive.com>
1295 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1296 instead of GP_REG_LAST-1.
1297 (riscv_adjust_libcall_cfi_prologue): Likewise.
1298 (riscv_adjust_libcall_cri_epilogue): Likewise.
1299 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1302 2017-12-04 Luis Machado <luis.machado@linaro.org>
1304 * ipa-pure-const.c (check_decl): Add missing newline.
1305 (state_from_flags): Likewise.
1307 2017-12-04 Jeff Law <law@redhat.com>
1309 PR tree-optimizatin/78496
1310 * gimple-ssa-evrp-analyze.h
1311 (evrp_range_analyzer::get_vr_values): Simplify.
1312 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1313 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1314 and gimple-ssa-evrp-analyze.h.
1315 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1316 (simplify_stmt_for_jump_threading): Copy a blob of code from
1317 tree-vrp.c to use ranges to simplify statements.
1318 (dom_opt_dom_walker::before_dom_children): Call
1319 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1320 (dom_opt_dom_walker::after_dom_children): Similarly for
1321 evrp_range_analyzer::leave.
1322 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1325 * gimple-ssa-evrp-analyze.c
1326 (evrp_range_analyzer::extract_range_from_stmt): Always use
1327 vr_values::update_value_range so preexisting range info is
1328 medged with new range info, even if the new range is VR_VARYING.
1330 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1332 * combine.c: Adjust comment.
1333 (use_crosses_set_p): Delete.
1334 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1335 (try_combine): Ditto.
1337 2017-12-04 Richard Biener <rguenther@suse.de>
1339 PR tree-optimization/83255
1340 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1341 Re-add zero-iteration check.
1343 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1345 PR rtl-optimization/83245
1346 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1347 hard registers as earlyclobber, also if not in an asm.
1349 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1353 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1357 rs6000_cpu based on cpu_index, not tune_index.
1359 2017-12-04 Richard Biener <rguenther@suse.de>
1361 PR tree-optimization/83238
1362 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1363 code match comment, rejecting invalid SESE regions.
1365 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1367 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1368 require base operand is a REG_POINTER prior to reload on targets
1369 with non-equivalent space registers.
1371 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1373 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1374 (update_specialized_profile): Fix updating of counts.
1375 (perhaps_add_new_callers): Likewise.
1377 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1380 * x86-tune.def: Remove obsolette FIXMEs.
1381 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1382 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1383 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1385 (X86_TUNE_PAD_RETURNS): Disable for generic.
1387 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1389 PR tree-optimization/83170
1390 PR tree-optimization/83241
1391 * gimple-ssa-store-merging.c
1392 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1393 gimple_vuse (ins_stmt) in case it has changed.
1394 (imm_store_chain_info::output_merged_store): Likewise.
1396 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1400 * tree-cfg.c (pass_warn_function_return::execute): Handle
1401 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1402 with BUILTINS_LOCATION.
1406 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1407 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1408 instead of DECL_MODE.
1410 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1411 last argument to rtx pointer.
1412 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1413 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1414 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1415 depending on the chosen ISAs.
1416 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1417 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1418 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1420 * config/i386/sse.md (mov<mode>_internal): Likewise.
1421 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1423 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1425 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1427 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1429 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1430 parameters from prototype.
1431 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1432 parameters. Don't print a comment.
1433 (emit_fusion_gpr_load): Adjust.
1434 (emit_fusion_load_store): Adjust.
1435 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1436 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1437 comment on the second line.
1439 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1443 rs6000_cpu based on cpu_index, not tune_index.
1445 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1447 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1448 which_alternative instead of which_alternative + 1.
1449 (output_asm_insn): Print an extra tab if the template is short.
1451 2017-12-01 Jim Wilson <jimw@sifive.com>
1453 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1455 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1456 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1458 * doc/tm.texi: Regenerate.
1460 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1463 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1464 whether we can allocate pseudos before trying to fix an address.
1465 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1466 memory address is indexed or indirect.
1467 (floatuns_<mode>si2_hw2): Likewise.
1469 2017-12-01 Jason Merrill <jason@redhat.com>
1471 * Makefile.in (TAGS): Add c-family/*.cc.
1473 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1475 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1476 (call_value_insn): Likewise.
1477 (sibcall_insn): Likewise.
1478 (sibcall_value_insn): Likewise.
1479 (movsi_aarch64): Likewise.
1480 (movdi_aarch64): Likewise.
1481 (add_losym_): Likewise.
1482 (ldr_got_small_): Likewise.
1483 (ldr_got_small_sidi): Likewise.
1484 (ldr_got_small_28k_): Likewise.
1485 (ldr_got_small_28k_sidi): Likewise.
1486 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1487 Move output_addr_const to symbolic case. Add error check.
1489 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1491 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1492 * config/s390/s390.md: Disable MVC merging peephole if it would
1493 disable operand forwarding.
1494 (new peephole2): Split MVCs if it would turn them into up to 2
1497 2017-12-01 Richard Biener <rguenther@suse.de>
1499 PR tree-optimization/83232
1500 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1501 detection of same access. Instead of breaking the group here
1502 do not consider the duplicate. Add comment explaining real fix.
1504 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1506 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1508 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1510 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1512 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1514 * function.h (struct function): Remove cilk_frame_decl,
1515 is_cilk_function and calls_cilk_spawn fields.
1516 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1518 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1519 cilk_elemental field.
1520 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1521 * target.def: Adjust comment.
1522 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1523 Don't test cilk_elemental.
1525 PR tree-optimization/83233
1526 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1527 bswap_stat name for the struct.
1530 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1531 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1532 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1533 SWITCH_BREAK_LABEL_P set on the label.
1534 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1535 added for default case if it was missing and not all cases covered.
1536 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1537 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1539 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1540 to avoid -Wimplicit-fallthrough warning.
1541 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1542 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1544 PR tree-optimization/83221
1545 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1547 (init_reassoc): Formatting fix.
1550 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1551 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1555 * tree-cfg.c: Include asan.h.
1556 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1558 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1560 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1562 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1564 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1565 vpcmpeqd instruction.
1567 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1569 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1570 registers type for 128bit mode.
1572 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1574 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1576 * gengtype-state.c (read_state_pair): Likewise.
1577 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1578 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1579 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1580 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1581 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1582 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1583 * builtins.c (fold_builtin_3): Likewise.
1584 * graphite-scop-detection.c
1585 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1586 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1589 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1592 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1593 parameter. Return true if ignore_decl_rtl_set_p is true and other
1594 conditions are satisfied.
1595 * asan.h (asan_protect_global): Add new parameter.
1596 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1597 to asan_protect_global calls.
1599 2017-11-30 Jim Wilson <jimw@sifive.com>
1601 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1602 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1603 -mstrict-align, add info on default value. Delete redundant lines for
1604 -mabi. Add missing -mexplicit-relocs docs.
1606 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1608 * config/arc/arc.md (trap): New pattern.
1610 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1612 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1613 instruction to end into a delay slot.
1614 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1615 can be placed into a delay slot against reg_note.
1617 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1619 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1620 labels number of usages.
1622 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1624 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1626 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1628 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1631 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1632 multiplication by power of 2 constant into two shifts + comparison.
1634 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1637 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1639 2017-11-30 Richard Biener <rguenther@suse.de>
1641 PR tree-optimization/83202
1642 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1643 allow_peel argument and guard peeling.
1644 (canonicalize_loop_induction_variables): Likewise.
1645 (canonicalize_induction_variables): Pass false.
1646 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1647 peeling from cunrolli.
1649 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1651 * combine.c (try_combine): Print a message to dump file whenever
1652 I0, I1, or I2 cannot be combined into I3.
1654 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1656 PR rtl-optimization/83156
1657 PR rtl-optimization/82621
1658 * combine.c (try_combine): Don't split an I2 if one of the dests is
1659 set again before I3. Allow unused dests.
1661 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1663 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1665 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1667 PR rtl-optimization/80818
1668 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1669 recursively. Use insn code for clobber.
1670 (lra_set_insn_recog_data): Pass the new arg to
1671 collect_non_operand_hard_regs.
1672 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1674 (lra_update_insn_regno_info): Pass insn to
1675 add_regs_to_insn_regno_info.
1677 2017-11-29 Jim Wilson <jimw@sifive.com>
1678 Andrew Waterman <andrew@sifive.com>
1680 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1681 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1682 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1683 use SINGLE_SHIFT_COST.
1684 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1685 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1687 2017-11-29 Julia Koval <julia.koval@intel.com>
1689 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1690 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1691 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1692 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1693 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1694 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1695 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1696 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1697 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1698 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1699 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1700 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1701 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1702 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1703 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1704 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1705 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1706 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1707 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1708 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1709 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1710 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1711 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1712 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1713 __builtin_ia32_vpshldv_v2di_mask,
1714 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1715 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1716 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1718 2017-11-29 Julia Koval <julia.koval@intel.com>
1720 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1721 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1722 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1723 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1724 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1725 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1726 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1727 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1728 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1729 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1730 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1731 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1732 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1733 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1734 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1735 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1736 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1737 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1738 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1739 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1740 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1742 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1744 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1745 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1748 2017-11-29 Martin Aberg <maberg@gaisler.com>
1750 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1751 to prevent b2bst errata sequence.
1752 (sqrtdf2_fix): Likewise.
1754 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1756 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1757 (div_sqrt_insn_p): New function.
1758 (sparc_do_work_around_errata): Insert NOP instructions to
1759 prevent sequences that could trigger the TN-0013 errata for
1760 certain LEON3 processors.
1761 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1762 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1763 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1764 (in_branch_delay): Prevent div and sqrt in delay slot if
1766 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1768 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1770 * config/sparc/sparc.c (atomic_insn_p): New function.
1771 (sparc_do_work_around_errata): Insert NOP instructions to
1772 prevent sequences that could trigger the TN-0010 errata for
1774 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1775 instruction referable in atomic_insns_p.
1777 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1779 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1780 (atomic_compare_and_swap_leon3_1): Likewise.
1783 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1785 * config/sparc/sparc.c (fpop_insn_p): New function.
1786 (sparc_do_work_around_errata): Insert NOP instructions to
1787 prevent sequences that could trigger the TN-0012 errata for
1789 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1790 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1791 (in_branch_annul_delay): Prevent floating-point instructions
1792 in delay slot of annulled integer branch.
1794 2017-11-29 Richard Biener <rguenther@suse.de>
1796 PR tree-optimization/83202
1797 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1799 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1800 nodes vectorized to the same stmts multiple times.
1801 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1802 (vect_analyze_slp_instance): Adjust.
1803 (scalar_stmts_to_slp_tree_map_t): New typedef.
1804 (vect_schedule_slp_instance): Add a map recording the SLP node
1805 representing the vectorized stmts for a set of scalar stmts.
1806 Avoid code-generating redundancies.
1807 (vect_schedule_slp): Allocate map and pass it down.
1809 2017-11-29 Nathan Sidwell <nathan@acm.org>
1812 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1815 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1817 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1818 cases for printing LDP/STP memory addresses.
1819 (aarch64_print_address_internal): Renamed from
1820 aarch64_print_operand_address, added parameter, add Pmode check.
1821 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1822 (aarch64_print_operand_address): Indirect to
1823 aarch64_print_address_internal.
1824 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1825 'y' operand output specifier.
1827 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1830 * tree.c (build_simple_mem_ref_loc): Handle
1831 get_addr_base_and_unit_offset returning a MEM_REF.
1834 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1837 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1840 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1842 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1844 (layout::move_to_column): Likewise.
1846 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1849 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1850 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1852 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1853 Martin Jambor <mjambor@suse.cz>
1856 * tree.h (expr_type_first_operand_type_p): Declare
1857 * tree.c (expr_type_first_operand_type_p): New function.
1858 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1859 (ipa_value_from_jfunc): Adjust declaration.
1860 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1861 Use it as result type for arithmetics, unless it is NULL in which case
1862 be more conservative.
1863 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1864 ipa_get_jf_pass_through_result.
1865 (propagate_vals_across_pass_through): Likewise.
1866 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1867 is to propagate_vals_across_pass_through.
1868 (propagate_constants_across_call): Pass PARM_TYPE to
1869 propagate_scalar_across_jump_function.
1870 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1871 ipa_value_from_jfunc.
1872 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1873 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1874 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1875 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1876 target_type, pass it to ipa_value_from_jfunc.
1877 (update_indirect_edges_after_inlining): Pass parameter type to
1878 try_make_edge_direct_simple_call.
1880 2017-11-28 Jeff Law <law@redhat.com>
1882 * gimple-ssa-evrp-analyze.c
1883 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1884 refine ranges if scev_initialized_p returns true.
1885 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1887 2017-11-28 Julia Koval <julia.koval@intel.com>
1889 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1890 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1891 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1892 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1893 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1894 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1895 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1896 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1897 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1898 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1899 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1900 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1901 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1902 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1903 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1904 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1905 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1906 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1907 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1908 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1909 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1910 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1912 2017-11-28 Julia Koval <julia.koval@intel.com>
1914 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1915 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1916 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1917 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1918 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1919 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1920 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1921 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1922 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1923 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1924 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1925 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1926 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1927 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1928 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1929 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1930 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1931 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1932 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1933 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1934 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1935 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1936 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1937 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1938 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1939 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1940 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1941 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1942 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1943 __builtin_ia32_vpshld_v2di_mask): New builtins.
1944 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1945 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1947 2017-11-28 Richard Biener <rguenther@suse.de>
1949 PR tree-optimization/80776
1950 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1952 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1954 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1955 If the incoming edge is an effective fallthru because the other
1956 edge only reaches a __builtin_unreachable () then record ranges
1957 derived from the controlling condition in SSA info.
1958 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1959 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1961 2017-11-28 Olivier Hainque <hainque@adacore.com>
1963 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1964 on mingw build hosts.
1966 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1969 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1971 2017-11-28 Richard Biener <rguenther@suse.de>
1974 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1975 copies generated from memcpy use a character array as reference
1978 2017-11-28 Julia Koval <julia.koval@intel.com>
1979 Sebastian Peryt <sebastian.peryt@intel.com>
1981 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1982 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1983 cilk-common.o, cilk.h, cilk-common.c): Remove.
1985 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1986 * builtins.c (is_builtin_name): Remove cilkplus condition.
1987 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1988 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1989 cilk-builtins.def, cilkplus.def): Remove.
1990 * cif-code.def (CILK_SPAWN): Remove.
1991 * cilk-builtins.def: Delete.
1992 * cilk-common.c: Ditto.
1994 * cilkplus.def: Ditto.
1995 * config/darwin.h (fcilkplus): Delete.
1996 * cppbuiltin.c: Ditto.
1997 * doc/extend.texi: Remove cilkplus doc.
1998 * doc/generic.texi: Ditto.
1999 * doc/invoke.texi: Ditto.
2000 * doc/passes.texi: Ditto.
2001 * gcc.c (fcilkplus): Remove.
2002 * gengtype.c (cilk.h): Remove.
2003 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2005 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2007 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2009 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2010 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2011 cilkplus conditions.
2012 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2013 inline_read_section): Ditto.
2014 * ipa-inline-analysis.c (cilk.h): Remove.
2015 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2016 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2017 append_linker_options): Remove condition for fcilkplus.
2018 * lto/lto-lang.c (cilk.h): Remove.
2019 (lto_init): Remove condition for fcilkplus.
2020 * omp-expand.c (expand_cilk_for_call): Delete.
2021 (expand_omp_taskreg, expand_omp_for_static_chunk,
2022 expand_omp_for): Remove cilkplus
2024 (expand_cilk_for): Delete.
2025 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2026 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2027 execute_lower_omp, diagnose_sb_0): Ditto.
2028 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2029 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2030 * tree-nested.c: Ditto.
2031 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2032 (dump_generic_node): Ditto.
2033 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2034 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2036 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2038 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2040 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2041 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2042 for vector mode and !TARGET_SIMD.
2044 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2046 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2048 * tree.h (SWITCH_LABELS): Remove.
2049 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2050 assert SWITCH_BODY is non-NULL.
2051 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2053 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2055 PR tree-optimization/80788
2056 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2057 has TREE_OVERFLOW set, call drop_tree_overflow.
2059 2017-11-28 Richard Biener <rguenther@suse.de>
2061 PR tree-optimization/83158
2062 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2064 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2067 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2068 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2070 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2072 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2073 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2075 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2078 * varasm.c (compare_constant): Take the mode of the constants into
2079 account when comparing floating point constants.
2081 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2083 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2084 from explicit instantiation of debug_helper.
2085 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2087 2017-11-27 Richard Biener <rguenther@suse.de>
2089 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2092 2017-11-27 Richard Biener <rguenther@suse.de>
2094 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2095 (type_hash_canon): Also clear int_cst_hash_table entry for
2097 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2099 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2101 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2102 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2103 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2105 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2107 * hash-map.h (gt_cleare_cache): Avoid UB.
2109 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2111 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2112 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2113 (decide_unroll_runtime_iterations): Likewise.
2114 (decide_unroll_stupid): Likewise.
2116 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2119 * config/i386/i386.md: Add a loop with incssp.
2121 2017-11-27 Martin Jambor <mjambor@suse.cz>
2123 PR tree-optimization/81248
2124 * tree-sra.c (splice_param_accesses): Remove size check.
2125 (decide_one_param_reduction): Fix size check.
2126 * gimple-pretty-print.c (dump_profile): Silence warning.
2127 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2129 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2132 * dbxout.c (lastlineno): New variable.
2133 (dbx_debug_hooks): Use dbxout_switch_text_section as
2134 switch_text_section debug hook.
2135 (dbxout_function_end): Switch to current_function_section
2136 rather than function_section. If crtl->has_bb_partition,
2137 output just one N_FUN, depending on in_cold_section_p.
2138 (dbxout_source_line): Remember last lineno in lastlineno.
2139 (dbxout_switch_text_section): New function.
2140 (dbxout_function_decl): Adjust dbxout_block caller.
2141 (dbx_block_with_cold_children): New function.
2142 (dbxout_block): Return true if any LBRAC/RBRAC have been
2143 emitted. Use dbx_block_with_cold_children at depth == 0
2144 in second partition. Add PARENT_BLOCKNUM argument, pass
2145 it optionally adjusted to children. Output LBRAC/RBRAC
2146 around recursive call only if the block is in the current
2147 partition, if not and anything was output, emit empty
2149 * final.c (final_scan_insn): Compute cold_function_name
2150 before calling switch_text_section debug hook. Call
2151 that hook even if dwarf2out_do_frame if not emitting
2155 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2156 TREE_READONLY decls.
2158 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2160 PR rtl-optimization/82488
2161 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2163 2017-11-26 Julia Koval <julia.koval@intel.com>
2165 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2167 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2170 2017-11-26 Julia Koval <julia.koval@intel.com>
2172 * config/i386/driver-i386.c (host_detect_local_cpu):
2173 Detect skylake-avx512.
2175 2017-11-26 Julia Koval <julia.koval@intel.com>
2177 * config.gcc: Add -march=cannonlake.
2178 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2179 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2180 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2181 (PTA_CANNONLAKE): New.
2182 (processor_target_table): Add cannonlake.
2183 (ix86_option_override_internal): Ditto.
2184 (fold_builtin_cpu): Ditto.
2185 (get_builtin_code_for_version): Handle cannonlake.
2186 (M_INTEL_COREI7_CANNONLAKE): New.
2187 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2188 * doc/invoke.texi: Add -march=cannonlake.
2190 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2192 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2193 (try_init_one_plugin): Alternative implementation for MinGW.
2194 * Makefile.in (plugin_implib): New.
2195 (gengtype-lex.c): Fix broken AIX workaround.
2196 * configure: Regenerate.
2197 * doc/plugins.texi: Document support for MinGW.
2199 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2201 PR rtl-optimization/81553
2202 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2203 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2204 is a shift where C1 has different mode than the whole shift, use C1's
2205 mode for MULT rather than the shift's mode.
2208 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2209 builtins not enabled in the currently selected ISA.
2211 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2213 PR tree-optimization/71026
2214 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2215 (insert_reciprocals): Change to insert reciprocals before a division
2216 by a square and to insert the square of a reciprocal.
2217 (execute_cse_reciprocals_1): Change to consider division by a square.
2218 (register_division_in): Add importance parameter.
2220 2017-11-24 Richard Biener <rguenther@suse.de>
2222 PR tree-optimization/82402
2223 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2224 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2226 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2228 * match.pd (0-ptr): New transformation.
2230 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2233 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2234 when skipping checking.
2236 2017-11-24 Richard Biener <rguenther@suse.de>
2238 PR tree-optimization/83128
2239 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2240 (vn_reference_lookup_3): Likewise.
2242 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2245 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2246 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2249 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2250 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2251 check for the latter with gimple_call_builtin_p. Do not handle
2252 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2254 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2256 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2259 2017-11-23 Julia Koval <julia.koval@intel.com>
2261 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2262 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2263 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2264 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2265 _mm512_maskz_expandloadu_epi16): New intrinsics.
2266 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2267 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2268 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2269 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2270 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2271 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2272 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2273 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2274 _mm256_maskz_expandloadu_epi8): New intrinsics.
2275 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2276 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2277 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2278 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2279 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2280 * config/i386/sse.md (VI248_VLBW): New iterator.
2281 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2283 2017-11-23 Julia Koval <julia.koval@intel.com>
2285 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2286 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2287 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2288 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2289 _mm512_mask_compressstoreu_epi16): New.
2290 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2291 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2292 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2293 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2294 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2295 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2296 _mm256_mask_compressstoreu_epi8): New.
2297 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2298 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2299 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2300 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2301 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2302 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2303 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2304 __builtin_ia32_compresshi128_mask,
2305 __builtin_ia32_compressstoreuqi512_mask,
2306 __builtin_ia32_compressstoreuhi512_mask,
2307 __builtin_ia32_compressstoreuqi256_mask,
2308 __builtin_ia32_compressstoreuqi128_mask,
2309 __builtin_ia32_compressstoreuhi256_mask,
2310 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2311 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2313 (ix86_expand_special_args_builtin): Handle new types.
2314 (s4fma_expand): Handle new builtin array.
2315 * config/i386/immintrin.h: Include new headers.
2316 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2317 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2319 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2321 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2322 fixes. Declare temp and g variables at the top in order to avoid
2323 {} in most of the cases.
2325 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2327 * match.pd (ptr-0): New transformation.
2329 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2331 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2332 (struct addr_mode_cost_table): New.
2333 (struct tune_params): Add field addr_mode_costs.
2334 * config/arm/arm.c (generic_addr_mode_costs): New.
2335 (arm_slowmul_tune): Initialise addr_mode_costs field.
2336 (arm_fastmul_tune): Likewise.
2337 (arm_strongarm_tune): Likewise.
2338 (arm_xscale_tune): Likewise.
2339 (arm_9e_tune): Likewise.
2340 (arm_marvell_pj4_tune): Likewise.
2341 (arm_v6t2_tune): Likewise.
2342 (arm_cortex_tune): Likewise.
2343 (arm_cortex_a8_tune): Likewise.
2344 (arm_cortex_a7_tune): Likewise.
2345 (arm_cortex_a15_tune): Likewise.
2346 (arm_cortex_a35_tune): Likewise.
2347 (arm_cortex_a53_tune): Likewise.
2348 (arm_cortex_a57_tune): Likewise.
2349 (arm_exynosm1_tune): Likewise.
2350 (arm_xgene1_tune): Likewise.
2351 (arm_cortex_a5_tune): Likewise.
2352 (arm_cortex_a9_tune): Likewise.
2353 (arm_cortex_a12_tune): Likewise.
2354 (arm_cortex_a73_tune): Likewise.
2355 (arm_v7m_tune): Likewise.
2356 (arm_cortex_m7_tune): Likewise.
2357 (arm_v6m_tune): Likewise.
2358 (arm_fa726te_tune): Likewise.
2359 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2361 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2363 * config/arm/arm.c (arm_mem_costs): New function.
2364 (arm_rtx_costs_internal): Use arm_mem_costs.
2366 2017-11-23 Mark Wielaard <mark@klomp.org>
2368 * dwarf2out.c (init_sections_and_labels): Use generation to create
2369 unique ranges_section_label and ranges_base_label. Return generation.
2370 (output_rnglists): Add generation argument. Use generation to create
2371 unique ranges labels.
2372 (dwarf2out_finish): Get generation from init_sections_and_labels
2373 and pass generation to output_rnglists.
2375 2017-11-23 Mike Stump <mikestump@comcast.net>
2376 Eric Botcazou <ebotcazou@adacore.com>
2378 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2379 * cfgloop.h (struct loop): Add unroll field.
2380 * function.h (struct function): Add has_unroll bitfield.
2381 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2382 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2383 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2384 (pass_rtl_unroll_loops::gate): Likewise.
2385 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2386 for which loop->unroll==1.
2387 (decide_unroll_constant_iterations): Use note for consistency and
2388 take loop->unroll into account. Return early if loop->unroll is set.
2389 Fix thinko in existing test.
2390 (decide_unroll_runtime_iterations): Use note for consistency and
2391 take loop->unroll into account.
2392 (decide_unroll_stupid): Likewise.
2393 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2394 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2395 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2397 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2398 (print_loop): Print loop->unroll if set.
2399 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2400 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2401 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2403 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2404 loop->unroll is set and smaller than the trip count. Otherwise bypass
2405 entirely the heuristics if loop->unroll is set. Remove dead note.
2406 Fix off-by-one bug in other note.
2407 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2408 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2410 (tree_unroll_loops_completely): Make static.
2411 (pass_complete_unroll::execute): Use correct type for variable.
2412 (pass_complete_unrolli::execute): Fix formatting.
2413 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2415 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2417 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2418 enable when TARGET_PREFER_AVX128 is set.
2420 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2422 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2424 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2425 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2427 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2429 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2430 frequencies. Fix estimation of aggregate parameters.
2432 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2434 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2437 2017-11-23 Tom de Vries <tom@codesourcery.com>
2439 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2441 2017-11-23 Tom de Vries <tom@codesourcery.com>
2443 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2446 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2449 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2450 sibcall_value_pcrel_fdpic): Use local variable instead of
2452 (calli_tbr_rel): Add missing operand 2.
2453 (call_valuei_tbr_rel): Add missing operand 3.
2455 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2458 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2459 bitpos/bitsize covering the whole destination, use store_expr only if
2460 the complex mode is the same. Otherwise, use expand_normal and if
2461 it returns CONCAT, subreg each part separately instead of trying to
2462 subreg the whole result.
2464 2017-11-23 Richard Biener <rguenther@suse.de>
2466 PR tree-optimization/23094
2467 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2468 come here from walking over backedges in the first iteration.
2469 (vn_reference_lookup_3): Skip clobbers that store the same value.
2471 2017-11-23 Richard Biener <rguenther@suse.de>
2473 PR tree-optimization/81403
2474 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2475 a block we need a leader relative to.
2476 (phi_translate_1): For nary processing require a leader from
2477 get_representative_for given we run expression simplification
2478 using match-and-simplify. Remove previous fix.
2480 2017-11-22 Jeff Law <law@redhat.com>
2482 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2483 Use new method allocate_value_range rather than accessing the
2484 vrp_value_range_pool data member directly.
2485 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2486 to use extract_range_from_stmt method to avoid need for
2487 extract_range_from_assignment method.
2488 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2489 method rather than setting values_propgated data member directly.
2490 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2491 and values propagated data members and extract_range_from_assignment
2492 method. Reorder private data members to conform to standards.
2493 Add new methods set_lattice_propagation_complete and
2494 allocate_value_range.
2496 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2498 PR rtl-optimization/83030
2499 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2500 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2501 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2502 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2503 CROSSING_JUMP_P flag is already set before setting it.
2504 * cfgrtl.c (fixup_partition_crossing): Likewise.
2505 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2508 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2510 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2511 Handle the case where both arguments are using gen_const_vec_series.
2513 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2516 * pretty-print.c (pp_format): Move quoting implementation to
2517 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2518 to pass address of "quote" local.
2519 (pp_begin_quote): New function.
2520 (pp_end_quote): New function.
2521 * pretty-print.h (printer_fn): Convert penultimate param from bool
2523 (pp_begin_quote): New decl.
2524 (pp_end_quote): New decl.
2525 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2526 param from bool to bool *.
2527 * tree-diagnostic.h (default_tree_printer): Likewise.
2529 2017-11-22 Jeff Law <law@redhat.com>
2531 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2532 Initialize vr_values.
2533 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2534 vr_values via vr_values class instance rather than delegators.
2535 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2536 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2537 (evrp_range_analyzer::push_value_range): Likewise.
2538 (evrp_range_analyzer::pop_value_range): Likewise.
2539 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2540 most delegators. Those remaining are exposed as public interfaces.
2541 Make vr_values a pointer and private.
2542 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2544 (evrp_range_analyzer::get_vr_value): New method.
2545 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2546 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2547 (class evrp_dom_walker): Attach evrp_folder class, initialize
2548 it in the ctor. Remove temporary delegators.
2549 (evrp_dom_walker::before_dom_children): Call methods in attached
2550 evrp_range_analyzer class via class instance pointer. Use
2551 free value_range_constant_singleton to remove need for
2552 op_with_constant_singleton_value delegator method. Do not
2553 create a vrp_prop class instance for every call! Narrow
2554 scope of a couple variables.
2555 (evrp_dom_walker::cleanup): Call methods in attached
2556 evrp_range_analyzer class via class instance pointer.
2557 * vr-values.h (class vr_values): Privatize many methods and
2560 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2562 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2563 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2564 * expr.c (expand_expr_real_2): Likewise.
2565 * fold-const.c (const_unop): Likewise.
2566 * optabs-tree.c (optab_for_tree_code): Likewise.
2567 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2568 * tree-inline.c (estimate_operator_cost): Likewise.
2569 * tree-pretty-print.c (dump_generic_node): Likewise.
2570 (op_code_prio): Likewise.
2571 (op_symbol_code): Likewise.
2572 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2573 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2574 * internal-fn.c (direct_internal_fn_optab): New function.
2575 (direct_internal_fn_array, direct_internal_fn_supported_p
2576 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2577 * fold-const-call.c (fold_const_reduction): New function.
2578 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2580 * tree-vect-loop.c: Include internal-fn.h.
2581 (reduction_code_for_scalar_code): Rename to...
2582 (reduction_fn_for_scalar_code): ...this and return an internal
2584 (vect_model_reduction_cost): Take an internal_fn rather than
2586 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2588 (vectorizable_reduction): Use internal functions rather than tree
2589 codes for the reduction operation. Update calls to the functions
2591 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2592 Use calls to internal functions rather than REDUC tree codes.
2593 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2595 2017-11-22 Olivier Hainque <hainque@adacore.com>
2597 * config/vxworks.c (vxworks_override_options): Pick default
2598 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2599 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2600 DWARF_GNAT_ENCODINGS_DEFAULT.
2601 * config/vxworksae.h: Likewise.
2603 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2605 PR tree-optimization/83104
2606 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2609 2017-11-22 Marek Polacek <polacek@redhat.com>
2610 H.J. Lu <hongjiu.lu@intel.com>
2611 Jason Merrill <jason@redhat.com>
2616 * calls.c (initialize_argument_information): Call
2617 warn_parameter_passing_abi target hook.
2618 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2619 argument onto stack.
2620 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2621 * common.opt: Update -fabi-version description.
2622 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2623 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2625 (ix86_is_empty_record): New function.
2626 (ix86_warn_parameter_passing_abi): New function.
2627 (TARGET_EMPTY_RECORD_P): Redefine.
2628 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2629 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2630 * doc/tm.texi: Regenerated.
2631 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2632 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2633 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2634 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2635 instead of int_size_in_bytes.
2636 * expr.c (copy_blkmode_to_reg): Likewise.
2637 * function.c (aggregate_value_p): Return 0 for empty types.
2638 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2640 (locate_and_pad_parm): Call arg size_in_bytes instead
2642 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2643 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2644 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2646 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2647 (std_gimplify_va_arg_expr): Skip empty records. Call
2648 arg_size_in_bytes instead size_in_bytes.
2649 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2650 * tree-core.h (tree_type_common): Add empty_flag.
2651 (tree_decl_common): Update comments.
2652 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2654 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2655 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2657 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2658 * tree.c (default_is_empty_type): New function.
2659 (default_is_empty_record): New function.
2660 (arg_int_size_in_bytes): New function.
2661 (arg_size_in_bytes): New function.
2662 (get_ultimate_context): New function.
2663 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2664 TRANSLATION_UNIT_WARN_EMPTY_P.
2665 (default_is_empty_record, arg_int_size_in_bytes,
2666 arg_size_in_bytes, get_ultimate_context): Declare.
2668 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2670 * config/arm/arm.c (cmse_clear_registers): New function.
2671 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2672 code by call to cmse_clear_registers.
2673 (cmse_nonsecure_entry_clear_before_return): Likewise.
2675 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2677 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2678 (vdot_s32, vdotq_s32): New.
2679 (vdot_lane_u32, vdotq_lane_u32): New.
2680 (vdot_lane_s32, vdotq_lane_s32): New.
2683 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2686 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2687 for unsigned values with fewer HWIs than the precision.
2688 (test_overflow): New function.
2689 (wide_int_cc_tests): Call it.
2691 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2693 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2694 has the same mode class as Pmode.
2696 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2698 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2699 padding_bits_to_clear_ptr.
2700 (cmse_nonsecure_entry_clear_before_return): Likewise.
2702 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2704 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2705 auto_sbitap instead of integer bitfield to control register needing
2708 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2710 PR tree-optimization/83044
2711 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2712 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2713 that need the upper bound. Subtract offset from
2714 get_addr_base_and_unit_offset only if positive and subtract it
2715 before division by eltsize rather than after it.
2718 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2719 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2720 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2724 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2725 expand_binop with *mul_widen_optab, make sure at least one of the
2726 operands doesn't have VOIDmode.
2729 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2731 PR rtl-optimization/82044
2732 PR tree-optimization/82042
2733 * dse.c (record_store): Check for overflow.
2734 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2735 add_wild_read instead of clear_rhs_from_active_local_stores on
2736 overflow and log it into dump_file.
2738 2017-11-22 Richard Biener <rguenther@suse.de>
2740 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2741 fake edges to exit when looking for a place to insert.
2742 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2744 (insert_into_preds_of_block): Commit edge insertion immediately,
2745 assert that doesn't require new BBs.
2746 (fini_pre): Release expressions.
2747 (pass_pre::execute): Shuffle things around a bit, if the fn
2748 is too large do not compute AVAIL either as this is really the
2751 2017-11-22 Richard Biener <rguenther@suse.de>
2753 PR tree-optimization/83089
2754 * tree-if-conv.c (pass_if_conversion::execute): If anything
2755 changed reset SCEV and free the number of iteration estimates.
2757 2017-11-21 Martin Sebor <msebor@redhat.com>
2759 PR tree-optimization/82945
2760 * calls.h (warn_nonstring_bound): Remove unused function.
2762 2017-11-21 Martin Sebor <msebor@redhat.com>
2764 PR tree-optimization/82945
2765 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2766 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2767 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2769 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2770 * calls.h (get_attr_nonstring_decl): Declare new function.
2771 * doc/extend.texi (attribute nonstring): Update.
2772 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2773 get_attr_nonstring_decl and handle it.
2774 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2775 detection of nul-termination.
2776 (strlen_to_stridx): Change to a pointer.
2777 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2778 (pass_strlen::execute): Same.
2780 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2782 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2783 for the new option -mprefer-vector-width=[none|128|256|512].
2784 * config/i386/i386.c (ix86_target_string): remove old style options
2785 -mprefer-avx256 and make -mprefer-avx128 as alias.
2786 (ix86_option_override_internal): Apply defaults for the
2787 -mprefer-vector-width=[128|256] option.
2788 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2789 Implement macros to work with -mprefer-vector-width=.
2790 * config/i386/i386.opt: Implemented option
2791 -mprefer-vector-width=[none|128|256|512].
2792 * doc/invoke.texi: Documentation for
2793 -mprefer-vector-width=[none|128|256|512].
2795 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2797 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2799 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2801 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2803 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2805 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2807 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2808 * expr.c (expand_expr_real_2): Likewise.
2809 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2810 fold_binary_loc): Likewise.
2811 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2812 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2813 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2814 MINUS_EXPR transformations.
2815 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2816 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2817 * tree-inline.c (estimate_operator_cost): Likewise.
2818 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2819 op_symbol_code): Likewise.
2820 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2821 * vr-values.c (extract_range_from_binary_expr): Likewise.
2822 * varasm.c (initializer_constant_valid_p_1): Likewise.
2823 * tree.def: New tree code POINTER_DIFF_EXPR.
2825 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2827 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2828 integer suffix to movbe mnemonic.
2829 (*bswaphi2_movbe): Ditto.
2830 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2832 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2835 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2836 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2837 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2839 2017-11-21 Martin Liska <mliska@suse.cz>
2841 * tree-inline.c (expand_call_inline): Remove not needed
2844 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2845 Jakub Jelinek <jakub@redhat.com>
2848 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2849 Remove static keyword from f variable.
2851 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2853 PR tree-optimization/83086
2854 * gimple-ssa-store-merging.c
2855 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2856 rather than n.base_addr.
2858 2017-11-21 Martin Liska <mliska@suse.cz>
2860 PR rtl-optimization/82044
2861 PR tree-optimization/82042
2862 * dse.c (check_mem_read_rtx): Check for overflow.
2864 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2866 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2869 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2871 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2872 and its corresponding call arg location note.
2874 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2875 Andrew Burgess <andrew.burgess@embecosm.com>
2877 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2879 (arc_return_slot_offset): Likewise.
2880 (arc_eh_return_address_location): New declaration.
2881 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2882 (MUST_SAVE_REGISTER): Add exception handler case.
2883 (MUST_SAVE_RETURN_ADDR): Likewise.
2884 (arc_frame_pointer_required): Likewise.
2885 (arc_frame_pointer_needed): New function.
2886 (arc_compute_frame_size): Changed.
2887 (arc_expand_prologue): Likewise.
2888 (arc_expand_epilogue): Likewise.
2889 (arc_initial_elimination_offset): Likewise.
2890 (arc_return_slot_offset): Delete.
2891 (arc_eh_return_address_location): New function.
2892 (arc_builtin_setjmp_frame_value): Likewise.
2893 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2894 (EH_RETURN_STACKADJ_RTX): Define.
2895 (EH_RETURN_HANDLER_RTX): Likewise.
2896 * config/arc/arc.md (eh_return): Delete.
2898 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2900 * print-tree.h (debug_vec_tree): Remove prototype.
2901 * gdbinit.in (pvt): Remove macro.
2903 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2905 PR tree-optimization/83047
2906 * gimple-ssa-store-merging.c
2907 (imm_store_chain_info::output_merged_store): If the loads with the
2908 same vuse are in different basic blocks, for load_gsi pick a load
2909 location that is dominated by the other loads.
2912 * config/i386/i386.c (ix86_memmodel_check): Start
2913 -Winvalid-memory-model diagnostics with lowercase letter.
2916 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2917 set in_cold_section_p to the partition containing loc_list->first.
2918 When seeing loc_list->last_before_switch node, update secname and
2919 perform range_across_switch second partition handling only after that.
2922 * run-rtl-passes.c: Include debug.h.
2923 (run_rtl_passes): Call debug_hooks->assembly_start.
2924 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2928 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2929 OPTAB_DIRECT in calls to expand_simple_binop.
2931 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2934 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2936 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2939 * spellcheck.h (best_match::blithely_get_best_candidate): New
2942 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2944 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2945 with lower case letter.
2947 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2949 * config/i386/i386.md (bswaphi2): New expander.
2950 (*bswaphi2_movbe): New insn pattern.
2951 (bswaphi -> rorhi pepehole2): New peephole pattern.
2953 2017-11-20 Jeff Law <law@redhat.com>
2955 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2956 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2957 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2958 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2959 gimple-ssa-evrp-analyze.h.
2961 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2962 set BB_VISITED here.
2963 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2965 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2967 * tree-predcom.c: Add general comment on Store-Store chains.
2968 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2970 (get_chain_last_ref_at): Rename into...
2971 (get_chain_last_write_at): ...this.
2972 (get_chain_last_write_before_load): New function.
2973 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2974 CT_STORE_STORE when write reference is added.
2975 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2976 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2977 (initialize_root_vars_store_elim_1): Ditto.
2978 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2979 definition is created.
2980 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2981 chain by replacing it with dominant stored value.
2983 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2985 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2987 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2989 * vr-values.c (extract_range_from_binary_expr): Use a full range
2992 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2994 * config/arm/arm.md (R4_REGNUM): Define constant.
2995 (nonsecure_call_internal): Remove r4 clobber.
2996 (nonsecure_call_value_internal): Likewise.
2997 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2998 clobber and resequence match_operands.
2999 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3000 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3001 (nonsecure_call_value_reg_thumb2): Likewise.
3003 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3005 PR tree-optimization/78821
3006 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3007 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3008 to the address of the base rather than the base itself.
3009 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3010 (find_bswap_or_nop_finalize): New function.
3011 (find_bswap_or_nop): Use it.
3012 (bswap_replace): Return a tree rather than bool, change first
3013 argument from gimple * to gimple_stmt_iterator, allow inserting
3014 into an empty sequence, allow ins_stmt to be NULL - then emit
3015 all stmts into gsi. Fix up MEM_REF address gimplification.
3016 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3017 (struct store_immediate_info): Add N and INS_STMT non-static
3019 (store_immediate_info::store_immediate_info): Initialize them
3020 from newly added ctor args.
3021 (merged_store_group::apply_stores): Formatting fixes. Sort by
3023 (stmts_may_clobber_ref_p): For stores call also
3024 refs_anti_dependent_p.
3025 (gather_bswap_load_refs): New function.
3026 (imm_store_chain_info::try_coalesce_bswap): New method.
3027 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3028 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3029 (imm_store_chain_info::output_merged_store): Fail if number of
3030 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3031 and NOP_EXPR rhs_code.
3032 (pass_store_merging::process_store): Compute n and ins_stmt, if
3033 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3034 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3036 (pass_store_merging::execute): Calculate dominators.
3038 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3039 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3040 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3041 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3042 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3043 class pass_optimize_bswap, bswap_replace,
3044 pass_optimize_bswap::execute): Moved to ...
3045 * gimple-ssa-store-merging.c: ... this file.
3046 Include optabs-tree.h.
3047 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3048 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3049 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3050 anonymous namespace, remove static keywords.
3051 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3052 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3054 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3057 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3059 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3061 * vec.h (debug_helper): New function.
3062 (DEFINE_DEBUG_VEC): New macro.
3063 * hash-set.h (debug_helper): New function.
3064 (DEFINE_DEBUG_HASH_SET): New macro.
3065 * cfg.c (debug_slim (edge)): New function.
3066 Call DEFINE_DEBUG_VEC for edges.
3067 Call DEFINE_DEBUG_HASH_SET for edges.
3068 * cfghooks.c (debug_slim (basic_block)): New function.
3069 Call DEFINE_DEBUG_VEC for basic blocks.
3070 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3071 * print-tree.c (debug_slim): New function to handle trees.
3072 Call DEFINE_DEBUG_VEC for trees.
3073 Call DEFINE_DEBUG_HASH_SET for trees.
3074 (debug (vec<tree, va_gc>) &): Remove.
3075 (debug (<vec<tree, va_gc>) *): Remove.
3076 * print-rtl.c (debug_slim): New function to handle const_rtx.
3077 Call DEFINE_DEBUG_VEC for rtx_def.
3078 Call DEFINE_DEBUG_VEC for rtx_insn.
3079 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3080 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3081 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3082 (debug (vec<rtx_insn *> *ptr): Remove.
3083 (debug_insn_vector): Remove.
3084 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3086 2017-11-20 Tom de Vries <tom@codesourcery.com>
3088 PR rtl-optimization/82020
3089 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3090 IF_THEN_ELSE condition.
3092 2017-11-19 Jeff Law <law@redhat.com>
3094 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3095 of degenerates resulting from ignoring an edge.
3097 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3100 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3102 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3105 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3106 for uninitialied counts.
3108 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3111 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3113 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3116 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3118 2017-11-19 Tom de Vries <tom@codesourcery.com>
3120 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3123 2017-11-19 Tom de Vries <tom@codesourcery.com>
3125 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3128 2017-11-19 Tom de Vries <tom@codesourcery.com>
3130 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3133 2017-11-19 Tom de Vries <tom@codesourcery.com>
3135 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3137 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3139 2017-11-19 Tom de Vries <tom@codesourcery.com>
3141 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3143 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3144 semicolon after MCORE_EXPORT_NAME call.
3146 2017-11-19 Tom de Vries <tom@codesourcery.com>
3149 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3150 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3152 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3154 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3155 (cgraph_edge::clone): Cleanup updating of profile.
3156 * ipa-cp.c (update_profiling_info): Likewise.
3157 * ipa-inline-transform.c (inline_transform): Likewise.
3158 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3159 * ipa-split.c (execute_split_functions): Do not split when function
3161 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3162 * profile-count.c (profile_count::dump): Add global0.
3163 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3165 (profile_count::to_sreal_scale): Likewise.
3166 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3167 (profile_count::combine_with_ipa_count): New function.
3168 * profile-count.h (profile_guessed_global0adjusted): New.
3169 (profile_count::adjusted_zero): New.
3170 (profile_count::global0adjusted): New.
3171 (profile_count::combine_with_ipa_count): New.
3172 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3173 correct profile of return block of split functions.
3174 (copy_cfg_body): Remove unused profile_count.
3175 (copy_body): Likewise.
3176 (expand_call_inline): Update.
3177 (tree_function_versioning): Update.
3179 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3181 * hash-set.h (hash_set::empty): New.
3182 * tree-ssa-threadbackward.h: Delete.
3183 * tree-ssa-threadbackward.c (class thread_jumps): New.
3184 Move max_threaded_paths into class.
3185 (fsm_find_thread_path): Remove arguments that are now in class.
3186 (profitable_jump_thread_path): Rename to...
3187 (thread_jumps::profitable_jump_thread_path): ...this.
3188 (convert_and_register_jump_thread_path): Rename to...
3189 (thread_jumps::convert_and_register_current_path): ...this.
3190 (check_subpath_and_update_thread_path): Rename to...
3191 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3192 (register_jump_thread_path_if_profitable): Rename to...
3193 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3194 (handle_phi): Rename to...
3195 (thread_jumps::handle_phi): ...this.
3196 (handle_assignment): Rename to...
3197 (thread_jumps::handle_assignment): ...this.
3198 (fsm_find_control_statement_thread_paths): Rename to...
3199 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3200 (find_jump_threads_backwards): Rename to...
3201 (thread_jumps::find_jump_threads_backwards): ...this.
3202 Initialize path local data.
3203 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3204 from within thread_jumps class.
3205 (pass_early_thread_jumps::execute): Same.
3207 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3209 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3211 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3213 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3214 to canonical location.
3216 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3219 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3221 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3223 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3225 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3226 rcrt1.o%s/grcrt1.o%s for -static-pie.
3228 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3230 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3231 ix86_shift_rotate_cost): Break out from ...
3232 (ix86_rtx_costs): ... here.
3233 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3236 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3238 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3239 when entry block was promoted unlikely.
3240 (estimate_bb_frequencies): Increase frequency scale.
3241 * profile-count.h (profile_count): Export precision info.
3243 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3245 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3246 disturbin profile of entry block.
3248 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3250 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3253 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3255 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3256 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3258 2017-11-17 Jeff Law <law@redhat.com>
3260 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3261 from evrp_dom_walker class. Various methods moved into new class.
3262 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3263 (evrp_range_analyzer::enter): New method.
3264 (evrp_range_analyzer::leave): New method.
3265 (evrp_dom_walker): Remove delegators no longer needed by this class.
3266 Replace vr_values data member with evrp_range_analyzer
3268 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3269 method extracted from evrp_dom_walker::before_dom_children.
3270 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3271 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3273 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3274 Add private copy constructor and move assignment operators.
3275 Privatize methods and class data where trivially possible.
3276 (evrp_dom_walker::cleanup): New function, extracted from
3277 execute_early_vrp. Simplify access to class data.
3279 * vr-values.h (get_output_for_vrp): Prototype.
3280 * vr-values.c (get_output_for_vrp): New function extracted from
3281 vrp_visit_assignment_or_call and extract_range_from_stmt.
3282 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3284 2017-11-17 Luis Machado <luis.machado@linaro.org>
3286 * config/aarch64/aarch64.c
3287 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3288 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3289 tune_params::AUTOPREFETCHER_WEAK.
3291 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3294 * config/arm/arm.c (arm_valid_target_attribute_rec):
3295 Parse "arch=" and "+<ext>".
3296 (arm_valid_target_attribute_tree): Re-init global options.
3297 (arm_option_override): Make non-static.
3298 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3299 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3300 (__ARM_FEATURE_CRC32): Support undef.
3301 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3302 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3304 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3306 * gdbinit.in (break-on-diagnostic): New command.
3308 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3310 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3311 return to indirect jump for EH return if control-flow
3312 protection is enabled. Change explicit 'false' argument in
3313 pro_epilogue_adjust_stack with a value of
3315 * config/i386/i386.md (simple_return_indirect_internal):
3316 Remove SImode restriction to support 64-bit.
3318 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3320 * combine.c (added_notes_insn): New.
3321 (try_combine): Handle added_notes_insn like added_links_insn.
3322 Rewrite return value code.
3323 (distribute_notes): Set added_notes_insn to the earliest insn we added
3326 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3328 PR rtl-optimization/82621
3329 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3330 dest of one of those SETs is unused.
3332 2017-11-17 Richard Biener <rguenther@suse.de>
3335 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3336 * tree-pretty-print.c (dump_generic_node): Handle
3337 annot_expr_parallel_kind.
3338 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3339 * gimplify.c (gimple_boolify): Likewise.
3341 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3343 * config.gcc (extra_headers): Add cet.h for x86 targets.
3344 * config/i386/cet.h: New file.
3345 * doc/install.texi: Add --enable-cet/--disable-cet.
3347 2017-11-17 Richard Biener <rguenther@suse.de>
3349 PR tree-optimization/83017
3350 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3351 (gen_parallel_loop): Properly count iterations.
3352 (parallelize_loops): Handle loop->can_be_parallel independent
3353 of flag_loop_parallelize_all. Make static profitability test match
3355 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3356 * invoke.texi (parloops-min-per-thread): Document.
3358 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3360 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3361 upstreaming review comments.
3363 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3365 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3366 with fast unaligned access.
3367 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3369 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3371 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3372 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3374 2017-11-17 Richard Biener <rguenther@suse.de>
3376 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3377 folding of references.
3379 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3382 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3383 of replacing call to strncmp with corresponding call to strcmp when
3386 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3388 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3389 option prefer-avx256 for skylake-avx512 configuration.
3390 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3391 (get_builtin_code_for_version): Ditto.
3393 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3394 Monk Chiang <sh.chiang04@gmail.com>
3396 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3397 (FIXED_REGISTERS): Reserve more register numbers.
3398 (CALL_USED_REGISTERS): Likewise.
3399 (REG_ALLOC_ORDER): Likewise.
3400 (REG_CLASS_CONTENTS): Likewise.
3401 (REGISTER_NAMES): Likewise.
3403 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3404 Kito Cheng <kito.cheng@gmail.com>
3406 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3408 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3410 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3412 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3413 parameter list for vec_splats.
3415 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3417 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3419 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3421 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3424 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3425 with the required version of automake 1.11.6.
3427 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3429 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3431 (DU_C2_3_power9): Correct reservation combinations.
3432 (FP_DIV_power9, VEC_DIV_power9): New.
3433 (power9-alu): Split out rotate/shift...
3434 (power9-rot): ...to here, correct dispatch resource.
3435 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3437 (power9-fp): Correct latency.
3438 (power9-sdiv): Add div/sqrt resource.
3439 (power9-ddiv): Correct latency, add div/sqrt resource.
3440 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3441 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3443 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3445 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3447 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3448 switch statement mapping KF built-ins to TF built-ins if we don't
3449 have the proper ISA 3.0 assembler support.
3451 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3453 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3454 (lower_emutls_function_body): Do not compute it.
3456 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3458 * ipa-split.c (split_bb_info): Turn time to sreal.
3459 (split_point): Likewise.
3460 (dump_split_point): Likewise.
3461 (fine_split_points): Likewise.
3462 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3464 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3466 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3469 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3471 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3473 (expensive_function_p): Remove useless assert.
3474 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3476 2017-11-16 Martin Sebor <msebor@redhat.com>
3478 PR tree-optimization/82588
3479 PR tree-optimization/82583
3480 * tree-vrp.c (check_array_ref): Handle flexible array members,
3481 string literals, and inner indices.
3482 (search_for_addr_array): Add detail to diagnostics.
3484 2017-11-16 Nathan Sidwell <nathan@acm.org>
3488 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3489 (SET_DECL_ASSEMBLER_NAME): Forward to
3490 overwrite_decl_assembler_name.
3491 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3492 (overwrite_decl_assembler_name): Declare.
3493 * tree.c (overwrite_decl_assembler_name): New.
3494 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3495 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3496 (LANG_HOOKS_INITIALIZER): Add it.
3497 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3498 * langhooks.c (lhd_set_decl_assembler_name): Use
3499 SET_DECL_ASSEMBLER_NAME.
3500 (lhd_overwrite_decl_assembler_name): Default implementation.
3502 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3503 Jackson Woodruff <jackson.woodruff@arm.com>
3505 PR tree-optimization/71026
3506 * match.pd: Canonicalize constant multiplies in division.
3508 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3510 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3511 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3512 Remove OPT_fomit_frame_pointer entry.
3513 * common/config/arc/arc-common.c: Likewise.
3514 * common/config/arm/arm-common.c: Likewise.
3515 * common/config/avr/avr-common.c: Likewise.
3516 * common/config/c6x/c6x-common.c: Likewise.
3517 * common/config/cr16/cr16-common.c: Likewise.
3518 * common/config/cris/cris-common.c: Likewise.
3519 * common/config/epiphany/epiphany-common.c: Likewise.
3520 * common/config/fr30/fr30-common.c: Likewise.
3521 * common/config/frv/frv-common.c: Likewise.
3522 * common/config/ia64/ia64-common.c: Likewise.
3523 * common/config/iq2000/iq2000-common.c: Likewise.
3524 * common/config/lm32/lm32-common.c: Likewise.
3525 * common/config/m32r/m32r-common.c: Likewise.
3526 * common/config/mcore/mcore-common.c: Likewise.
3527 * common/config/microblaze/microblaze-common.c: Likewise.
3528 * common/config/mips/mips-common.c: Likewise.
3529 * common/config/mmix/mmix-common.c: Likewise.
3530 * common/config/mn10300/mn10300-common.c: Likewise.
3531 * common/config/nios2/nios2-common.c: Likewise.
3532 * common/config/pa/pa-common.c: Likewise.
3533 * common/config/pdp11/pdp11-common.c: Likewise.
3534 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3535 * common/config/riscv/riscv-common.c: Likewise.
3536 * common/config/rs6000/rs6000-common.c: Likewise.
3537 * common/config/rx/rx-common.c: Likewise.
3538 * common/config/s390/s390-common.c: Likewise.
3539 * common/config/sh/sh-common.c: Likewise.
3540 * common/config/sparc/sparc-common.c: Likewise.
3541 * common/config/tilegx/tilegx-common.c: Likewise.
3542 * common/config/tilepro/tilepro-common.c: Likewise.
3543 * common/config/v850/v850-common.c: Likewise.
3544 * common/config/visium/visium-common.c: Likewise.
3545 * common/config/xstormy16/xstormy16-common.c: Likewise.
3546 * common/config/xtensa/xtensa-common.c: Likewise.
3547 * invoke.texi (-fomit-frame-pointer): Update documentation.
3549 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3551 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3553 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3555 * cfg.c (scale_bbs_frequencies_int,
3556 cale_bbs_frequencies_gcov_type): Remove.
3557 * cfg.h (scale_bbs_frequencies_int,
3558 cale_bbs_frequencies_gcov_type): Remove.
3560 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3562 * tree-ssa-loop-manip.c
3563 (scale_dominated_blocks_in_loop): Update to profile counts.
3564 (tree_transform_and_unroll_loop): Likewise.
3566 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3568 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3569 scale_bbs_frequencies_int.
3571 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3573 * final.c (compute_alignments): Use counts rather than frequencies.
3575 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3577 * cfgloopanal.c: Include sreal.h
3578 (average_num_loop_insns): Use counts and sreal for accounting.
3580 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3582 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3585 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3586 Kito Cheng <kito.cheng@gmail.com>
3588 * config/nds32/constraints.md: Provide more constraints.
3589 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3590 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3591 support constraints usage.
3593 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3595 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3597 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3598 Kito Cheng <kito.cheng@gmail.com>
3600 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3601 * config/nds32/nds32.opt: Refine the layout.
3602 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3603 TARGET_EXT_STRING): Support new options.
3604 * config/nds32/nds32.h: Likewise.
3605 * config/nds32/nds32.md: Likewise.
3606 * config/nds32/nds32-predicates.c: Likewise.
3607 * config/nds32/constraints.md: Likewise.
3608 * common/config/nds32/nds32-common.c: Likewise.
3610 2017-11-16 Julia Koval <julia.koval@intel.com>
3613 * config/i386/gfniintrin.h: Add sse check.
3614 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3616 2017-11-16 Julia Koval <julia.koval@intel.com>
3618 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3619 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3620 (ix86_handle_option): Handle -mavx512vbmi2.
3621 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3622 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3623 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3624 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3625 (ix86_valid_target_attribute_inner_p): Ditto.
3626 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3627 * config/i386/i386.opt (mavx512vbmi2): New option.
3628 * doc/invoke.texi: Add new option.
3630 2017-11-16 Julia Koval <julia.koval@intel.com>
3632 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3633 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3634 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3635 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3636 _mm512_gf2p8mul_epi8): New intrinsics.
3637 * config/i386/i386-builtin-types.def
3638 (V64QI_FTYPE_V64QI_V64QI): New type.
3639 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3640 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3641 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3642 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3643 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3644 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3646 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3648 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3649 explicitly as a stream of bytes.
3651 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3653 * config/rs6000/altivec.h (vec_xst_be): New #define.
3654 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3655 and externalize from *altivec_vperm_<mode>_internal.
3656 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3658 (XL_BE_V8HI): Likewise.
3659 (XL_BE_V4SI): Likewise.
3660 (XL_BE_V4SI): Likewise.
3661 (XL_BE_V2DI): Likewise.
3662 (XL_BE_V4SF): Likewise.
3663 (XL_BE_V2DF): Likewise.
3664 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3665 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3666 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3667 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3668 VSX_BUILTIN_VEC_XST_BE.
3669 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3670 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3672 (altivec_init_builtins): Replace conditional calls to def_builtin
3673 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3674 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3675 with unconditional calls. Remove calls to def_builtin for
3676 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3677 __builtin_vec_xst_be.
3678 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3679 to define_expand, and add alternate RTL generation for P8.
3680 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3681 vsx_ld_elemrev_v8hi.
3682 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3683 add alternate RTL generation for P8.
3684 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3685 vsx_ld_elemrev_v16qi.
3686 (vsx_st_elemrev_v8hi): Convert define_insn
3687 to define_expand, and add alternate RTL generation for P8.
3688 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3689 vsx_st_elemrev_v8hi.
3690 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3691 add alternate RTL generation for P8.
3692 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3693 vsx_st_elemrev_v16qi.
3695 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3698 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3699 TARGET_AVX512ER check.
3700 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3701 neither -mzeroupper nor -mno-zeroupper is used and
3702 TARGET_EMIT_VZEROUPPER is set.
3703 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3704 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3706 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3708 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3709 folding of vector compares.
3710 (fold_build_vec_cmp): New helper function.
3711 (fold_compare_helper): New helper function.
3712 (builtin_function_type): Add compare builtins to the list of functions
3713 having unsigned arguments. Cosmetic updates to comment indentation.
3714 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3715 the not+eq combination.
3717 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3719 PR tree-optimization/82726
3720 PR tree-optimization/70754
3721 * tree-predcom.c (order_drefs_by_pos): New function.
3722 (combine_chains): Move code setting has_max_use_after to...
3723 (try_combine_chains): ...here. New parameter. Sort combined chains
3724 according to position information.
3725 (tree_predictive_commoning_loop): Update call to above function.
3726 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3728 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3730 PR tree-optimization/82726
3732 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3734 PR tree-optimization/70754
3735 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3736 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3737 combined stmt before it if not NULL.
3738 (combine_chains): Process refs reversely and compute dominance point
3742 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3744 PR tree-optimization/79663
3745 * tree-predcom.c (combine_chains): Process refs in reverse order
3746 only for ZERO length chains, and add explaining comment.
3748 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3750 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3751 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3752 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3754 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3756 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3758 2017-11-15 Martin Liska <mliska@suse.cz>
3760 * tree-cfg.c (pass_warn_function_return::execute):
3761 Compare warn_return_type for greater than zero.
3763 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3767 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3768 to return true on Xeon and not on Xeon Phi.
3769 (ix86_check_avx256_register): Changed to ...
3770 (ix86_check_avx_upper_register): ... this. Add extra check for
3771 VALID_AVX512F_REG_OR_XI_MODE.
3772 (ix86_avx_u128_mode_needed): Changed
3773 ix86_check_avx256_register to ix86_check_avx_upper_register.
3774 (ix86_check_avx256_stores): Changed to ...
3775 (ix86_check_avx_upper_stores): ... this. Changed
3776 ix86_check_avx256_register to ix86_check_avx_upper_register.
3777 (ix86_avx_u128_mode_after): Changed
3778 avx_reg256_found to avx_upper_reg_found. Changed
3779 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3780 (ix86_avx_u128_mode_entry): Changed
3781 ix86_check_avx256_register to ix86_check_avx_upper_register.
3782 (ix86_avx_u128_mode_exit): Ditto.
3783 * config/i386/i386.h: (host_detect_local_cpu): New define.
3785 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3787 * config/arm/xgene1.md (xgene1): Split into automatons
3788 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3789 (xgene1_f_load): Adjust reservations and/or types.
3790 (xgene1_f_store): Likewise.
3791 (xgene1_load_pair): Likewise.
3792 (xgene1_store_pair): Likewise.
3793 (xgene1_fp_load1): Likewise.
3794 (xgene1_load1): Likewise.
3795 (xgene1_store1): Likewise.
3796 (xgene1_move): Likewise.
3797 (xgene1_alu): Likewise.
3798 (xgene1_simd): Likewise.
3799 (xgene1_bfm): Likewise.
3800 (xgene1_neon_load1): Likewise.
3801 (xgene1_neon_store1): Likewise.
3802 (xgene1_neon_logic): Likewise.
3803 (xgene1_neon_st1): Likewise.
3804 (xgene1_neon_ld1r): Likewise.
3805 (xgene1_alu_cond): Added.
3806 (xgene1_shift_reg): Likwise.
3807 (xgene1_bfx): Likewise.
3808 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3810 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3813 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3815 (can_widen_mult_without_libcall): New function.
3816 (expand_mul_overflow): If only checking unsigned mul overflow,
3817 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3818 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3819 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3820 (expand_DIVMOD): Formatting fix.
3821 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3822 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3823 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3825 PR tree-optimization/82977
3826 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3827 constructed temporary to strlen_to_stridx.put.
3829 2017-11-15 Martin Liska <mliska@suse.cz>
3831 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3832 * configure: Regenerate.
3834 2017-11-15 Martin Liska <mliska@suse.cz>
3837 * config/sh/sh-mem.cc: Use proper probability for
3840 2017-11-14 Jeff Law <law@redhat.com>
3842 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3843 the red zone for stack_clash_protection_final_dynamic_probe targets
3844 when the total dynamic stack size is zero bytes.
3846 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3847 blocks is post order.
3849 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3851 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3852 * final.c (rest_of_clean_state): Set it for the
3853 -fcompare-debug dump.
3854 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3855 class when TDF_COMPARE_DEBUG is set.
3857 * dwarf2out.c (gen_producer_string): Discard
3860 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3863 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3864 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3865 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3866 (__TGMATH_REAL_2_3): Remove macros.
3867 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3868 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3870 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3872 (remquo): Define using __TGMATH_REAL_3.
3874 2017-11-14 Jeff Law <law@redhat.com>
3876 * vr-values.c: New file with contents extracted from tree-vrp.c.
3877 * Makefile.in (OBJS): Add vr-values.o
3878 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3879 (set_value_range, set_and_canonicalize_value_range): Likewise.
3880 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3881 (value_range_constant_singleton, symbolic_range_p): Likewise.
3882 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3883 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3884 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3885 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3886 (find_case_label_range, find_case_label_index): Likewise.
3887 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3888 (range_int_cst_singleton_p, value_inside_range): Likewise.
3889 (get_single_symbol): Likewise.
3890 (switch_update): Move structure definition here.
3891 (to_remove_edges, to_update_switch_stmts): Provide externs.
3892 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3893 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3894 (vrp_val_is_min, set_value_range): Likewise.
3895 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3896 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3897 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3898 (range_is_nonnull, range_int_cst_p): Likewwise.
3899 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3900 (get_single_symbol, operand_less_p): Likewise
3901 (compare_values_warnv, compare_values): Likewise.
3902 (value_inside_range, value_range_constant_singleton): Likewise.
3903 (zero_nonzero_bitgs_from_vr): Likewise.
3904 (extract_range_from_binary_expr_1): Likewise.
3905 (overflow_comparison_p): Likewise.
3906 (to_remove_edges, to_update_switch_stmts): Likewise.
3907 (find_case_label-index, find_case_label_range): Likewise.
3908 (switch_update, set_value_range_to_nonnegative): Remove.
3909 (set_value_range_to_truthvalue): Likewise.
3910 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3911 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3912 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3913 (find_case_label_ranges, test_for_singularity): Likewise.
3914 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3915 (x_vr_values): Move to its remaining use site.
3917 2017-11-10 Jeff Law <law@redhat.com>
3919 * vr-values.h (VR_INITIALIZER): Move #define here.
3920 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3921 * Makefile.in (OBJS): Add tree-evrp.o
3922 * tree-vrp.h (assert_info): Move structure definition here.
3923 (set_value_range_to_varying): Prototype.
3924 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3925 (infer_value_range, register_edge_assert_for): Likewise.
3926 (stmt_interesting_for_vrp): Likewise.
3927 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3928 (set_value_range_to_varying): No longer static.
3929 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3930 (infer_value_range, register_edge_assert_for): Likewise.
3932 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3934 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3935 generate the XXBRD instruction.
3937 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3938 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3939 as being compatible if -mabi=ieeelongdouble.
3940 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3941 to setup float128 built-ins with hardware support.
3942 (BU_FLOAT128_HW_2): Likewise.
3943 (BU_FLOAT128_HW_3): Likewise.
3944 (BU_FLOAT128_HW_VSX_1): Likewise.
3945 (BU_FLOAT128_HW_VSX_2): Likewise.
3946 (scalar_extract_expq): Change float128 built-in functions to
3947 accommodate having both KFmode and TFmode functions. Use the
3948 KFmode variant as the default.
3949 (scalar_extract_sigq): Likewise.
3950 (scalar_test_neg_qp): Likewise.
3951 (scalar_insert_exp_q): Likewise.
3952 (scalar_insert_exp_qp): Likewise.
3953 (scalar_test_data_class_qp): Likewise.
3954 (sqrtf128_round_to_odd): Delete processing the round to odd
3955 built-in functions as special built-in functions, and define them
3956 as float128 built-ins. Use the KFmode variant as the default.
3957 (truncf128_round_to_odd): Likewise.
3958 (addf128_round_to_odd): Likewise.
3959 (subf128_round_to_odd): Likewise.
3960 (mulf128_round_to_odd): Likewise.
3961 (divf128_round_to_odd): Likewise.
3962 (fmaf128_round_to_odd): Likewise.
3963 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3964 support for KFmode and TFmode xststdcqp calls.
3965 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3966 point, switch the built-in handlers for the get/set float128
3967 exponent, get float128 mantissa, float128 test built-ins, and the
3968 float128 round to odd built-in functions. Eliminate creating the
3969 float128 round to odd built-in functions as special built-ins.
3970 (rs6000_init_builtins): Eliminate special creation of the float128
3971 round to odd built-in functions.
3972 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3973 function insns to support both TFmode and KFmode variants.
3974 (xsxsigqp_<mode>): Likewise.
3975 (xsiexpqpf_<mode>): Likewise.
3976 (xsiexpqp_<mode>): Likewise.
3977 (xststdcqp_<mode>): Likewise.
3978 (xststdcnegqp_<mode>): Likewise.
3979 (xststdcqp_<mode>): Likewise.
3981 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3983 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3984 unused path_in_freq_ptr parameter.
3985 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3987 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3989 * ipa-inline.c (edge_badness): Dump sreal frequency.
3990 (compute_inlined_call_time): Match natural implementaiton ...
3991 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3992 forgotten division by CGRAPH_FREQ_BASE.
3994 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3996 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3997 Solaris 11. Update comment.
3998 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4000 * config/sol2.h (STARTFILE_SPEC): Likewise.
4001 * configure: Regenerate.
4003 2017-11-14 Carl Love <cel@us.ibm.com>
4005 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4006 le_ and be_ prefixes to swap* variables. Remove
4007 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4010 2017-11-14 Jason Merrill <jason@redhat.com>
4012 Support GTY((cache)) on hash_map.
4013 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4014 (ggc_cache_remove): Override it instead of ggc_mx.
4015 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4016 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4017 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4018 (simple_cache_map_traits): Override maybe_mx.
4019 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4020 (hash_map): Friend gt_cleare_cache.
4021 (gt_cleare_cache): New.
4022 * tree.h (tree_cache_traits): New hash_map traits class.
4023 (tree_cache_map): New typedef.
4025 2017-11-14 Richard Biener <rguenther@suse.de>
4027 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4028 paramter and handling.
4029 (cleanup_control_flow_bb): Likewise.
4030 (cleanup_control_flow_pre): New helper performing a DFS walk
4031 to call cleanup_control_flow_bb in PRE order.
4032 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4033 via cleanup_control_flow_pre.
4035 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4037 * config/aarch64/aarch64-simd.md
4038 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4039 (*aarch64_simd_bsl<mode>_alt): Likewise.
4040 (aarch64_simd_bsldi_internal): New.
4041 (aarch64_simd_bsldi_alt): Likewise.
4043 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4045 * tracer.c (better_p): Do not compare frequencies.
4046 * reg-stack.c (better_edge): Likewise.
4047 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4050 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4052 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4053 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4054 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4055 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4056 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4057 * predict.c (maybe_hot_frequency_p): Inline to ...
4058 (maybe_hot_count_p): ... here; rewrite to counts.
4059 (counts_to_freqs): Rename to ...
4060 (update_max_bb_count): ... this one.
4061 (expensive_function_p): Use counts.
4062 (estimate_bb_frequencies): Update.
4063 (rebuild_frequencies): Update.
4064 * predict.h (counts_to_freqs): Rename to ...
4065 (update_max_bb_count): ... this one.
4066 * profile.c (compute_branch_probabilities): Add debug info
4067 * tree-inline.c (expand_call_inline): Update debug info.
4068 (optimize_inline_calls): Use update_max_bb_count..
4069 (tree_function_versioning): Use update_max_bb_count..
4070 * value-prof.c (gimple_value_profile_transformations):
4071 Do not use update_max_bb_count.
4073 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4075 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4076 always use frequencies.
4078 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4080 * bb-reorder.c: Remove frequencies from comments.
4081 (better_edge_p): Use profile counts.
4082 (find_traces): Dump profile counts.
4083 (rotate_loop): Use profile counts.
4084 (find_traces_1_round): Likewise.
4085 (connect_better_edge_p): Use counts instead of probabilities for
4087 (copy_bb_p): Drop early check for non-0 frequency.
4088 (sanitize_hot_paths): Update comments.
4090 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4092 * ipa-split.c (struct split_point): Add count.
4093 (consider_split): Do not compute incoming frequency; compute incoming
4094 count and store it to split_point.
4095 (split_function): Set count of the call to split part correctly.
4097 2017-11-13 Carl Love <cel@us.ibm.com>
4099 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4101 2017-11-13 Tom Tromey <tom@tromey.com>
4103 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4105 2017-11-13 Carl Love <cel@us.ibm.com>
4107 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4108 Add support for builtins:
4109 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4110 vector {un,}signed {char,int,short},
4111 vector {un,}signed {char,int,short}) arguments.
4112 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4113 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4114 Add BU_P9V_AV_2 expansions for the builtins.
4115 * config/rs6000/altivec.h (vec_first_match_index,
4116 vec_first_mismatch_index, vec_first_match_or_eos_index,
4117 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4118 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4119 new extern declaration.
4120 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4121 * config/rs6000/vsx.md (first_match_index_<mode>,
4122 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4123 first_mismatch_or_eos_index_<mode>): Add define expand.
4124 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4125 * doc/extend.texi: Update the built-in documenation file for the new
4128 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4130 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4131 into the min/max operations for _Float<N> and _Float<N>X types.
4133 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4136 * dwarf2out.c (do_eh_frame): New static variable.
4137 (dwarf2out_begin_prologue): Set it.
4138 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4140 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4142 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4144 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4146 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4149 * debug.h (dwarf2out_do_eh_frame): Declare.
4150 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4151 (dwarf2out_do_frame): Use it.
4152 (dwarf2out_do_cfi_asm): Likewise.
4153 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4154 (dwarf2out_assembly_start): Likewise.
4155 (dwarf2out_begin_prologue): Fix comment.
4156 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4157 if the target needs either debug or unwind DWARF2 info.
4158 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4159 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4160 -ftrapping-math, -ftrapv and -fwrapv.
4162 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4164 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4165 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4166 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4167 (estimate_edge_size_and_time): Likewise.
4168 (ipa_merge_fn_summary_after_inlining): Likewise.
4169 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4170 (compute_uninlined_call_time): Use sreal_frequency.
4171 (compute_inlined_call_time): Likewise.
4172 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4173 * profile-count.c: Include sreal.h.
4174 (profile_count::to_sreal_scale): New.
4175 * profile-count.h: Forward declare sreal.
4176 (profile_count::to_sreal_scale): Declare.
4178 2017-11-13 Nathan Sidwell <nathan@acm.org>
4180 * diagnostic.c (maybe_line_and_column): New.
4181 (diagnostic_get_location_text): Use it.
4182 (diagnostic_report_current_module): Likewise.
4183 (test_diagnostic_get_location_text): Add tests.
4185 2017-11-13 Luis Machado <luis.machado@linaro.org>
4187 * doc/md.texi (Specifying processor pipeline description): Fix
4188 incorrect latency for the div instruction example.
4190 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4192 PR tree-optimization/78821
4193 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4194 that bit_not_p is the same.
4195 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4196 (split_group): Count precisely bit_not_p bits in each statement.
4197 (invert_op): New function.
4198 (imm_store_chain_info::output_merged_store): Use invert_op to
4199 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4200 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4202 2017-11-13 Martin Liska <mliska@suse.cz>
4204 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4205 (struct source_info): Likewise.
4206 (add_branch_counts): Likewise.
4207 (add_line_counts): Likewise.
4208 (function_summary): Likewise.
4209 (output_intermediate_line): Likewise.
4210 (generate_results): Likewise.
4212 2017-11-13 Martin Liska <mliska@suse.cz>
4214 * gcov.c (struct block_info): Remove typedef for block_t.
4215 (struct line_info): Likewise.
4216 (line_info::has_block): Likewise.
4217 (EXIT_BLOCK): Likewise.
4218 (unblock): Likewise.
4219 (circuit): Likewise.
4220 (get_cycles_count): Likewise.
4221 (process_file): Likewise.
4222 (read_graph_file): Likewise.
4223 (solve_flow_graph): Likewise.
4224 (find_exception_blocks): Likewise.
4225 (add_line_counts): Likewise.
4226 (accumulate_line_info): Likewise.
4227 (output_line_details): Likewise.
4229 2017-11-13 Martin Liska <mliska@suse.cz>
4231 * gcov.c (struct arc_info): Remove typedef for arc_t.
4232 (struct line_info): Likewise.
4233 (add_branch_counts): Likewise.
4234 (output_branch_count): Likewise.
4235 (function_info::~function_info): Likewise.
4236 (circuit): Likewise.
4237 (output_intermediate_line): Likewise.
4238 (read_graph_file): Likewise.
4239 (solve_flow_graph): Likewise.
4240 (find_exception_blocks): Likewise.
4241 (add_line_counts): Likewise.
4242 (accumulate_line_info): Likewise.
4243 (output_line_details): Likewise.
4244 (output_function_details): Likewise.
4246 2017-11-13 Martin Liska <mliska@suse.cz>
4248 * gcov.c (struct function_info): Remove typedef for function_t.
4249 (struct source_info): Likewise.
4250 (source_info::get_functions_at_location): Likewise.
4251 (solve_flow_graph): Likewise.
4252 (find_exception_blocks): Likewise.
4253 (add_line_counts): Likewise.
4254 (output_intermediate_file): Likewise.
4255 (process_file): Likewise.
4256 (generate_results): Likewise.
4257 (release_structures): Likewise.
4258 (read_graph_file): Likewise.
4259 (read_count_file): Likewise.
4260 (accumulate_line_counts): Likewise.
4261 (output_lines): Likewise.
4263 2017-11-13 Martin Liska <mliska@suse.cz>
4265 * gcov.c (function_info::function_info): Remove num_counts
4266 and add vector<gcov_type>.
4267 (function_info::~function_info): Use the vector.
4268 (process_file): Likewise.
4269 (read_graph_file): Likewise.
4270 (read_count_file): Likewise.
4271 (solve_flow_graph): Likewise.
4273 2017-11-13 Martin Liska <mliska@suse.cz>
4275 * gcov.c (function_info::is_artificial): New function.
4276 (process_file): Erase all artificial early.
4277 (generate_results): Skip as all artificial are already
4280 2017-11-13 Martin Liska <mliska@suse.cz>
4282 * gcov.c (read_graph_file): Store to global vector of functions.
4283 (read_count_file): Iterate the vector.
4284 (process_file): Likewise.
4285 (generate_results): Likewise.
4286 (release_structures): Likewise.
4288 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4290 PR tree-optimization/82954
4291 * gimple-ssa-store-merging.c
4292 (imm_store_chain_info::coalesce_immediate_stores): If
4293 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4295 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4297 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4298 Upddate call to ENDIAN_LANE_N.
4299 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4300 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4301 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4302 and use aarch64_endian_lane_rtx.
4303 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4305 2017-11-12 Tom de Vries <tom@codesourcery.com>
4307 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4309 2017-11-12 Tom de Vries <tom@codesourcery.com>
4311 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4313 2017-11-12 Tom de Vries <tom@codesourcery.com>
4315 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4317 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4318 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4319 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4321 2017-11-11 Martin Sebor <msebor@redhat.com>
4324 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4327 * prefic.c (translate_name): Replace strncpy with memcpy to
4328 avoid -Wstringop-truncation.
4330 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4332 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4334 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4336 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4338 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4340 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4343 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4345 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4346 enable generating XXBRH if the value is in a vector register.
4347 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4348 value is in a vector register.
4349 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4350 register to register bswap64's instead of doing the GPR sequence
4351 used on previous machines.
4352 (bswapdi2_xxbrd): New insn.
4353 (bswapdi2_reg): Disallow on ISA 3.0.
4354 (register to register bswap64 splitter): Do not split the insn on
4355 ISA 3.0 systems that use XXBRD.
4357 2017-11-10 Martin Sebor <msebor@redhat.com>
4360 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4362 (find_subframework_file): Same.
4364 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4366 * auto-profile.c (afdo_indirect_call): Drop frequency.
4367 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4368 (cgraph_node::create_edge): Drop frequency argument.
4369 (cgraph_node::create_indirect_edge): Drop frequency argument.
4370 (cgraph_edge::make_speculative): Drop frequency arguments.
4371 (cgraph_edge::resolve_speculation): Do not update frequencies
4372 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4373 (cgraph_node::dump): Check consistency in IPA mode.
4374 (cgraph_edge::maybe_hot_p): Use IPA counter.
4375 (cgraph_edge::verify_count_and_frequency): Rename to ...
4376 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4377 (cgraph_node::verify_node): Update.
4378 * cgraph.h (struct cgraph_edge): Drop frequency.
4379 (cgraph_edge::frequency): New function.
4380 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4382 (cgraph_edge::rebuild_edges): Likewise.
4383 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4384 (duplicate_thunk_for_node): Do not pass frequency.
4385 (cgraph_node::create_clone): Scale only counts.
4386 (cgraph_node::create_virtual_clone): Do not pass frequency.
4387 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4388 (cgraph_node::create_version_clone): Do not pass frequency.
4389 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4390 (cgraph_node::expand_thunk): Do not pass frequency.
4391 (cgraph_node::create_wrapper): Do not pass frequency.
4392 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4394 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4395 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4396 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4397 (gather_caller_stats): Use frequency function.
4398 (ipcp_cloning_candidate_p): Use frequency function.
4399 (ipcp_propagate_stage): Use frequency function.
4400 (get_info_about_necessary_edges): Use frequency function.
4401 (update_profiling_info): Update only IPA profile.
4402 (update_specialized_profile): Use frequency functoin.
4403 (perhaps_add_new_callers): Update only IPA profile.
4404 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4405 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4406 (dump_ipa_call_summary): Use frequency function.
4407 (estimate_edge_size_and_time): Use frequency function.
4408 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4409 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4410 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4411 (update_noncloned_counts): ... ths one; scale counts only.
4412 (clone_inlined_nodes): Do not scale frequency.
4413 (inline_call): Do not pass frequency.
4414 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4415 (compute_inlined_call_time): Use IPA profile.
4416 (want_inline_small_function_p): Use IPA profile.
4417 (want_inline_self_recursive_call_p): Use IPA profile.
4418 (edge_badness): Use IPA profile.
4419 (lookup_recursive_calls): Use IPA profile.
4420 (recursive_inlining): Do not pass frequency.
4421 (resolve_noninline_speculation): Do not update frequency.
4422 (inline_small_functions): Collect max of IPA profile.
4423 (dump_overall_stats): Dump IPA porfile.
4424 (dump_inline_stats): Dump IPA porfile.
4425 (ipa_inline): Collect IPA stats.
4426 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4427 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4428 (ipa_propagate_frequency): Use frequency function.
4429 (ipa_profile): Cleanup.
4430 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4431 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4432 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4433 (input_node): Do not stream frequency.
4434 (input_edge): Do not stream frequency.
4435 (merge_profile_summaries): Scale only IPA profiles.
4436 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4437 * predict.c (drop_profile): Do not recompute frequency.
4438 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4439 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4440 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4441 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4442 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4443 * tree-inline.c (copy_bb): Do not scale frequency.
4444 (expand_call_inline): Do not scale frequency.
4445 (tree_function_versioning): Do not scale frequency.
4446 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4448 2017-11-10 Julia Koval <julia.koval@intel.com>
4450 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4451 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4452 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4453 (_mm256_maskz_gf2p8affine_epi64_epi8)
4454 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4455 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4456 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4457 (__builtin_ia32_vgf2p8affineqb_v32qi)
4458 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4459 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4461 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4465 (arm_option_override): Refactor.
4466 (arm_option_reconfigure_globals): New.
4467 (arm_options_perform_arch_sanity_checks): New.
4468 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4470 (arm_options_perform_arch_sanity_checks): Likewise
4472 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4474 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4475 (power9-qpmul): New.
4476 * rs6000/rs6000.md ("type" attr): Add qmul.
4477 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4478 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4479 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4481 2017-11-10 Martin Sebor <msebor@redhat.com>
4484 * builtins.c (compute_objsize): Handle arrays that
4485 compute_builtin_object_size likes to fail for. Make extern.
4486 * builtins.h (compute_objsize): Declare.
4487 (check_strncpy_sizes): New function.
4488 (expand_builtin_strncpy): Call check_strncpy_sizes.
4489 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4490 -Wstringop-truncation.
4491 (gimple_fold_builtin_strncat): Same.
4492 * gimple.c (gimple_build_call_from_tree): Set call location.
4493 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4494 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4495 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4496 (handle_builtin_strlen): Use strlen_to_stridx.
4497 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4499 Use strlen_to_stridx.
4500 (pass_strlen::execute): Release strlen_to_stridx.
4501 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4502 (-Wstringop-truncation): Document new option.
4504 2017-11-10 Martin Liska <mliska@suse.cz>
4506 PR gcov-profile/82702
4507 * gcov.c (main): Handle intermediate files in a different
4509 (get_gcov_intermediate_filename): New function.
4510 (output_gcov_file): Remove support of intermediate files.
4511 (generate_results): Allocate intermediate file.
4512 (release_structures): Clean-up properly fn_end.
4513 (output_intermediate_file): Start iterating with line 1.
4515 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4517 PR tree-optimization/82929
4518 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4519 ops_swapped_p non-static data member.
4520 (store_immediate_info::store_immediate_info): Clear it.
4521 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4522 ops set ops_swapped_p.
4523 (count_multiple_uses): Handle ops_swapped_p.
4525 2017-11-10 Martin Liska <mliska@suse.cz>
4527 * coverage.c (coverage_init): Stream information about
4528 support of has_unexecuted_blocks.
4529 * doc/gcov.texi: Document that.
4530 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4531 * gcov.c (read_graph_file): Likewise.
4532 (output_line_beginning): Fix a small issue with
4535 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4537 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4538 reference of trivial component.
4540 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4543 * gimple-ssa-store-merging.c
4544 (pass_store_merging::terminate_all_aliasing_chains): For
4545 gimple_store_p stmts also call refs_output_dependent_p.
4547 PR rtl-optimization/82913
4548 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4551 2017-11-09 Jeff Law <law@redhat.com>
4553 * vr-values.h: New file with vr_values class.
4554 * tree-vrp.c: Include vr-values.h
4555 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4556 data objects into the vr_values class.
4557 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4558 (get_value_range): Make it a member function within vr_values class.
4559 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4560 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4561 (op_with_constant_singleton_value_range): Likewise.
4562 (extract_range_for_var_from_comparison_expr): Likewise.
4563 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4564 (extract_range_from_binary_expr): Likewise.
4565 (extract_range_from_unary_expr): Likewise.
4566 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4567 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4568 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4569 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4570 (compare_name_with_value, compare_names): Likewise.
4571 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4572 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4573 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4574 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4575 (extract_range_from_phi_node): Likewise.
4576 (simplify_truth_ops_using_ranges): Likewise.
4577 (simplify_div_or_mod_using_ranges): Likewise.
4578 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4579 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4580 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4581 (simplify_float_conversion_using_ranges): Likewise.
4582 (simplify_internal_call_using_ranges): Likewise.
4583 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4584 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4585 poitner via x_vr_values for calls into gimple folder.
4586 (vrp_initialize_lattice): Make this the vr_values ctor.
4587 (vrp_free_lattice): Make this the vr_values dtor.
4588 (set_vr_value): New function.
4589 (class vrp_prop): Add vr_values data member. Add various member
4590 functions as well as member functions that delegate to vr_values.
4591 (check_array_ref): Make a member function within vrp_prop class.
4592 (search_for_addr_array, vrp_initialize): Likewise.
4593 (vrp_finalize): Likewise. Revamp to avoid direct access to
4594 vr_value, values_propagated, etc.
4595 (check_array_bounds): Extract vrp_prop class instance pointer from
4596 walk info structure. Use it to call member functions.
4597 (check_all_array_refs): Make a member function within vrp_prop class.
4598 Smuggle class instance pointer via walk info structure.
4599 (x_vr_values): New local static.
4600 (vrp_valueize): Use x_vr_values to get class instance.
4601 (vr_valueize_1): Likewise.
4602 (class vrp_folder): Add vr_values data member. Add various member
4603 functions as well as member functions that delegate to vr_values.
4604 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4605 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4606 class instance from vr_values. Use it to call member functions.
4607 (vrp_dom_walker): Add vr_values data member.
4608 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4609 instance via x_vr_values.
4610 (identify_jump_threads): Accept vr_values as argument. Store
4611 it into the walker structure.
4612 (evrp_dom_walker): Add vr_values class data member. Add various
4614 (evrp_dom_walker::try_find_new_range): Use vr_values data
4615 member to access the memory allocator.
4616 (evrp_dom_walker::before_dom_children): Store vr_values class
4617 instance into the vrp_folder class.
4618 (evrp_dom_walker::push_value_range): Rework to avoid direct
4619 access to num_vr_values and vr_value.
4620 (evrp_dom_walker::pop_value_range): Likewise.
4621 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4622 Use vr_values to get to dump_all_value_ranges member function.
4623 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4624 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4625 Pass vr_values class instance down to identify_jump_threads.
4626 Remove call to vrp_free_lattice.
4627 (debug_all_value_ranges): Remove.
4629 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4630 (vrp_folder): Likewise.
4632 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4633 Get it from the existing bitmap instead.
4634 (vrp_intersect_ranges_1): Likewise.
4636 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4638 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4640 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4641 set bit_not_p to it.
4642 (imm_store_chain_info::coalesce_immediate_stores): Break group
4643 if bit_not_p is different.
4644 (count_multiple_uses, split_group,
4645 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4646 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4647 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4648 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4649 to store_immediate_info ctor.
4651 2017-11-09 Jim Wilson <jimw@sifive.com>
4653 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4654 (scan_prog_file): Likewise.
4656 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4658 * bb-reorder.c (max_entry_frequency): Remove.
4659 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4660 connect_traces, push_to_next_round_p): Remove prototypes.
4661 (find_traces_1_round): Use counts only.
4662 (push_to_next_round_p): Likewise.
4663 (find_traces): Likewise.
4664 (rotate_loop): Likewise.
4665 (find_traces_1_round): Likewise.
4666 (connect_traces): Likewise.
4667 (edge_order): Likewise.
4669 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4671 * config/arm/arm.c (output_return_instruction): Add comments to
4672 indicate requirement for cmse_nonsecure_entry return to account
4673 for the size of clearing instruction output here.
4674 (thumb_exit): Likewise.
4675 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4676 return in hardfloat mode.
4678 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4680 * config/rs6000/rs6000.c (machine_function): Add a bool,
4681 "toc_is_wrapped_separately".
4682 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4683 if it wasn't explicitly set or unset, we are optimizing for speed, and
4684 doing separate shrink-wrapping.
4685 (rs6000_get_separate_components): Enable the TOC component if
4686 saving the TOC register in the prologue.
4687 (rs6000_components_for_bb): Handle the TOC component.
4688 (rs6000_emit_prologue_components): Store the TOC register where needed.
4689 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4690 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4692 2017-11-09 Martin Jambor <mjambor@suse.cz>
4694 * ipa-param-manipulation.c: New file.
4695 * ipa-param-manipulation.h: Likewise.
4696 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4697 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4698 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4699 (ipa_parm_adjustment): Likewise.
4700 (ipa_parm_adjustment_vec): Likewise.
4701 (ipa_get_vector_of_formal_parms): Moved declaration to
4702 ipa-param-manipulation.h.
4703 (ipa_get_vector_of_formal_parm_types): Likewise.
4704 (ipa_modify_formal_parameters): Likewise.
4705 (ipa_modify_call_arguments): Likewise.
4706 (ipa_combine_adjustments): Likewise.
4707 (ipa_dump_param_adjustments): Likewise.
4708 (ipa_modify_expr): Likewise.
4709 (ipa_get_adjustment_candidate): Likewise.
4710 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4711 ipa-param-manipulation.c.
4712 (ipa_get_vector_of_formal_parm_types): Likewise.
4713 (ipa_modify_formal_parameters): Likewise.
4714 (ipa_modify_call_arguments): Likewise.
4715 (ipa_modify_expr): Likewise.
4716 (get_ssa_base_param): Likewise.
4717 (ipa_get_adjustment_candidate): Likewise.
4718 (index_in_adjustments_multiple_times_p): Likewise.
4719 (ipa_combine_adjustments): Likewise.
4720 (ipa_dump_param_adjustments): Likewise.
4721 * tree-sra.c: Also include ipa-param-manipulation.h
4722 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4725 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4726 Alan Hayward <alan.hayward@arm.com>
4727 David Sherwood <david.sherwood@arm.com>
4729 * doc/sourcebuild.texi (vect_masked_store): Document.
4731 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4732 Alan Hayward <alan.hayward@arm.com>
4733 David Sherwood <david.sherwood@arm.com>
4735 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4737 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4738 Alan Hayward <alan.hayward@arm.com>
4739 David Sherwood <david.sherwood@arm.com>
4741 * doc/sourcebuild.texi (vect_variable_length): Document.
4743 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4744 Alan Hayward <alan.hayward@arm.com>
4745 David Sherwood <david.sherwood@arm.com>
4747 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4749 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4750 Alan Hayward <alan.hayward@arm.com>
4751 David Sherwood <david.sherwood@arm.com>
4753 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4755 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4756 Alan Hayward <alan.hayward@arm.com>
4757 David Sherwood <david.sherwood@arm.com>
4759 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4760 previously undocumented selectors.
4761 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4763 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4765 * doc/rtl.texi (const_vector): Say that elements can be
4766 const_wide_ints too.
4767 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4768 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4769 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4770 * optabs.c (expand_vector_broadcast): Likewise.
4772 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4773 Alan Hayward <alan.hayward@arm.com>
4774 David Sherwood <david.sherwood@arm.com>
4776 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4777 scaled index even if the unscaled address was invalid.
4778 Don't increase the complexity of using a scale in that case.
4780 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4781 Alan Hayward <alan.hayward@arm.com>
4782 David Sherwood <david.sherwood@arm.com>
4784 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4785 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4786 UNITS_PER_WORD bytes.
4787 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4788 based on REGMODE_NATURAL_SIZE of the inner mode.
4789 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4790 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4791 * expmed.c (lowpart_bit_field_p): Divide the value up into
4792 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4793 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4794 whether something is likely to occupy more than one register.
4796 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4799 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4800 profile_count::adjust_for_ipa_scaling.
4801 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4802 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4804 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4806 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4808 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4809 (split_group): Add total_orig and total_new arguments, estimate the
4810 number of statements related to the store group without store merging
4811 and with store merging.
4812 (imm_store_chain_info::output_merged_store): Adjust split_group
4813 callers, punt if estimated number of statements with store merging
4814 is not smaller than estimated number of statements without it.
4816 (handled_load): Remove has_single_use checks.
4817 (pass_store_merging::process_store): Likewise.
4819 2017-11-09 Richard Biener <rguenther@suse.de>
4821 PR tree-optimization/82902
4822 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4824 2017-11-09 Martin Liska <mliska@suse.cz>
4827 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4830 2017-11-09 Martin Liska <mliska@suse.cz>
4832 PR tree-optimization/82669
4833 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4835 2017-11-09 Martin Liska <mliska@suse.cz>
4837 PR gcov-profile/48463
4838 * coverage.c (coverage_begin_function): Output also end locus
4839 of a function and information whether the function is
4841 * gcov-dump.c (tag_function): Parse and print the information.
4842 * gcov.c (INCLUDE_MAP): Add include.
4843 (INCLUDE_SET): Likewise.
4844 (struct line_info): Move earlier in the source file because
4845 of vector<line_info> in function_info structure.
4846 (line_info::line_info): Likewise.
4847 (line_info::has_block): Likewise.
4848 (struct source_info): Add new member index.
4849 (source_info::get_functions_at_location): New function.
4850 (function_info::group_line_p): New function.
4851 (output_intermediate_line): New function.
4852 (output_intermediate_file): Use the mentioned function.
4853 (struct function_start): New.
4854 (struct function_start_pair_hash): Likewise.
4855 (process_file): Add code that identifies group functions.
4856 Assign lines either to global or function scope.
4857 (generate_results): Skip artificial functions.
4858 (find_source): Assign index for each source file.
4859 (read_graph_file): Read new flag artificial and end_line.
4860 (add_line_counts): Assign it either to global of function scope.
4861 (accumulate_line_counts): Isolate core of the function to
4862 accumulate_line_info and call it for both function and global
4864 (accumulate_line_info): New function.
4865 (output_line_beginning): Fix GNU coding style.
4866 (print_source_line): New function.
4867 (output_line_details): Likewise.
4868 (output_function_details): Likewise.
4869 (output_lines): Iterate both source (global) scope and function
4871 (struct function_line_start_cmp): New class.
4872 * doc/gcov.texi: Reflect changes in documentation.
4874 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4877 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4878 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4879 and similarly for not instead of neg.
4881 2017-11-08 Andi Kleen <ak@linux.intel.com>
4883 * config/i386/i386.opt: Add -mforce-indirect-call.
4884 * config/i386/predicates.md: Check for flag_force_indirect_call.
4885 * doc/invoke.texi: Document -mforce-indirect-call
4887 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4889 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4891 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4892 config/riscv/riscv.c (predict.h): New include.
4893 (riscv_slow_unaligned_access_p): No longer static.
4894 (riscv_block_move_straight): Add require.
4895 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4897 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4900 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4901 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4902 nonimmediate_operand predicate for operand 1 instead of
4905 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4907 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4909 * config/aarch64/constraints.md (Uml): New constraint.
4910 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4913 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4915 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4916 of two vec_duplicates into a vec_concat.
4918 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4920 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4921 Simplify vec_merge of vec_duplicate and vec_concat.
4922 * config/aarch64/constraints.md (Utq): New constraint.
4923 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4926 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4928 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4929 Simplify vec_merge of vec_duplicate and const_vector.
4930 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4932 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4933 mode iterator. Update predicate on operand 1 to
4934 handle non-const_vec constants. Delete constraints.
4935 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4937 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4939 PR tree-optimization/78821
4940 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4942 (store_operand_info::store_operand_info): Initialize it to false.
4943 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4944 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4945 store in the group, and if chain_info is non-NULL, to ignore altogether
4947 (compatible_load_p): Fail if bit_not_p does not match.
4948 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4949 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4950 (pass_store_merging::process_store): Adjust
4951 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4952 call terminate_all_aliasing_chains newly when adding a store into
4953 a chain with non-NULL chain_info.
4955 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4957 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4959 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4961 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4963 (aarch64_layout_frame): Initialise emit_frame_chain.
4964 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4965 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4967 2017-11-08 Martin Liska <mliska@suse.cz>
4969 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4970 of gimple_call_internal_p.
4972 2017-11-07 Tom de Vries <tom@codesourcery.com>
4974 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4976 2017-11-07 Tom de Vries <tom@codesourcery.com>
4978 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4981 2017-11-08 Martin Liska <mliska@suse.cz>
4984 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4986 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4988 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4990 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4993 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4994 (*cmp<mode>_ccz_1): New insn with $k alternative.
4997 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4998 RTX_COMM_COMPARE as commutative as well.
4999 (ix86_binary_operator_ok): Formatting fix.
5000 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5001 *<code><mode>3<mask_name><round_saeonly_name>,
5002 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5003 *<s>mul<mode>3_highpart<mask_name>,
5004 *vec_widen_umult_even_v16si<mask_name>,
5005 *vec_widen_umult_even_v8si<mask_name>,
5006 *vec_widen_umult_even_v4si<mask_name>,
5007 *vec_widen_smult_even_v16si<mask_name>,
5008 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5009 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5010 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5011 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5012 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5013 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5014 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5015 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5016 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5017 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5018 ix86_binary_operator_ok. Formatting fixes.
5019 (*<plusminus_insn><mode>3<mask_name><round_name>,
5020 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5023 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5025 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5026 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5027 eq and ne if TARGET_ISEL.
5028 (cmp): New code_iterator.
5029 (UNS, UNSU_, UNSIK): New code_attrs.
5030 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5031 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5033 ("eq<mode>3"): ... this.
5034 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5036 ("ne<mode>3"): ... this.
5038 2017-11-07 Julia Koval <julia.koval@intel.com>
5041 * common/config/i386/i386-common.c
5042 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5043 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5044 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5045 * config/i386/i386.opt: Ditto.
5046 * config/i386/i386.c (ix86_target_string): Ditto.
5047 (ix86_option_override_internal): Ditto.
5048 (ix86_init_mpx_builtins): Move MPX to args2.
5049 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5050 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5051 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5052 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5053 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5054 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5055 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5057 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5060 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5061 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5062 (zero-extendsidi peephole2): Remove peephole.
5064 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5067 * stor-layout.c: Include attribs.h.
5068 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5069 explicit lookup of "aligned" attribute.
5071 2017-11-07 Andrew Waterman <andrew@sifive.com>
5073 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5074 (riscv_expand_block_move): Likewise.
5075 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5077 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5078 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5079 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5080 (riscv_adjust_block_mem): Likewise.
5081 (riscv_block_move_loop): Likewise.
5082 (riscv_expand_block_move): Likewise.
5083 * config/riscv/riscv.md (movmemsi): New pattern.
5085 2017-11-07 Michael Clark <michaeljclark@mac.com>
5087 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5088 (MUSL_DYNAMIC_LINKER): Likewise.
5090 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5092 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5093 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5096 2017-11-07 Richard Biener <rguenther@suse.de>
5098 * match.pd: Fix build.
5100 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5101 Jackson Woodruff <jackson.woodruff@arm.com>
5103 PR tree-optimization/71026
5104 * match.pd: Canonicalize negate in division.
5106 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5109 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5111 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5113 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5114 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5116 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5118 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5119 non-scalar integral types.
5120 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5121 (-(A-B), -(~A)): New transformations.
5123 2017-11-07 Tom de Vries <tom@codesourcery.com>
5125 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5126 semicolon after "do {} while (0)".
5127 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5128 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5129 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5130 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5131 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5133 2017-11-07 Tom de Vries <tom@codesourcery.com>
5135 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5136 after "do {} while (0)".
5137 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5138 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5139 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5140 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5141 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5143 2017-11-07 Tom de Vries <tom@codesourcery.com>
5146 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5148 (arm_rtx_costs_internal): Add missing semicolon after
5149 HANDLE_NARROW_SHIFT_ARITH call.
5151 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5153 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5154 disable isel if it was not set explicitly.
5156 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5158 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5159 (add_type_attribute) likewise.
5161 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5163 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5164 of struct ix86_frame.
5165 (ix86_initial_elimination_offset): Likewise.
5166 (ix86_expand_split_stack_prologue): Likewise.
5168 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5170 * tree-vrp.h (enum value_range_type): Update stale comment.
5172 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5173 Alan Hayward <alan.hayward@arm.com>
5174 David Sherwood <david.sherwood@arm.com>
5176 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5177 (aarch64_expand_vec_perm_const): Take the number of units too.
5178 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5179 (aarch64_expand_vec_perm_const): Likewise.
5180 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5181 (vec_perm<mode>): Update accordingly.
5183 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5184 Alan Hayward <alan.hayward@arm.com>
5185 David Sherwood <david.sherwood@arm.com>
5187 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5188 Take the number of units too.
5189 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5190 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5191 but check for a vector mode before rather than after the call.
5192 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5193 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5194 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5195 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5196 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5197 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5198 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5199 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5200 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5201 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5202 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5203 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5204 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5205 (aarch64_sqdmull2_n<mode>): Update accordingly.
5207 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5208 Alan Hayward <alan.hayward@arm.com>
5209 David Sherwood <david.sherwood@arm.com>
5211 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5212 the number of units too.
5213 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5214 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5215 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5216 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5217 (vec_store_lanesxi<mode>): Update accordingly.
5219 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5220 Alan Hayward <alan.hayward@arm.com>
5221 David Sherwood <david.sherwood@arm.com>
5223 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5224 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5225 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5226 of units rather than the mode.
5227 * config/aarch64/iterators.md (nunits): New mode attribute.
5228 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5229 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5230 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5231 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5232 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5233 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5234 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5235 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5236 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5237 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5238 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5239 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5240 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5241 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5242 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5243 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5244 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5245 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5246 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5247 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5248 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5249 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5250 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5251 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5252 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5253 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5254 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5255 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5256 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5257 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5258 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5259 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5260 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5261 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5262 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5263 (aarch64_simd_vec_setv2di): Likewise.
5265 2017-11-06 Carl Love <cel@us.ibm.com>
5267 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5269 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5270 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5272 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5274 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5275 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5276 Add power 8 macro expansions.
5277 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5278 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5279 power 8 instructions. (VSX_XXBR): Add iterator.
5281 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5283 * config/arm/arm.md (predicable_short_it): Change default to "no",
5284 improve documentation, remove uses that are identical to the default.
5285 (enabled_for_depr_it): Rename to enabled_for_short_it.
5286 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5287 Remove default uses.
5288 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5289 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5290 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5291 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5293 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5296 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5297 float128 helper macros, which are no longer used after deleting
5298 the old 'q' built-in functions, and moving the round to odd
5299 built-in functions to being special built-in functions.
5300 (BU_FLOAT128_2): Likewise.
5301 (BU_FLOAT128_1_HW): Likewise.
5302 (BU_FLOAT128_2_HW): Likewise.
5303 (BU_FLOAT128_3_HW): Likewise.
5304 (FABSQ): Delete old 'q' built-in functions.
5305 (COPYSIGNQ): Likewise.
5306 (SQRTF128_ODD): Move round to odd built-in functions to be
5307 special built-in functions, so that we can handle
5308 -mabi=ieeelongdouble.
5309 (TRUNCF128_ODD): Likewise.
5310 (ADDF128_ODD): Likewise.
5311 (SUBF128_ODD): Likewise.
5312 (MULF128_ODD): Likewise.
5313 (DIVF128_ODD): Likewise.
5314 (FMAF128_ODD): Likewise.
5315 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5316 built-in names to 'f128'.
5317 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5318 old 'q' built-in functions, as the machine independent code for
5319 'f128' built-in functions handles this.
5320 (rs6000_expand_builtin): Add expansion for float128 round to odd
5321 functions, keying off on -mabi=ieeelongdouble of whether to use
5322 the KFmode or TFmode variant.
5323 (rs6000_init_builtins): Initialize the _Float128 round to odd
5325 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5326 _Float128 'q' built-in functions are now mapped into the new
5327 'f128' built-in functions.
5329 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5331 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5332 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5334 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5337 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5338 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5339 * toplev.c: Include "ipa-fnsummary.h".
5340 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5342 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5344 PR tree-optimization/82838
5345 * gimple-ssa-store-merging.c
5346 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5347 on a separate gimple_seq which is then appended to seq.
5349 2017-11-06 Jeff Law <law@redhat.com>
5352 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5353 (get_probe_interval): New functions.
5354 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5355 (ix86_adjust_stack_and_probe): Likewise.
5356 (output_adjust_stack_and_probe): Likewise.
5357 (ix86_emit_probe_stack_range): Likewise.
5358 (ix86_expand_prologue): Likewise.
5360 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5362 PR tree-optimization/82816
5363 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5364 if the modes of the two types are the same.
5365 (convert_plusminus_to_widen): Likewise.
5367 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5369 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5371 (usadv16qi): New define_expand.
5372 (usadv8hi): New define_expand.
5374 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5377 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5379 (inline_transform): Likewise.
5380 * predict.c (counts_to_freqs): Remove useless conditional.
5381 * profile-count.h (profile_count::apply_scale): Move sanity check.
5382 * tree-inline.c (copy_bb): Always scale.
5383 (copy_cfg_body): Likewise.
5385 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5388 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5390 (*cmp_ite0): Add enabled_for_depr_it attribute.
5391 (*cmp_ite1): Likewise.
5393 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5395 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5398 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5400 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5401 return the value by pointer.
5402 (extract_range_from_multiplicative_op_1): Update accordingly.
5403 Return as soon as an operation fails.
5405 2017-11-05 Tom de Vries <tom@codesourcery.com>
5408 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5409 (DEF_SANITIZER_BUILTIN): ... here.
5410 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5411 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5413 2017-11-05 Tom de Vries <tom@codesourcery.com>
5416 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5418 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5419 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5420 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5422 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5423 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5424 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5426 2017-11-05 Tom de Vries <tom@codesourcery.com>
5429 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5432 2017-11-04 Michael Clark <michaeljclark@mac.com>
5434 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5435 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5437 (*addsi3_extended): Likewise.
5438 (*addsi3_extended2): Likewise.
5439 (<optab>si3): Likewise.
5440 (<optab>di3): Likewise.
5441 (<optab><mode>3): Likewise.
5442 (<*optabe>si3_internal): Likewise.
5443 (zero_extendqi<SUPERQI:mode>2): Likewise.
5444 (*add<mode>hi3): Likewise.
5445 (*xor<mode>hi3): Likewise.
5446 (<optab>di3): Likewise.
5447 (*<optab>si3_extend): Likewise.
5448 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5449 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5450 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5452 2017-11-04 Andrew Waterman <andrew@sifive.com>
5454 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5455 TARGET_STRICT_ALIGN based upon -mtune argument.
5457 2017-11-04 Andrew Waterman <andrew@sifive.com>
5459 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5461 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5463 * config/i386/i386.c (choose_basereg): Use optional scratch
5464 register and add assertion.
5465 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5466 needed, and don't allocate stack.
5467 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5468 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5469 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5471 2017-11-03 Jeff Law <law@redhat.com>
5473 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5474 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5475 to probe at the start of a noreturn function.
5477 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5479 PR tree-optimization/78821
5480 * gimple-ssa-store-merging.c: Update the file comment.
5481 (MAX_STORE_ALIAS_CHECKS): Define.
5482 (struct store_operand_info): New type.
5483 (store_operand_info::store_operand_info): New constructor.
5484 (struct store_immediate_info): Add rhs_code and ops data members.
5485 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5486 and op1r arguments to the ctor, initialize corresponding data members.
5487 (struct merged_store_group): Add load_align_base and load_align
5489 (merged_store_group::merged_store_group): Initialize them.
5490 (merged_store_group::do_merge): Update them.
5491 (merged_store_group::apply_stores): Pick the constant for
5492 encode_tree_to_bitpos from one of the two operands, or skip
5493 encode_tree_to_bitpos if neither operand is a constant.
5494 (class pass_store_merging): Add process_store method decl. Remove
5495 bool argument from terminate_all_aliasing_chains method decl.
5496 (pass_store_merging::terminate_all_aliasing_chains): Remove
5497 var_offset_p argument and corresponding handling.
5498 (stmts_may_clobber_ref_p): New function.
5499 (compatible_load_p): New function.
5500 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5501 if there is overlap and rhs_code is not INTEGER_CST. For
5502 non-overlapping stores terminate group if rhs is not mergeable.
5503 (get_alias_type_for_stmts): Change first argument from
5504 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5505 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5506 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5508 (get_location_for_stmts): Change first argument from
5509 auto_vec<gimple *> & to vec<gimple *> &.
5510 (struct split_store): Remove orig_stmts data member, add orig_stores.
5511 (split_store::split_store): Create orig_stores rather than orig_stmts.
5512 (find_constituent_stmts): Renamed to ...
5513 (find_constituent_stores): ... this. Change second argument from
5514 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5515 to info structures rather than the statements.
5516 (split_group): Rename ALLOW_UNALIGNED argument to
5517 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5518 it. Adjust find_constituent_stores caller.
5519 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5520 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5521 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5522 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5523 (mem_valid_for_store_merging): New function.
5524 (handled_load): New function.
5525 (pass_store_merging::process_store): New method.
5526 (pass_store_merging::execute): Use process_store method. Adjust
5527 terminate_all_aliasing_chains caller.
5529 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5531 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5532 Return true for more constants, symbols and label references.
5533 (aarch64_valid_floating_const): Remove unused function.
5535 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5538 * config/aarch64/aarch64.c (aarch64_layout_frame):
5539 Undo forcing of LR at bottom of frame.
5541 2017-11-03 Jeff Law <law@redhat.com>
5544 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5545 for int_registers_saved.
5547 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5548 extracted from tree-ssa-dom.c.
5549 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5550 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5551 (record_equivalences_from_incoming_edge): Add additional argument
5552 to single_pred_edge_ignoring_loop_edges call.
5553 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5554 (uncprop_dom_walker::before_dom_children): Add additional argument
5555 to single_pred_edge_ignoring_loop_edges call.
5556 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5557 single_pred_edge_ignoring_loop_edges rather than open coding.
5558 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5560 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5562 * match.pd (-(-A)): Rewrite.
5564 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5566 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5567 (rs6000_emit_int_cmove): New declaration.
5568 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5569 (rs6000_emit_sISEL): Delete.
5570 (rs6000_emit_int_cmove): Make non-static.
5571 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5572 instead of rs6000_emit_sISEL.
5574 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5576 * asan.c (create_cond_insert_point): Maintain profile.
5577 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5579 * basic-block.h (struct basic_block_def): Remove frequency.
5580 (EDGE_FREQUENCY): Use to_frequency
5581 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5583 (find_traces): Update to use to_frequency.
5584 (find_traces_1_round): Likewise; use only IPA counts.
5585 (bb_to_key): Likewise.
5586 (connect_traces): Use IPA counts only.
5587 (copy_bb_p): Update to use to_frequency.
5588 (fix_up_crossing_landing_pad): Likewise.
5589 (sanitize_hot_paths): Likewise.
5590 * bt-load.c (basic_block_freq): Likewise.
5591 * cfg.c (init_flow): Set count_max to uninitialized.
5592 (check_bb_profile): Remove frequencies; check counts.
5593 (dump_bb_info): Do not dump frequencies.
5594 (update_bb_profile_for_threading): Update counts only.
5595 (scale_bbs_frequencies_int): Likewise.
5596 (MAX_SAFE_MULTIPLIER): Remove.
5597 (scale_bbs_frequencies_gcov_type): Update counts only.
5598 (scale_bbs_frequencies_profile_count): Update counts only.
5599 (scale_bbs_frequencies): Update counts only.
5600 * cfg.h (struct control_flow_graph): Add count-max.
5601 (update_bb_profile_for_threading): Update prototype.
5602 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5603 (find_many_sub_basic_blocks): Likewise.
5604 * cfgcleanup.c (try_forward_edges): Likewise.
5605 (try_crossjump_to_edge): Likewise.
5606 * cfgexpand.c (expand_gimple_cond): Likewise.
5607 (expand_gimple_tailcall): Likewise.
5608 (construct_init_block): Likewise.
5609 (construct_exit_block): Likewise.
5610 * cfghooks.c (verify_flow_info): Check consistency of counts.
5611 (dump_bb_for_graph): Do not dump frequencies.
5612 (split_block_1): Do not update frequencies.
5613 (split_edge): Do not update frequencies.
5614 (make_forwarder_block): Do not update frequencies.
5615 (duplicate_block): Do not update frequencies.
5616 (account_profile_record): Do not update frequencies.
5617 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5618 for global heuristics.
5619 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5620 (expected_loop_iterations_unbounded): Use counts only.
5621 * cfgloopmanip.c (scale_loop_profile): Simplify.
5622 (create_empty_loop_on_edge): Simplify
5624 (duplicate_loop_to_header_edge): Simplify
5625 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5626 (update_br_prob_note): Take care of removing note when profile
5628 (relink_block_chain): Do not dump frequency.
5629 (rtl_account_profile_record): Use to_frequency.
5630 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5631 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5632 (cgraph_update_edges_for_call_stmt_node): Likewise.
5633 (cgraph_edge::verify_count_and_frequency): Update.
5634 (cgraph_node::verify_node): Temporarily disable frequency verification.
5635 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5636 to_cgraph_frequency.
5637 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5638 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5640 (cgraph_node::expand_thunk): Update profile.
5641 * except.c (dw2_build_landing_pads): Do not update frequency.
5642 * final.c (compute_alignments): Use to_frequency.
5643 (dump_basic_block_info): Do not dump frequency.
5644 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5645 (dump_gimple_bb_header): Do not dump frequency.
5646 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5648 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5649 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5650 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5651 (init_before_recovery): Do not update frequency.
5652 (sched_create_recovery_edges): Do not update frequency.
5653 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5654 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5655 (ipa_cp_c_finalize): Set max_count to uninitialized.
5656 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5657 (param_change_prob): Use counts.
5658 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5660 * ipa-split.c (consider_split): Use to_frequency.
5661 (split_function): Use to_frequency.
5662 * ira-build.c (loop_compare_func): Likewise.
5663 (mark_loops_for_removal): Likewise.
5664 (mark_all_loops_for_removal): Likewise.
5665 * loop-doloop.c (doloop_modify): Do not update frequency.
5666 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5668 * lto-streamer-in.c (input_function): Update count_max.
5669 * omp-expand.c (expand_omp_taskreg): Update count_max.
5670 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5671 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5672 (maybe_hot_count_p): Use ipa counts only.
5673 (maybe_hot_bb_p): Simplify.
5674 (maybe_hot_edge_p): Simplify.
5675 (probably_never_executed): Do not take frequency argument.
5676 (probably_never_executed_bb_p): Do not pass frequency.
5677 (probably_never_executed_edge_p): Likewise.
5678 (combine_predictions_for_bb): Check that profile is nonzero.
5679 (propagate_freq): Do not set frequency.
5680 (drop_profile): Simplify.
5681 (counts_to_freqs): Simplify.
5682 (expensive_function_p): Use to_frequency.
5683 (propagate_unlikely_bbs_forward): Simplify.
5684 (determine_unlikely_bbs): Simplify.
5685 (estimate_bb_frequencies): Add hack to silence graphite issues.
5686 (compute_function_frequency): Use ipa counts.
5687 (pass_profile::execute): Update.
5688 (rebuild_frequencies): Use counts only.
5689 (force_edge_cold): Use counts only.
5690 * profile-count.c (profile_count::dump): Dump new count types.
5691 (profile_count::differs_from_p): Check compatiblity.
5692 (profile_count::to_frequency): New function.
5693 (profile_count::to_cgraph_frequency): New function.
5694 * profile-count.h (struct function): Declare.
5695 (enum profile_quality): Add profile_guessed_local and
5696 profile_guessed_global0.
5697 (class profile_proability): Decrease number of bits to 29;
5698 update from_reg_br_prob_note and to_reg_br_prob_note.
5699 (class profile_count: Update comment; decrease number of bits
5700 to 61. Check compatibility.
5701 (profile_count::compatible_p): New private member function.
5702 (profile_count::ipa_p): New member function.
5703 (profile_count::operator<): Handle global zero correctly.
5704 (profile_count::operator>): Handle global zero correctly.
5705 (profile_count::operator<=): Handle global zero correctly.
5706 (profile_count::operator>=): Handle global zero correctly.
5707 (profile_count::nonzero_p): New member function.
5708 (profile_count::force_nonzero): New member function.
5709 (profile_count::max): New member function.
5710 (profile_count::apply_scale): Handle IPA scalling.
5711 (profile_count::guessed_local): New member function.
5712 (profile_count::global0): New member function.
5713 (profile_count::ipa): New member function.
5714 (profile_count::to_frequency): Declare.
5715 (profile_count::to_cgraph_frequency): Declare.
5716 * profile.c (OVERLAP_BASE): Delete.
5717 (compute_frequency_overlap): Delete.
5718 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5719 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5720 * sched-ebb.c (rank): Use counts only.
5721 * shrink-wrap.c (handle_simple_exit): Use counts only.
5722 (try_shrink_wrapping): Use counts only.
5723 (place_prologue_for_one_component): Use counts only.
5724 * tracer.c (find_best_predecessor): Use to_frequency.
5725 (find_trace): Use to_frequency.
5726 (tail_duplicate): Use to_frequency.
5727 * trans-mem.c (expand_transaction): Do not update frequency.
5728 * tree-call-cdce.c: Do not update frequency.
5729 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5730 (gimple_merge_blocks): Likewise.
5731 (gimple_split_edge): Likewise.
5732 (gimple_duplicate_sese_region): Likewise.
5733 (gimple_duplicate_sese_tail): Likewise.
5734 (move_sese_region_to_fn): Likewise.
5735 (gimple_account_profile_record): Likewise.
5736 (insert_cond_bb): Likewise.
5737 * tree-complex.c (expand_complex_div_wide): Likewise.
5738 * tree-eh.c (lower_resx): Update profile.
5739 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5741 (initialize_cfun): Do not initialize frequencies
5742 (freqs_to_counts): Delete.
5743 (copy_cfg_body): Ignore count parameter.
5744 (copy_body): Update.
5745 (expand_call_inline): Update count_max.
5746 (optimize_inline_calls): Update count_max.
5747 (tree_function_versioning): Update count_max.
5748 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5749 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5751 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5752 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5753 (try_peel_loop): Likewise.
5754 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5756 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5757 (tree_transform_and_unroll_loop): Do not use frequencies
5758 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5759 Use reliable prediction only.
5760 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5761 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5762 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5763 probability scaling.
5764 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5765 not update frequency
5766 (any_remaining_duplicated_blocks): Likewise.
5767 (update_profile): Likewise.
5768 (estimated_freqs_path): Delete.
5769 (freqs_to_counts_path): Delete.
5770 (clear_counts_path): Delete.
5771 (ssa_fix_duplicate_block_edges): Likewise.
5772 (duplicate_thread_path): Likewise.
5773 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5774 * tree-tailcall.c (decrease_profile): Do not update frequency.
5775 (eliminate_tail_call): Likewise.
5776 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5777 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5778 (optimize_mask_stores): Likewise.
5779 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5780 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5781 (ubsan_expand_ptr_ifn): Update profile.
5782 * value-prof.c (gimple_ic): Simplify.
5783 * value-prof.h (gimple_ic): Update prototype.
5784 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5785 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5787 (want_inline_self_recursive_call_p): Likewise.
5788 (resolve_noninline_speculation): Only cummulate defined counts.
5789 (inline_small_functions): Use nonzero_p.
5790 (ipa_inline): Do not access freed node.
5792 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5794 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5795 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5797 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5799 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5800 non-legitimate address.
5802 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5804 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5805 (*lt0_<mode>di, *lt0_<mode>si): New.
5807 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5809 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5810 TARGET_PAIRED_FLOAT.
5812 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5813 Jim Wilson <jim.wilson@linaro.org>
5815 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5816 * config/aarch64/aarch64-tune.md: Regenerated.
5817 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5818 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5820 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5822 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5823 (arc_expand_prologue): Restore blink for millicode.
5824 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5826 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5829 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5830 gen_vec_duplicate after forcing the scalar into a register.
5832 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5834 * combine (try_combine): Print the insns input to try_combine to the
5837 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5840 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5841 Remove second argument from aarch64_process_target_attr call.
5842 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5844 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5846 (aarch64_handle_attr_arch): Remove second argument.
5847 (aarch64_handle_attr_cpu): Ditto.
5848 (aarch64_handle_attr_tune): Ditto.
5849 (aarch64_handle_attr_isa_flags): Ditto.
5850 (aarch64_process_one_target_attr): Ditto.
5851 (aarch64_process_target_attr): Ditto.
5852 (aarch64_option_valid_attribute_p): Remove second argument.
5853 on aarch64_process_target_attr call.
5855 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5857 * diagnostic.c: Include "selftest-diagnostic.h".
5858 (selftest::assert_location_text): New function.
5859 (selftest::test_diagnostic_get_location_text): New function.
5860 (selftest::diagnostic_c_tests): Call it.
5862 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5864 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5865 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5866 (class selftest::test_diagnostic_context): Move to...
5867 * selftest-diagnostic.c: New file.
5868 * selftest-diagnostic.h: New file.
5870 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5872 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5873 offset range for FT32B.
5874 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5875 * config/ft32/ft32.md: Add TARGET_NOPM.
5876 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5877 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5879 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5881 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5883 2017-11-02 Jeff Law <law@redhat.com>
5885 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5886 virtual keyword on FINAL OVERRIDE members.
5888 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5889 virtuals together. Add virtual destructor.
5890 (substitute_and_fold_engine): Similarly.
5892 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5894 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5896 2017-11-02 Richard Biener <rguenther@suse.de>
5898 PR tree-optimization/82795
5899 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5901 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5903 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5905 (gcc_GAS_CHECK_FEATURE): Remove.
5906 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5907 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5908 * configure: Regenerate.
5910 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5912 * config/arc/arc.c (hwloop_optimize): Account for empty
5915 2017-11-02 Richard Biener <rguenther@suse.de>
5918 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5919 Truncate ARRAY_REF index and element size.
5921 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5923 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5925 2017-11-01 Jeff Law <law@redhat.com>
5927 * tree-ssa-ccp.c (ccp_folder): New class derived from
5928 substitute_and_fold_engine.
5929 (ccp_folder::get_value): New member function.
5930 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5931 (ccp_fold_stmt): Remove prototype.
5932 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5933 * tree-ssa-copy.c (copy_folder): New class derived from
5934 substitute_and_fold_engine.
5935 (copy_folder::get_value): Renamed from get_value.
5936 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5937 * tree-vrp.c (vrp_folder): New class derived from
5938 substitute_and_fold_engine.
5939 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5940 (vrp_folder::get_value): New member function.
5941 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5942 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5943 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5944 provide a class interface to folder/substitute routines.
5945 (ssa_prop_fold_stmt_fn): Remove typedef.
5946 (ssa_prop_get_value_fn): Likewise.
5947 (subsitute_and_fold): Remove prototype.
5948 (replace_uses_in): Likewise.
5949 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5950 Renamed from replace_uses_in. Call the virtual member function
5951 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5952 (substitute_and_fold_dom_walker): Remove initialization of
5953 data member entries for calbacks. Add substitute_and_fold_engine
5954 member and initialize it.
5955 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5956 member functions for get_value, replace_phi_args_in c
5957 replace_uses_in, and fold_stmt calls.
5958 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5959 substitute_and_fold. Remove assert. Update ctor call.
5961 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5962 (ssa_prop_visit_phi_fn): Likewise.
5963 (class ssa_propagation_engine): New class to provide an interface
5965 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5967 (ssa_prop_visit_phi): Likewise.
5968 (ssa_propagation_engine::simulate_stmt): Moved into class.
5969 Call visit_phi/visit_stmt from the class rather than via
5970 file scoped static variables.
5971 (ssa_propagation_engine::simulate_block): Moved into class.
5972 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5973 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5974 set file scoped statics for the visit_stmt/visit_phi callbacks.
5975 * tree-complex.c (complex_propagate): New class derived from
5976 ssa_propagation_engine.
5977 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5978 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5979 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5981 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5982 ssa_propagation_engine.
5983 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5984 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5985 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5986 * tree-ssa-copy.c (copy_prop): New class derived from
5987 ssa_propagation_engine.
5988 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5989 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5990 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5991 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5992 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5993 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5994 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5996 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5998 PR rtl-optimization/82778
5999 PR rtl-optimization/82597
6000 * compare-elim.c (struct comparison): Add in_a_setter field.
6001 (find_comparison_dom_walker::before_dom_children): Remove killed
6002 bitmap and df_simulate_find_defs call, instead walk the defs.
6003 Compute last_setter and initialize in_a_setter. Merge definitions
6004 with first initialization for a few variables.
6005 (try_validate_parallel): Use insn_invalid_p instead of
6006 recog_memoized. Return insn rather than just the pattern.
6007 (try_merge_compare): Fix up comment. Don't uselessly test if
6008 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6010 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6013 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6014 Alan Hayward <alan.hayward@arm.com>
6015 David Sherwood <david.sherwood@arm.com>
6017 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6018 aarch64_hard_regno_nregs to get the number of registers
6021 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6022 Alan Hayward <alan.hayward@arm.com>
6023 David Sherwood <david.sherwood@arm.com>
6025 * config/aarch64/constraints.md (Upl): Rename to...
6027 * config/aarch64/aarch64.md
6028 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6031 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6032 Alan Hayward <alan.hayward@arm.com>
6033 David Sherwood <david.sherwood@arm.com>
6035 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6036 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6039 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6040 Alan Hayward <alan.hayward@arm.com>
6041 David Sherwood <david.sherwood@arm.com>
6043 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6044 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6045 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6047 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6048 of a single element.
6049 * config/aarch64/iterators.md: Add a comment above the permute
6050 unspecs to say that they are generated directly by
6051 aarch64_expand_vec_perm_const.
6052 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6054 2017-11-01 Nathan Sidwell <nathan@acm.org>
6056 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6058 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6060 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6061 and medany code models, and describe what they do.
6063 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6065 Revert accidental duplicate:
6067 * combine.c (can_change_dest_mode): Reject changes in
6068 REGMODE_NATURAL_SIZE.
6070 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6072 PR rtl-optimization/64682
6073 PR rtl-optimization/69567
6074 PR rtl-optimization/69737
6075 PR rtl-optimization/82683
6076 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6077 register mentioned in the note, drop the note, unless it came from I3,
6078 in which case it should go to I3 again.
6080 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6082 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6083 and return false if not.
6084 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6086 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6088 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6089 range is known to be empty.
6091 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6092 Alan Hayward <alan.hayward@arm.com>
6093 David Sherwood <david.sherwood@arm.com>
6095 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6096 and CONST_VECTOR_NUNITS instead of computing the number of units from
6097 the byte sizes of the vector and element.
6098 (simplify_binary_operation_1): Likewise.
6099 (simplify_const_binary_operation): Likewise.
6100 (simplify_ternary_operation): Likewise.
6102 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6103 Alan Hayward <alan.hayward@arm.com>
6104 David Sherwood <david.sherwood@arm.com>
6106 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6107 (int_mem_offset): ...this new function.
6108 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6109 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6110 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6113 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6114 Alan Hayward <alan.hayward@arm.com>
6115 David Sherwood <david.sherwood@arm.com>
6117 * lower-subreg.c (interesting_mode_p): New function.
6118 (compute_costs, find_decomposable_subregs, decompose_register)
6119 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6120 (resolve_clobber, dump_choices): Use it.
6122 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6123 Alan Hayward <alan.hayward@arm.com>
6124 David Sherwood <david.sherwood@arm.com>
6126 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6128 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6129 Alan Hayward <alan.hayward@arm.com>
6130 David Sherwood <david.sherwood@arm.com>
6132 * alias.c (find_base_value, find_base_term): Only process integer
6133 truncations. Check the precision rather than the size.
6135 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6136 Alan Hayward <alan.hayward@arm.com>
6137 David Sherwood <david.sherwood@arm.com>
6139 * machmode.h (is_narrower_int_mode): New function
6140 * optabs.c (expand_float, expand_fix): Use it.
6141 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6143 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6144 Alan Hayward <alan.hayward@arm.com>
6145 David Sherwood <david.sherwood@arm.com>
6147 * rtl.h (narrower_subreg_mode): New function.
6148 * ira-color.c (update_costs_from_allocno): Use it.
6150 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6151 Alan Hayward <alan.hayward@arm.com>
6152 David Sherwood <david.sherwood@arm.com>
6154 * optabs-query.h (convert_optab_p): New function, split out from...
6155 (convert_optab_handler): ...here.
6156 (widening_optab_handler): Delete.
6157 (find_widening_optab_handler): Remove permit_non_widening parameter.
6158 (find_widening_optab_handler_and_mode): Likewise. Provide an
6159 override that operates on mode class wrappers.
6160 * optabs-query.c (widening_optab_handler): Delete.
6161 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6162 parameter. Assert that the two modes are the same class and that
6163 the "from" mode is narrower than the "to" mode. Use
6164 convert_optab_handler instead of widening_optab_handler.
6165 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6166 instead of widening_optab_handler.
6167 * expr.c (expand_expr_real_2): Update calls to
6168 find_widening_optab_handler.
6169 * optabs.c (expand_widen_pattern_expr): Likewise.
6170 (expand_binop_directly): Take the insn_code as a parameter.
6171 (expand_binop): Only call find_widening_optab_handler for
6172 conversion optabs; use optab_handler otherwise. Update calls
6173 to find_widening_optab_handler and expand_binop_directly.
6174 Use convert_optab_handler instead of widening_optab_handler.
6175 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6176 find_widening_optab_handler and use scalar_mode rather than
6178 (convert_plusminus_to_widen): Likewise.
6180 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6181 Alan Hayward <alan.hayward@arm.com>
6182 David Sherwood <david.sherwood@arm.com>
6184 * machmode.h (fixed_size_mode): New class.
6185 * rtl.h (get_pool_mode): Return fixed_size_mode.
6186 * gengtype.c (main): Add fixed_size_mode.
6187 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6188 (get_raw_arg_mode): Likewise.
6189 * doc/tm.texi: Regenerate.
6190 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6191 * targhooks.c (default_get_reg_raw_mode): Likewise.
6192 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6193 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6194 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6195 (msp430_get_raw_result_mode): Likewise.
6196 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6197 * dbxout.c (dbxout_parms): Require fixed-size modes.
6198 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6199 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6200 * omp-low.c (lower_oacc_reductions): Likewise.
6201 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6202 (simplify_subreg): Update accordingly.
6203 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6204 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6205 that aren't fixed-size.
6206 (get_pool_mode): Return a fixed_size_mode.
6207 (output_constant_pool_2): Take a fixed_size_mode.
6209 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6210 Alan Hayward <alan.hayward@arm.com>
6211 David Sherwood <david.sherwood@arm.com>
6213 * doc/rtl.texi (vec_series): Document.
6214 (const): Say that the operand can be a vec_series.
6215 * rtl.def (VEC_SERIES): New rtx code.
6216 * rtl.h (const_vec_series_p_1): Declare.
6217 (const_vec_series_p): New function.
6218 * emit-rtl.h (gen_const_vec_series): Declare.
6219 (gen_vec_series): Likewise.
6220 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6221 (gen_vec_series): Likewise.
6222 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6223 * simplify-rtx.c (simplify_unary_operation): Handle negations
6225 (simplify_binary_operation_series): New function.
6226 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6227 (test_vector_ops_series): New function.
6228 (test_vector_ops): Call it.
6229 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6230 gen_const_vec_series.
6231 (altivec_lvsr): Likewise.
6232 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6234 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6235 Alan Hayward <alan.hayward@arm.com>
6236 David Sherwood <david.sherwood@arm.com>
6238 * doc/rtl.texi (const): Update description of address constants.
6239 Say that vector constants are allowed too.
6240 * common.md (E, F): Use CONSTANT_P instead of checking for
6242 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6243 checking for CONST_VECTOR.
6244 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6246 * expr.c (expand_expr_real_2): Check for vector modes instead
6247 of checking for CONST_VECTOR.
6248 * rtl.h (const_vec_p): New function.
6249 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6250 (unwrap_const_vec_duplicate): Handle them here too.
6252 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6253 David Malcolm <dmalcolm@redhat.com>
6254 Alan Hayward <alan.hayward@arm.com>
6255 David Sherwood <david.sherwood@arm.com>
6257 * rtl.h (vec_duplicate_p): New function.
6258 * selftest-rtl.c (assert_rtx_eq_at): New function.
6259 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6260 (assert_rtx_eq_at): Declare.
6261 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6262 * selftest-run-tests.c (selftest::run_tests): Call it.
6263 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6264 (simplify_unary_operation_1): Recursively handle vector duplicates.
6265 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6267 (simplify_subreg): Handle subregs of vector duplicates.
6268 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6269 (selftest::simplify_rtx_c_tests): New functions.
6271 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6272 Alan Hayward <alan.hayward@arm.com>
6273 David Sherwood <david.sherwood@arm.com>
6275 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6276 (gen_vec_duplicate): Likewise.
6277 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6279 (gen_const_vector): ...here.
6280 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6281 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6282 whose elements are all equal.
6283 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6284 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6285 (simplify_relational_operation): Likewise.
6286 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6288 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6289 (aarch64_expand_vector_init): Likewise.
6290 * config/arm/arm.c (neon_vdup_constant): Likewise.
6291 (neon_expand_vector_init): Likewise.
6292 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6293 (arm_block_set_unaligned_vect): Likewise.
6294 (arm_block_set_aligned_vect): Likewise.
6295 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6296 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6297 (expand_vec_perm_even_odd_pack): Likewise.
6298 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6299 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6300 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6301 gen_const_vec_duplicate.
6302 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6303 * config/mips/mips.c (mips_gen_const_int_vector): Use
6304 gen_const_vec_duplicate.
6305 (mips_expand_vector_init): Use CONST0_RTX.
6306 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6307 (define_split): Use gen_const_vec_duplicate.
6308 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6309 (define_split): Use gen_const_vec_duplicate.
6310 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6311 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6312 * config/spu/spu.c (spu_const): Likewise.
6314 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6315 Alan Hayward <alan.hayward@arm.com>
6316 David Sherwood <david.sherwood@arm.com>
6318 * combine.c (can_change_dest_mode): Reject changes in
6319 REGMODE_NATURAL_SIZE.
6321 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6323 * configure.ac (--enable-libssp): New.
6324 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6325 trying to determine target-specific default. Adjust indentation.
6326 * configure: Regenerated.
6327 * doc/install.texi (Configuration): Expand --disable-libssp
6330 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6332 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6335 2017-10-31 Martin Jambor <mjambor@suse.cz>
6338 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6340 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6342 * auto-profile.c (autofdo_source_profile::read): Use
6343 UNKNOWN_LOCATION rather than 0.
6344 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6345 (warning_at): ...this overload.
6346 (warning_at_rich_loc_n): Rename to...
6347 (warning_n): ...this overload.
6348 (error_at_rich_loc): Rename to...
6349 (error_at): ...this overload.
6350 (pedwarn_at_rich_loc): Rename to...
6351 (pedwarn): ...this overload.
6352 (permerror_at_rich_loc): Rename to...
6353 (permerror): ...this overload.
6354 (inform_at_rich_loc): Rename to...
6355 (inform): ...this overload.
6356 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6357 (diagnostic_n_impl_richloc): Rename to...
6358 (diagnostic_n_impl): ...this rich_location *-based decl.
6359 (inform_at_rich_loc): Rename to...
6360 (inform): ...this, and add an assertion.
6361 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6362 (warning_at_rich_loc): Rename to...
6363 (warning_at): ...this, and add an assertion.
6364 (warning_at_rich_loc_n): Rename to...
6365 (warning_n): ...this, and add an assertion.
6366 (warning_n): Update location_t-based implementation for removal of
6367 location_t-based diagnostic_n_impl.
6368 (pedwarn_at_rich_loc): Rename to...
6369 (pedwarn): ...this, and add an assertion.
6370 (permerror_at_rich_loc): Rename to...
6371 (permerror): ...this, and add an assertion.
6372 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6373 (error_at_rich_loc): Rename to...
6374 (error_at): ...this, and add an assertion.
6375 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6376 (driver::do_spec_on_infiles): Likewise.
6377 * substring-locations.c (format_warning_va): Update for renaming
6378 of inform_at_rich_loc.
6380 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6382 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6383 _Float<N>X built-in functions so that the variant without the
6384 "__builtin_" prefix is only enabled for the GNU C and Objective C
6385 languages when they are in non-strict ANSI/ISO mode.
6386 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6387 * target.def (floatn_builtin_p): Add a target hook to control
6388 whether _Float<N> and _Float<N>X built-in functions without the
6389 "__builtin_" prefix are enabled, and return true for C and
6390 Objective C in the default hook. Include langhooks.h in
6392 * targhooks.h (default_floatn_builtin_p): Likewise.
6393 * targhooks.c (default_floatn_builtin_p): Likewise.
6394 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6395 floatn_builtin_p target hook.
6396 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6398 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6399 Eric Botcazou <ebotcazou@adacore.com>
6401 PR rtl-optimization/81803
6402 * lra-constraints.c (curr_insn_transform): Also reload the whole
6403 register for a strict subreg no wider than a word if this is for
6404 a WORD_REGISTER_OPERATIONS target.
6406 2017-10-31 Jason Merrill <jason@redhat.com>
6408 * gdbinit.in: Skip over inlines from timevar.h.
6410 2017-10-31 Martin Liska <mliska@suse.cz>
6412 * doc/gcov.texi: Document new option.
6413 * gcov.c (print_usage): Likewise print it.
6414 (process_args): Support the argument.
6415 (format_count): New function.
6416 (format_gcov): Use the function.
6418 2017-10-31 Martin Liska <mliska@suse.cz>
6420 * gcov.c (struct name_map): do not use typedef.
6421 Define operator== and operator<.
6422 (name_search): Remove.
6423 (name_sort): Remove.
6424 (main): Do not allocate names.
6425 (process_file): Add vertical space.
6426 (generate_results): Use std::find.
6427 (release_structures): Do not release memory.
6428 (find_source): Use std::find.
6430 2017-10-31 Martin Liska <mliska@suse.cz>
6432 * gcov.c (struct line_info): Remove it's typedef.
6433 (line_info::line_info): Add proper ctor.
6434 (line_info::has_block): Do not use a typedef.
6435 (struct source_info): Do not use typedef.
6436 (circuit): Likewise.
6437 (get_cycles_count): Likewise.
6438 (output_intermediate_file): Iterate via vector iterator.
6439 (add_line_counts): Use std::vector methods.
6440 (accumulate_line_counts): Likewise.
6441 (output_lines): Likewise.
6443 2017-10-31 Martin Liska <mliska@suse.cz>
6445 * gcov.c (struct source_info): Remove typedef.
6446 (source_info::source_info): Add proper ctor.
6447 (accumulate_line_counts): Use struct, not it's typedef.
6448 (output_gcov_file): Likewise.
6449 (output_lines): Likewise.
6450 (main): Do not allocate an array.
6451 (output_intermediate_file): Use size of vector container.
6452 (process_file): Resize the vector.
6453 (generate_results): Do not preallocate, use newly added vector
6455 (release_structures): Do not release sources.
6456 (find_source): Use vector methods.
6457 (add_line_counts): Do not use typedef.
6459 2017-10-31 Martin Liska <mliska@suse.cz>
6461 * doc/gcov.texi: Document that.
6462 * gcov.c (add_line_counts): Mark lines with a non-executed
6464 (output_line_beginning): Handle such lines.
6465 (output_lines): Pass new argument.
6466 (output_intermediate_file): Print it in intermediate format.
6468 2017-10-31 Martin Liska <mliska@suse.cz>
6470 * color-macros.h: New file.
6471 * diagnostic-color.c: Factor out color related to macros to
6473 * doc/gcov.texi: Document -k option.
6474 * gcov.c (INCLUDE_STRING): Include string.h.
6475 (print_usage): Add -k option.
6476 (process_args): Parse it.
6477 (pad_count_string): New function.
6478 (output_line_beginning): Likewise.
6479 (DEFAULT_LINE_START): New macro.
6480 (output_lines): Support color output.
6482 2017-10-31 Martin Liska <mliska@suse.cz>
6484 PR gcov-profile/82633
6485 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6486 their interaction with GCOV infrastructure.
6487 * configure.ac: Add -fkeep-{inline,static}-functions to
6489 * configure: Regenerate.
6491 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6494 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6496 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6499 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6500 into a register if it does not fit into an immediate offset field.
6502 2017-10-31 Olivier Hainque <hainque@adacore.com>
6504 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6506 2017-10-31 Julia Koval <julia.koval@intel.com>
6508 * config.gcc: Add gfniintrin.h.
6509 * config/i386/gfniintrin.h: New.
6510 * config/i386/i386-builtin-types.def
6511 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6512 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6513 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6514 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6515 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6516 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6517 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6518 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6519 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6520 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6521 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6522 * config/i386/immintrin.h: Include gfniintrin.h.
6523 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6525 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6527 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6529 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6531 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6532 (arm_ashldi3_1bit): Remove pattern.
6533 (ashrdi3): Remove shift by 1 expansion.
6534 (arm_ashrdi3_1bit): Remove pattern.
6535 (lshrdi3): Remove shift by 1 expansion.
6536 (arm_lshrdi3_1bit): Remove pattern.
6537 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6538 cost of ashldi3 by 1.
6539 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6540 (<shift>di3_neon): Likewise.
6542 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6544 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6545 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6546 and (*aarch64_simd_mov<VQ:mode>).
6547 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6548 pattern alternative.
6549 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6550 attributes to match pattern alternative.
6552 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6554 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6555 * config/rs6000/emmintrin.h: New file.
6556 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6558 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6560 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6561 * (movdi_vfp_cortexa8): Remove pattern.
6563 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6565 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6567 (Specific, alpha*-dec-osf5.1): Remove.
6568 (Specific, mips-sgi-irix5): Remove.
6569 (Specific, mips-sgi-irix6): Remove.
6571 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6574 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6575 arguments to clear_bit_region_be.
6577 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6579 * gimplify.c: Include memmodel.h.
6581 2017-10-30 Martin Jambor <mjambor@suse.cz>
6583 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6584 condition whether loop should be executed at all.
6586 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6588 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6589 gimple folding of vec_madd() intrinsics.
6590 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6591 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6592 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6594 2017-10-30 Richard Biener <rguenther@suse.de>
6596 PR tree-optimization/82762
6598 2017-10-23 Richard Biener <rguenther@suse.de>
6600 PR tree-optimization/82129
6602 2017-08-01 Richard Biener <rguenther@suse.de>
6604 PR tree-optimization/81181
6605 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6606 (compute_antic): ... end of iteration here.
6608 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6610 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6611 -std=iso9899:2017 and -std=gnu17.
6612 * doc/standards.texi (C Language): Document C17 support.
6613 * doc/cpp.texi (Overview): Mention -std=c17.
6614 (Standard Predefined Macros): Document C11 and C17 values of
6615 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6616 * doc/extend.texi (Inline): Do not list individual options for
6617 standards newer than C99.
6618 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6620 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6623 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6625 * asan.c (asan_finish_file): Align asan globals array by shadow
6628 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6631 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6632 (struct store_immediate_info): Add bitregion_start and bitregion_end
6634 (store_immediate_info::store_immediate_info): Add brs and bre
6635 arguments and initialize bitregion_{start,end} from those.
6636 (struct merged_store_group): Add bitregion_start, bitregion_end,
6637 align_base and mask fields. Drop unnecessary struct keyword from
6638 struct store_immediate_info. Add do_merge method.
6639 (clear_bit_region_be): Use memset instead of loop storing zeros.
6640 (merged_store_group::do_merge): New method.
6641 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6642 stores as long as the surrounding bitregions have no gaps.
6643 (merged_store_group::merge_overlapping): Use do_merge.
6644 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6645 is byte aligned, rather than requiring that start and width are
6646 byte aligned. Drop unnecessary struct keyword from
6647 struct store_immediate_info. Allocate and populate also mask array.
6648 Make start of the arrays relative to bitregion_start rather than
6649 start and size them according to bitregion_{end,start} difference.
6650 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6651 struct store_immediate_info.
6652 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6653 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6654 struct keyword from struct store_immediate_info.
6655 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6656 between stores as long as the surrounding bitregions have no gaps.
6658 (struct split_store): Add orig non-static data member.
6659 (split_store::split_store): Initialize orig to false.
6660 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6661 if there is exactly a single original stmt. Change stmts argument
6662 to pointer from reference, if NULL, don't push anything to it. Add
6663 first argument, use it to optimize skipping over orig stmts that
6664 are known to be before bitpos already. Simplify.
6665 (split_group): Return unsigned int count how many stores are or
6666 would be needed rather than a bool. Add allow_unaligned argument.
6667 Change split_stores argument from reference to pointer, if NULL,
6668 only do a dry run computing how many stores would be produced.
6669 Rewritten algorithm to use both alignment and misalign if
6670 !allow_unaligned and handle bitfield stores with gaps.
6671 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6672 from bitregion_start instead of start. Compute allow_unaligned
6673 here, if true, do 2 split_group dry runs to compute which one
6674 produces fewer stores and prefer aligned if equal. Punt if
6675 new count is bigger or equal than original before emitting any
6676 statements, rather than during that. Remove no longer needed
6677 new_ssa_names tracking. Replace num_stmts with
6678 split_stores.length (). Use 32-bit stack allocated entries
6679 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6680 if possible. Handle bitfields with gaps.
6681 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6682 Compute bitregion_{start,end} for the stores and construct
6683 store_immediate_info with that. Formatting fixes.
6685 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6688 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6689 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6691 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6693 * gimplify.c: Include tm_p.h.
6695 * common.opt (gcoff): Re-add as ignored option.
6696 (gcoff1, gcoff2, gcoff3): Likewise.
6698 * Makefile.in (OBJS): Delete sdbout.o.
6699 (GTFILES): Delete $(srcdir)/sdbout.c.
6700 * debug.h: Delete sdb_debug_hooks.
6701 * final.c: Delete sdbout.h include.
6702 (final_scan_insn): Delete SDB_DEBUG check.
6703 (rest_of_clean_state): Likewise.
6704 * output.h: Delete sdb_begin_function_line.
6707 * toplev.c: Delete sdbout.h include.
6708 (process_options): Delete SDB_DEBUG check.
6709 * tree-core.h (tree_type_common): Delete pointer field of
6711 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6712 TYPE_SYMTAB_POINTER.
6713 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6714 (TYPE_SYMTAB_IS_POINTER): Delete.
6715 (TYPE_SYMTAB_IS_DIE): Renumber.
6716 * xcoffout.c: Refer to former sdbout.c file.
6717 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6719 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6720 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6721 (Debugging Options): Delete -gcoff.
6722 (-gstabs): Delete SDB reference.
6724 (-gcoff@var{level}): Delete.
6725 * doc/passes.texi (Debugging information output): Delete SDB and
6726 sdbout.c references.
6727 * doc/tm.texi: Regenerate.
6728 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6729 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6731 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6732 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6733 (SDB_DEBUGGING_INFO): Delete.
6734 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6735 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6736 * target.def (output_source_filename): Delete COFF reference.
6738 * common.opt (gcoff): Delete.
6739 (gxcoff+): Update Negative chain.
6740 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6742 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6743 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6744 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6746 (expand_function_start): Change sdb reference to past tense.
6747 (expand_function_end): Change sdb reference to past tense.
6748 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6749 * opts.c (debug_type_names): Delete coff entry.
6750 (common_handle_option): Delete OPT_gcoff case.
6751 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6753 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6754 * config/cris/cris.h: Delete SDB reference in comment.
6755 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6756 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6757 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6758 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6760 (ix86_expand_prologue): Likewise.
6761 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6762 * config/ia64/ia64.h: Likewise.
6763 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6764 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6766 * config/mmix/mmix.h: Likewise.
6767 * config/nds32/nds32.c: Likewise.
6768 * config/stormy/storym16.h: Likewise.
6769 * config/visium/visium.h: Likewise.
6770 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6772 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6774 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6775 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6776 FRAME_POINTER_REGNUM point at high end of local var area.
6778 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6780 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6781 Move comment around. Do not reset best_edge for a copiable
6782 destination if the copy would cause a partition change.
6783 (better_edge_p): Remove redundant check.
6785 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6787 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6789 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6791 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6792 for math functions that have _Float<N> and _Float<N>X variants.
6793 (mathfn_built_in_2): Add support for math functions that have
6794 _Float<N> and _Float<N>X variants.
6795 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6796 (expand_builtin_mathfn_ternary): Add support for fma with
6797 _Float<N> and _Float<N>X variants.
6798 (expand_builtin): Likewise.
6799 (fold_builtin_3): Likewise.
6800 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6801 create math function _Float<N> and _Float<N>X variants as external
6803 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6804 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6805 the __builtin_ prefix and if not strict ansi, without the prefix.
6806 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6807 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6808 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6809 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6810 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6811 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6812 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6813 function signatures for fma _Float<N> and _Float<N>X variants.
6814 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6815 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6816 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6817 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6818 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6819 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6820 * gencfn-macros.c (print_case_cfn): Add support for math functions
6821 that have _Float<N> and _Float<N>X variants.
6822 (print_define_operator_list): Likewise.
6823 (fltfn_suffixes): Likewise.
6825 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6826 for math functions that have _Float<N> and _Float<N>X variants.
6827 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6828 and _Float<N>X variants.
6829 (COPYSIGN): Likewise.
6832 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6833 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6835 (integer_valued_read_call_p): Likewise.
6836 * fold-const-call.c (fold_const_call_ss): Likewise.
6837 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6838 _Float<N> and _Float<N>X variants.
6839 (fold_const_call_ssss): Add support for fma _Float<N> and
6840 _Float<N>X variants.
6841 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6842 support for copysign and fma _Float<N> and _Float<N>X variants.
6843 (backprop::process_builtin_call_use): Likewise.
6844 * tree-call-cdce.c (can_test_argument_range); Add support for
6845 sqrt _Float<N> and _Float<N>X variants.
6846 (edom_only_function): Likewise.
6847 (get_no_error_domain): Likewise.
6848 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6849 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6850 copysign _Float<N> and _Float<N>X variants.
6851 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6852 handled by machine independent code.
6853 (FMAF128): Likewise.
6854 * doc/cpp.texi (Common Predefined Macros): Document defining
6855 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6856 fma _Float<N> and _Float<N>X variants.
6858 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6861 * config/i386/i386-modes.def (CCFPU): Remove definition.
6862 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6863 (ix86_cc_modes_compatible): Ditto.
6864 (ix86_expand_carry_flag_compare): Ditto.
6865 (ix86_expand_int_movcc): Ditto.
6866 (ix86_expand_int_addcc): Ditto.
6867 (ix86_reverse_condition): Ditto.
6868 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6869 Return true/false for unordered/ordered fp comparisons.
6870 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6871 (ix86_prepare_fp_compare_args): Update for rename.
6872 (ix86_expand_fp_compare): Update for rename. Generate unordered
6873 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6874 (ix86_expand_sse_compare_and_jump): Ditto.
6875 * config/i386/predicates.md (fcmov_comparison_operator):
6876 Remove CCFPU mode handling.
6877 (ix86_comparison_operator): Ditto.
6878 (ix86_carry_flag_operator): Ditto.
6879 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6880 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6881 (*cmpu<mode>_cc_i387): Ditto.
6882 (FPCMP): Remove mode iterator.
6883 (unord): Remove mode attribute.
6884 (unord_subst): New define_subst transformation
6885 (unord): New define_subst attribute.
6887 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6888 (*cmpi<unord>xf_i387): Ditto.
6889 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6890 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6891 using unord_subst transformation.
6892 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6893 (round_saeonly): Also handle CCFP mode.
6894 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6895 Remove UNSPEC_SAHF unspec handling.
6897 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6899 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6901 2017-10-27 Jeff Law <law@redhat.com>
6903 * gimple-ssa-sprintf.c: Include domwalk.h.
6904 (class sprintf_dom_walker): New class, derived from dom_walker.
6905 (sprintf_dom_walker::before_dom_children): New function.
6906 (struct call_info): Moved into sprintf_dom_walker class
6907 (compute_formath_length, handle_gimple_call): Likewise.
6908 (sprintf_length::execute): Call the dominator walker rather
6909 than walking the statements.
6911 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6912 gimple statement locations.
6913 (check_array_bounds): Corresponding changes. Get the statement's
6914 location directly from wi->stmt.
6916 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6919 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6921 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6923 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6924 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6926 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6929 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6930 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6931 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6932 maybe_get_pool_constant.
6933 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6936 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6938 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6939 2.26 caveat. Update gas and gld versions.
6940 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6943 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6945 * cgraph.h (set_malloc_flag): Declare.
6946 * cgraph.c (set_malloc_flag_1): New function.
6947 (set_malloc_flag): Likewise.
6948 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6949 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6951 (read_ipa_call_summary): Add support for reading is_return_callee.
6952 (write_ipa_call_summary): Stream is_return_callee.
6953 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6954 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6955 ipa-prop.h, ipa-fnsummary.h.
6956 (pure_const_names): Change to static.
6957 (malloc_state_e): Define.
6958 (malloc_state_names): Define.
6959 (funct_state_d): Add field malloc_state.
6960 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6961 (check_retval_uses): New function.
6962 (malloc_candidate_p): Likewise.
6963 (analyze_function): Add support for malloc attribute.
6964 (pure_const_write_summary): Stream malloc_state.
6965 (pure_const_read_summary): Add support for reading malloc_state.
6966 (dump_malloc_lattice): New function.
6967 (propagate_malloc): New function.
6968 (warn_function_malloc): New function.
6969 (ipa_pure_const::execute): Call propagate_malloc and
6970 ipa_free_fn_summary.
6971 (pass_local_pure_const::execute): Add support for malloc attribute.
6972 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6973 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6975 2017-10-27 Martin Liska <mliska@suse.cz>
6977 PR gcov-profile/82457
6978 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6979 for fork-like functions to be properly instrumented.
6981 2017-10-27 Richard Biener <rguenther@suse.de>
6984 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6985 info when we redirected EH.
6987 2017-10-26 Michael Collison <michael.collison@arm.com>
6989 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6991 (<optab>_trunchf<GPI:mode>2: New pattern.
6992 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6993 * config/aarch64/iterators.md (wv): New mode attribute.
6994 (vf, VF): New mode attributes.
6995 (vgp, VGP): New mode attributes.
6996 (s): Update attribute with SImode and DImode prefixes.
6998 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7000 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7001 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7002 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7003 (nios2_option_overide): Initialize it. Don't allow R0-relative
7004 addressing with PIC.
7005 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7006 (nios2_symbolic_constant_p): Likewise.
7007 (nios2_legitimate_address_p): Likewise.
7008 (nios2_r0rel_section_name_p): New.
7009 (nios2_symbol_ref_in_r0rel_data_p): New.
7010 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7011 (r0rel_constant_p): New.
7012 (nios2_print_operand_address): Handle r0rel_constant_p.
7013 (nios2_cdx_narrow_form_p): Likewise.
7014 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7015 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7016 (Nios II Options): Document -mr0rel-sec.
7018 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7020 * config/nios2/nios2.c: Include xregex.h.
7021 (nios2_gprel_sec_regex): New.
7022 (nios2_option_overide): Initialize it. Don't allow GP-relative
7023 addressing with PIC.
7024 (nios2_small_section_name_p): Check for regex match.
7025 * config/nios2/nios2.opt (mgprel-sec=): New option.
7026 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7027 (Nios II Options): Document -mgprel-sec.
7029 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7031 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7033 2017-10-26 Tom de Vries <tom@codesourcery.com>
7035 PR tree-optimization/82707
7036 * gimple.c (gimple_copy): Fix unsharing of
7037 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7039 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7041 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7042 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7043 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7044 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7045 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7046 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7047 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7048 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7049 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7050 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7051 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7052 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7053 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7054 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7055 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7056 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7057 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7059 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7061 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7063 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7064 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7065 warning to rs6000.c. Remove the Undocumented flag, since it has
7067 (-mabi=ibmlongdouble): Likewise.
7068 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7069 already set, set the default format for long double.
7070 (rs6000_debug_reg_global): Print whether long double is IBM or
7072 (rs6000_option_override_internal): Rework setting long double
7073 format. Only warn if the user is changing the long double default
7074 and they did not use -Wno-psabi.
7075 * doc/invoke.texi (PowerPC options): Update the documentation for
7076 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7078 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7079 Alan Hayward <alan.hayward@arm.com>
7080 David Sherwood <david.sherwood@arm.com>
7082 * rtl.h (wider_subreg_mode): New function.
7083 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7084 rather than an unsigned int *.
7085 * ira-color.c (regno_max_ref_width): Replace with...
7086 (regno_max_ref_mode): ...this new variable.
7087 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7088 Use wider_subreg_mode.
7089 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7090 rather than an unsigned int *.
7091 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7092 (process_alt_operands): Likewise.
7093 (invariant_p): Likewise.
7094 * lra-spills.c (assign_mem_slot): Likewise.
7095 (add_pseudo_to_slot): Likewise.
7096 * lra.c (collect_non_operand_hard_regs): Likewise.
7097 (add_regs_to_insn_regno_info): Likewise.
7098 * reload1.c (regno_max_ref_width): Replace with...
7099 (regno_max_ref_mode): ...this new variable.
7100 (reload): Update accordingly. Update call to
7101 ira_sort_regnos_for_alter_reg.
7102 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7103 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7104 (scan_paradoxical_subregs): Likewise.
7106 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7108 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7109 (aarch64_frame): Add emit_frame_chain boolean.
7110 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7111 Move eh_return case to aarch64_layout_frame.
7112 (aarch64_layout_frame): Initialize emit_frame_chain.
7113 (aarch64_expand_prologue): Use emit_frame_chain.
7115 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7117 * config/aarch64/aarch64.c (aarch64_layout_frame):
7118 Ensure LR is always stored at the bottom of the callee-saves.
7119 Remove rarely used frame layout which saves callee-saves at top of
7120 frame, so the store of LR can be used as a valid probe in all cases.
7122 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7124 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7125 Improve unaligned TImode/TFmode base/offset split.
7127 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7128 Alan Hayward <alan.hayward@arm.com>
7129 David Sherwood <david.sherwood@arm.com>
7131 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7132 * combine.c (find_single_use_1): Likewise.
7133 (expand_field_assignment): Likewise.
7134 (move_deaths): Likewise.
7135 * lra-constraints.c (simplify_operand_subreg): Likewise.
7136 (curr_insn_transform): Likewise.
7137 * lra.c (collect_non_operand_hard_regs): Likewise.
7138 (add_regs_to_insn_regno_info): Likewise.
7139 * rtlanal.c (reg_referenced_p): Likewise.
7140 (covers_regno_no_parallel_p): Likewise.
7142 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7144 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7145 Don't print any bits outside the precision of the value.
7146 * wide-int.cc (test_printing): Add some new tests.
7148 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7150 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7151 supports -xbrace_comment option.
7152 * configure: Regenerate.
7153 * config.in: Regenerate.
7154 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7155 (ASM_CPU_SPEC): Use it.
7157 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7159 * target.def (static_rtx_alignment): New hook.
7160 * targhooks.h (default_static_rtx_alignment): Declare.
7161 * targhooks.c (default_static_rtx_alignment): New function.
7162 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7163 * doc/tm.texi: Regenerate.
7164 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7165 instead of targetm.constant_alignment. Remove call to
7167 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7168 (cris_preferred_mininum_alignment): New function, split out from...
7169 (cris_constant_alignment): ...here.
7170 (cris_static_rtx_alignment): New function.
7171 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7173 (ix86_constant_alignment): ...here.
7174 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7175 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7176 (mmix_static_rtx_alignment): New function.
7177 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7178 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7180 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7183 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7184 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7186 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7188 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7189 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7190 mark as a sign-extending load.
7191 (local_pic_load_u): Define.
7193 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7196 * fold-const.c (operand_equal_for_comparison_p): Also return true
7197 if ARG0 is a simple variant of ARG1 with narrower precision.
7198 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7200 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7202 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7204 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7205 scatter_static, scatter_per_elt.
7206 * x86-tune-costs.h: Add new cost entries.
7208 2017-10-25 Richard Biener <rguenther@suse.de>
7210 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7211 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7212 class pass_fre): Move to ...
7213 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7214 class pass_fre): ... here and adjust for statistics.
7216 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7219 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7220 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7221 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7222 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7225 2017-10-25 Richard Biener <rguenther@suse.de>
7227 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7228 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7229 eliminate_push_avail, eliminate_insert): Move inside...
7230 (class eliminate_dom_walker): ... this class in preparation
7232 (fini_eliminate): Remove by merging with ...
7233 (eliminate): ... this function. Adjust for class changes.
7234 (pass_pre::execute): Remove fini_eliminate call.
7235 (pass_fre::execute): Likewise.
7237 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7240 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7241 (VPERMI2, VPERMI2I): New mode iterators.
7242 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7243 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7245 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7246 mode iterator. Remove 3 old define_insn patterns.
7247 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7248 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7249 VPERMI2 mode iterator, remove the other two expanders.
7250 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7251 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7252 instructions, remove the other two patterns.
7253 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7254 mode iterator, remove the other two patterns.
7255 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7256 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7257 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7258 and adjust argument order accordingly.
7259 (ix86_expand_vec_perm): Adjust caller.
7260 (expand_vec_perm_1): Likewise.
7261 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7262 (expand_vec_perm_vpermt2_vpshub2): ... this.
7263 (ix86_expand_vec_perm_const_1): Adjust caller.
7264 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7267 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7268 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7269 (vec_shl_<mode>): Remove unused expander.
7270 (avx512bw_<shift_insn><mode>3): New define_insn.
7271 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7272 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7274 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7277 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7280 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7282 PR rtl-optimization/82396
7283 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7284 (autopref_multipass_init): Simplify initialization.
7285 (autopref_rank_data): Simplify sort order.
7286 * gcc/sched-int.h (autopref_multipass_data_): Remove
7287 multi_mem_insn_p, min_offset and max_offset.
7289 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7292 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7293 Check special value of flag_omit_frame_pointer.
7294 (aarch64_can_eliminate): Likewise.
7295 (aarch64_override_options_after_change_1): Simplify handling of
7296 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7298 2017-10-24 Richard Biener <rguenther@suse.de>
7300 PR tree-optimization/82697
7301 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7302 zero for conditional load and unconditional store.
7304 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7306 * doc/install.texi: Document bootstrap-cet.
7308 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7311 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7312 ENDBR instruction at function entrance if function is only
7315 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7318 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7319 patterns to better describe from which operation the CF is computed.
7320 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7321 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7322 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7323 is 0, use _0 suffixed expanders instead of emitting a comparison
7326 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7328 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7329 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7331 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7334 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7335 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7336 * loop-iv.c (iv_get_reaching_def): Likewise.
7337 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7338 variable is promoted and the partition contains undefined values.
7340 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7342 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7344 (nios2_address_cost): Define.
7345 (nios2_legitimize_address): Recognize (exp + constant) directly.
7346 (TARGET_ADDRESS_COST): Define.
7348 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7350 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7351 (nios2_symbolic_memory_operand_p): Declare.
7352 (nios2_split_large_constant): Declare.
7353 (nios2_split_symbolic_memory_operand): Declare.
7354 * config/nios2/nios2.c: Adjust includes.
7355 (nios2_symbolic_constant_allowed): New.
7356 (nios2_symbolic_constant_p): New.
7357 (nios2_plus_symbolic_constant_p): New.
7358 (nios2_valid_addr_expr_p): Recognize addresses involving
7360 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7361 (nios2_symbolic_memory_operand_p): New.
7362 (nios2_large_constant_p): New.
7363 (nios2_split_large_constant): New.
7364 (nios2_split_plus_large_constant): New.
7365 (nios2_split_symbolic_memory_operand): New.
7366 (nios2_legitimize_address): Code refactoring. Handle addresses
7367 involving symbolic constants.
7368 (nios2_emit_move_sequence): Likewise.
7369 (nios2_print_operand): Improve error output.
7370 (nios2_print_operand_address): Handle LO_SUM.
7371 (nios2_cdx_narrow_form_p): Likewise.
7372 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7373 operands involving symbolic constants.
7374 (movhi_internal, movsi_internal): Likewise.
7375 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7376 (extendhisi2, extendqi<mode>2): Likewise.
7378 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7380 * tree-pass.h (PROP_rtl_split_insns): Define.
7381 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7383 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7385 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7387 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7390 * target.def (const_not_ok_for_debug_p): Default to
7391 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7392 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7393 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7394 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7395 which targetm.const_not_ok_for_debug_p returned true.
7396 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7398 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7400 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7401 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7402 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7404 (ix86_const_not_ok_for_debug_p): New function.
7405 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7406 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7408 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7411 * system.h: Conditionally include "unique-ptr.h" if
7412 INCLUDE_UNIQUE_PTR is defined.
7413 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7414 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7416 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7418 * config/rl78/rl78.md: New define_expand "subdi3".
7420 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7423 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7424 DF_REF_INSN if DF_REF_INSN_INFO is false.
7426 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7428 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7429 xmm_move instead of sse_move.
7430 (sse_store_index): New function.
7431 (ix86_register_move_cost): Be more sensible about mismatch stall;
7432 model AVX moves correctly; make difference between sse->integer and
7434 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7435 moves; make difference between SSE and AVX.
7436 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7437 and zmm_move. Increase size of sse load and store tables;
7438 add unaligned load and store tables; add ssemmx_to_integer.
7439 * x86-tune-costs.h: Update all entries according to real
7440 move latencies from Agner Fog's manual and chip documentation.
7442 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7445 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7446 * config/i386/constraints.md (Wf): New constraint.
7447 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7448 (cmp<dwi>_doubleword): Removed.
7449 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7450 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7451 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7452 expand with cmp<dwi>_doubleword. For LTU and GEU use
7453 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7455 * common.opt (gcolumn-info): Enable by default.
7456 * doc/invoke.texi (gcolumn-info): Document new default.
7458 2017-10-23 Richard Biener <rguenther@suse.de>
7460 PR tree-optimization/82672
7461 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7462 Fold the stmt if we propagated into it.
7464 2017-10-23 Richard Biener <rguenther@suse.de>
7466 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7467 (bitmap_remove_expr_from_set): ... this. All callers call this
7468 for non-constant values.
7469 (bitmap_set_subtract): Rename to...
7470 (bitmap_set_subtract_expressions): ... this. Adjust and
7472 (bitmap_set_contains_value): Remove superfluous check.
7473 (bitmap_set_replace_value): Inline into single caller ...
7474 (bitmap_value_replace_in_set): ... here and simplify.
7475 (dependent_clean): Merge into ...
7476 (clean): ... this using an overload. Adjust.
7477 (prune_clobbered_mems): Adjust.
7478 (compute_antic_aux): Likewise.
7479 (compute_partial_antic_aux): Likewise.
7481 2017-10-23 Richard Biener <rguenther@suse.de>
7483 PR tree-optimization/82129
7485 2017-08-01 Richard Biener <rguenther@suse.de>
7487 PR tree-optimization/81181
7488 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7489 (compute_antic): ... end of iteration here.
7491 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7493 * target.def (starting_frame_offset): New hook.
7494 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7495 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7496 * doc/tm.texi.in: Regenerate.
7497 * hooks.h (hook_hwi_void_0): Declare.
7498 * hooks.c (hook_hwi_void_0): New function.
7499 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7500 STARTING_FRAME_OFFSET.
7501 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7502 * reload1.c (reload): Likewise.
7503 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7504 instead of STARTING_FRAME_OFFSET.
7505 * function.c (try_fit_stack_local): Likewise.
7506 (assign_stack_local_1): Likewise
7507 (instantiate_virtual_regs): Likewise.
7508 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7509 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7510 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7511 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7512 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7513 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7514 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7515 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7516 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7517 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7518 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7519 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7520 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7521 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7522 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7523 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7524 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7525 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7526 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7527 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7528 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7529 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7530 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7531 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7532 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7533 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7534 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7535 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7536 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7537 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7538 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7539 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7540 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7541 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7542 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7543 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7544 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7545 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7546 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7547 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7548 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7549 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7550 return a HOST_WIDE_INT.
7551 (avr_builtin_setjmp_frame_value): Use it instead of
7552 STARTING_FRAME_OFFSET.
7553 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7554 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7555 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7557 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7558 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7559 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7560 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7561 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7562 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7563 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7564 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7565 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7566 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7567 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7568 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7570 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7571 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7572 * config/mips/mips.c (mips_compute_frame_info): Refer to
7573 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7574 (mips_starting_frame_offset): New function.
7575 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7576 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7577 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7578 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7579 and return a HOST_WIDE_INT.
7580 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7581 (mmix_initial_elimination_offset): Refer to
7582 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7583 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7584 * config/pa/pa.c (pa_starting_frame_offset): New function.
7585 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7586 (pa_expand_prologue): Likewise.
7587 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7588 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7589 !FRAME_GROWS_DOWNWARD handling to...
7590 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7591 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7592 !FRAME_GROWS_DOWNWARD handling to...
7593 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7594 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7595 !FRAME_GROWS_DOWNWARD handling to...
7596 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7597 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7599 (rs6000_starting_frame_offset): New function.
7600 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7601 !FRAME_GROWS_DOWNWARD handling to...
7602 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7603 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7604 !FRAME_GROWS_DOWNWARD handling to...
7605 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7606 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7607 !FRAME_GROWS_DOWNWARD handling to...
7608 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7609 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7610 (rs6000_starting_frame_offset): New function.
7611 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7612 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7613 * config/vax/vax.c (vax_starting_frame_offset): New function.
7614 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7615 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7616 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7617 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7618 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7619 * system.h (STARTING_FRAME_OFFSET): Poison.
7621 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7623 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7624 SCALAR_TYPE_MODE instead of TYPE_MODE.
7626 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7627 Alan Hayward <alan.hayward@arm.com>
7628 David Sherwood <david.sherwood@arm.com>
7630 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7632 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7633 Alan Hayward <alan.hayward@arm.com>
7634 David Sherwood <david.sherwood@arm.com>
7636 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7638 2017-10-23 Richard Biener <rguenther@suse.de>
7640 PR tree-optimization/82129
7641 * tree-ssa-pre.c (bitmap_set_and): Remove.
7642 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7643 canonicalizing expressions in the set to those with lowest
7644 ID rather than taking that from the first edge.
7646 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7648 * combine.c (rtx_equal_for_field_assignment_p): Use
7649 byte_lowpart_offset.
7651 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7652 Alan Hayward <alan.hayward@arm.com>
7653 David Sherwood <david.sherwood@arm.com>
7655 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7656 to a SUBREG_PROMOTED_VAR.
7658 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7659 Alan Hayward <alan.hayward@arm.com>
7660 David Sherwood <david.sherwood@arm.com>
7662 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7663 (expand_debug_source_expr): Likewise.
7664 * combine.c (combine_simplify_rtx): Likewise.
7665 * cse.c (fold_rtx): Likewise.
7666 * optabs.c (expand_float): Likewise.
7667 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7668 (simplify_binary_operation_1): Likewise.
7670 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7671 Alan Hayward <alan.hayward@arm.com>
7672 David Sherwood <david.sherwood@arm.com>
7674 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7675 (record_promoted_value): Likewise.
7676 * expr.c (expand_expr_real_2): Likewise.
7677 * ree.c (update_reg_equal_equiv_notes): Likewise.
7678 (combine_set_extension): Likewise.
7679 * rtlanal.c (low_bitmask_len): Likewise.
7680 * simplify-rtx.c (neg_const_int): Likewise.
7681 (simplify_binary_operation_1): Likewise.
7683 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7684 Alan Hayward <alan.hayward@arm.com>
7685 David Sherwood <david.sherwood@arm.com>
7687 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7688 * regcprop.c (maybe_mode_change): Likewise.
7689 * reload1.c (alter_reg): Likewise.
7691 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7693 * inchash.h (inchash::hash::add_wide_int): New function.
7694 * lto-streamer-out.c (hash_tree): Use it.
7696 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7698 * inchash.h (inchash::hash::add_wide_int): Rename to...
7699 (inchash::hash::add_hwi): ...this.
7700 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7701 (polymorphic_call_target_hasher::hash): Likewise.
7702 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7703 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7704 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7705 * lto-streamer-out.c (hash_tree): Likewise.
7706 * optc-save-gen.awk: Likewise.
7707 * tree.c (add_expr): Likewise.
7709 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7712 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7713 for ordered inequality comparisons even with TARGET_IEEE_FP.
7715 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7718 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7719 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7720 Expand with cmp<dwi>_doubleword.
7722 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7724 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7725 List CET intrinsics.
7726 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7727 specific to -fcf-protection option.
7729 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7731 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7732 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7733 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7734 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7735 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7736 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7737 (extra_objs): Add cet.o for Linux/x86 targets.
7738 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7739 * config/i386/cet.c: New file.
7740 * config/i386/cetintrin.h: Likewise.
7741 * config/i386/t-cet: Likewise.
7742 * config/i386/cpuid.h (bit_SHSTK): New.
7743 (bit_IBT): Likewise.
7744 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7745 pass IBT and SHSTK bits.
7746 * config/i386/i386-builtin-types.def
7747 (VOID_FTYPE_UNSIGNED_PVOID): New.
7748 (VOID_FTYPE_UINT64_PVOID): Likewise.
7749 * config/i386/i386-builtin.def: Add CET intrinsics.
7750 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7751 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7752 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7753 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7755 * config/i386/i386.c (rest_of_insert_endbranch): New.
7756 (pass_data_insert_endbranch): Likewise.
7757 (pass_insert_endbranch): Likewise.
7758 (make_pass_insert_endbranch): Likewise.
7759 (ix86_notrack_prefixed_insn_p): Likewise.
7760 (ix86_target_string): Add -mibt, -mshstk flags.
7761 (ix86_option_override_internal): Add flag_cf_protection
7763 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7764 (ix86_print_operand): Add 'notrack' prefix output.
7765 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7766 (ix86_expand_builtin): Expand CET intrinsics.
7767 (x86_output_mi_thunk): Add 'endbranch' instruction.
7768 * config/i386/i386.h (TARGET_IBT): New.
7769 (TARGET_IBT_P): Likewise.
7770 (TARGET_SHSTK): Likewise.
7771 (TARGET_SHSTK_P): Likewise.
7772 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7773 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7774 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7775 (builtin_setjmp_setup): New pattern.
7776 (builtin_longjmp): Likewise.
7777 (rdssp<mode>): Likewise.
7778 (incssp<mode>): Likewise.
7779 (saveprevssp): Likewise.
7780 (rstorssp): Likewise.
7781 (wrss<mode>): Likewise.
7782 (wruss<mode>): Likewise.
7783 (setssbsy): Likewise.
7784 (clrssbsy): Likewise.
7785 (nop_endbr): Likewise.
7786 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7788 * config/i386/immintrin.h: Include <cetintrin.h>.
7789 * config/i386/linux-common.h
7790 (file_end_indicate_exec_stack_and_cet): New prototype.
7791 (TARGET_ASM_FILE_END): New.
7793 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7795 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7796 latencies instead of having separate table; make difference between
7797 integer and float costs.
7798 * i386.h (processor_costs): Remove scalar_stmt_cost,
7799 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7800 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7802 * x86-tune-costs.h: Remove entries which has been removed in
7803 procesor_costs from all tables; make cond_taken_branch_cost
7804 and cond_not_taken_branch_cost COST_N_INSNS based.
7806 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7808 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7810 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7812 * config/i386/i386.md (isa): Remove fma_avx512f.
7813 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7814 <avx512>_fmadd_<mode>_mask3<round_name>,
7815 <avx512>_fmsub_<mode>_mask<round_name>,
7816 <avx512>_fmsub_<mode>_mask3<round_name>,
7817 <avx512>_fnmadd_<mode>_mask<round_name>,
7818 <avx512>_fnmadd_<mode>_mask3<round_name>,
7819 <avx512>_fnmsub_<mode>_mask<round_name>,
7820 <avx512>_fnmsub_<mode>_mask3<round_name>,
7821 <avx512>_fmaddsub_<mode>_mask<round_name>,
7822 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7823 <avx512>_fmsubadd_<mode>_mask<round_name>,
7824 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7825 (*vec_widen_umult_even_v16si<mask_name>,
7826 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7827 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7829 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7831 * extend.texi: Add 'nocf_check' documentation.
7832 * gimple.texi: Add second parameter to
7833 gimple_build_call_from_tree.
7834 * invoke.texi: Add -fcf-protection documentation.
7835 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7837 2017-10-20 Richard Biener <rguenther@suse.de>
7839 PR tree-optimization/82473
7840 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7841 the largest input type.
7843 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7845 * c-attribs.c (handle_nocf_check_attribute): New function.
7846 (c_common_attribute_table): Add 'nocf_check' handling.
7847 * gimple-parser.c: Add second argument NULL to
7848 gimple_build_call_from_tree.
7849 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7850 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7852 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7853 * common.opt: Add fcf-protection flag.
7854 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7855 * flag-types.h: Add enum cf_protection_level.
7856 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7857 Add 'nocf_check' attribute propagation to gimple call.
7858 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7859 (gimple_build_call_from_tree): Update prototype.
7860 (gimple_call_nocf_check_p): New function.
7861 (gimple_call_set_nocf_check): Likewise.
7862 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7863 * ipa-icf.c: Add nocf_check attribute in statement hash.
7864 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7865 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7866 * toplev.c (process_options): Add flag_cf_protection handling.
7868 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7870 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7872 2017-10-20 Richard Biener <rguenther@suse.de>
7874 PR tree-optimization/82603
7875 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7876 remove false predicated stores.
7878 2017-10-20 Richard Biener <rguenther@suse.de>
7880 * graphite-isl-ast-to-gimple.c
7881 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7882 Remove return value and simplify, dump copied stmt after lhs
7884 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7885 Reduce dump verbosity.
7886 (gsi_insert_earliest): Likewise.
7887 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7888 * graphite.c (print_global_statistics): Adjust dumping.
7889 (print_graphite_scop_statistics): Likewise.
7890 (print_graphite_statistics): Do not dump loops here.
7891 (graphite_transform_loops): But here.
7893 2017-10-20 Nicolas Roche <roche@adacore.com>
7895 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7896 * configure: Regenerate.
7898 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7901 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7902 functions when optimizing replace GIMPLE_RETURN stmts with
7903 calls to __builtin_unreachable ().
7906 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7907 for -fsanitize=thread link of executables.
7908 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7909 link of executables.
7912 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7914 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7915 define_insns for logical vector shifts to use VI248_AVX512BW
7916 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7917 condition, useless isa and prefix attributes. Change the first
7918 2 of these define_insns to ...
7919 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7920 define_insn for avx512vl.
7921 (<shift_insn><mode>3): ... and this, new define_insn without
7922 masking for non-avx512vl.
7925 * config/i386/sse.md (*andnot<mode>3,
7926 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7927 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7928 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7929 not applied use empty suffix even for TARGET_AVX512VL.
7930 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7931 is applied, supply evex,evex or evex,evex,evex instead of just
7934 2017-10-20 Julia Koval <julia.koval@intel.com>
7936 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7937 (OPTION_MASK_ISA_GFNI_UNSET): New.
7938 (ix86_handle_option): Handle OPT_mgfni.
7939 * config/i386/cpuid.h (bit_GFNI): New.
7940 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7941 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7942 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7943 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7944 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7945 * config/i386/i386.opt: Add mgfni.
7947 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7949 * config/msp430/msp430.c (msp430_option_override): Disable
7950 -fdelete-null-pointer-checks.
7951 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7953 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7955 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7956 of x87 and SSE instructions.
7958 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7960 * asan.c (create_cond_insert_point): Do not update edge count.
7961 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7962 (afdo_propagate_circuit): Likewise.
7963 (afdo_calculate_branch_prob): Likewise.
7964 (afdo_annotate_cfg): Likewise.
7965 * basic-block.h (struct edge_def): Remove count.
7966 (edge_def::count): New accessor.
7967 * bb-reorder.c (rotate_loop): Update.
7968 (find_traces_1_round): Update.
7969 (connect_traces): Update.
7970 (sanitize_hot_paths): Update.
7971 * cfg.c (unchecked_make_edge): Update.
7972 (make_single_succ_edge): Update.
7973 (check_bb_profile): Update.
7974 (dump_edge_info): Update.
7975 (update_bb_profile_for_threading): Update.
7976 (scale_bbs_frequencies_int): Update.
7977 (scale_bbs_frequencies_gcov_type): Update.
7978 (scale_bbs_frequencies_profile_count): Update.
7979 (scale_bbs_frequencies): Update.
7980 * cfganal.c (connect_infinite_loops_to_exit): Update.
7981 * cfgbuild.c (compute_outgoing_frequencies): Update.
7982 (find_many_sub_basic_blocks): Update.
7983 * cfgcleanup.c (try_forward_edges): Update.
7984 (try_crossjump_to_edge): Update
7985 * cfgexpand.c (expand_gimple_cond): Update
7986 (expand_gimple_tailcall): Update
7987 (construct_exit_block): Update
7988 * cfghooks.c (verify_flow_info): Update
7989 (redirect_edge_succ_nodup): Update
7990 (split_edge): Update
7991 (make_forwarder_block): Update
7992 (duplicate_block): Update
7993 (account_profile_record): Update
7994 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7995 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7996 * cfgloopmanip.c (scale_loop_profile): Update.
7998 (lv_adjust_loop_entry_edge): Update.
7999 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8000 (force_nonfallthru_and_redirect): Update.
8001 (purge_dead_edges): Update.
8002 (rtl_flow_call_edges_add): Update.
8003 * cgraphunit.c (init_lowered_empty_function): Update.
8004 (cgraph_node::expand_thunk): Update.
8005 * gimple-pretty-print.c (dump_probability): Update.
8006 (dump_edge_probability): Update.
8007 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8008 * haifa-sched.c (sched_create_recovery_edges): Update.
8009 * hsa-gen.c (convert_switch_statements): Update.
8010 * ifcvt.c (dead_or_predicable): Update.
8011 * ipa-inline-transform.c (inline_transform): Update.
8012 * ipa-split.c (split_function): Update.
8013 * ipa-utils.c (ipa_merge_profiles): Update.
8014 * loop-doloop.c (add_test): Update.
8015 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8016 * lto-streamer-in.c (input_cfg): Update.
8017 (input_function): Update.
8018 * lto-streamer-out.c (output_cfg): Update.
8019 * modulo-sched.c (sms_schedule): Update.
8020 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8021 * predict.c (maybe_hot_edge_p): Update.
8022 (unlikely_executed_edge_p): Update.
8023 (probably_never_executed_edge_p): Update.
8024 (dump_prediction): Update.
8025 (drop_profile): Update.
8026 (propagate_unlikely_bbs_forward): Update.
8027 (determine_unlikely_bbs): Update.
8028 (force_edge_cold): Update.
8029 * profile.c (compute_branch_probabilities): Update.
8030 * reg-stack.c (better_edge): Update.
8031 * shrink-wrap.c (handle_simple_exit): Update.
8032 * tracer.c (better_p): Update.
8033 * trans-mem.c (expand_transaction): Update.
8034 (split_bb_make_tm_edge): Update.
8035 * tree-call-cdce.c: Update.
8036 * tree-cfg.c (gimple_find_sub_bbs): Update.
8037 (gimple_split_edge): Update.
8038 (gimple_duplicate_sese_region): Update.
8039 (gimple_duplicate_sese_tail): Update.
8040 (gimple_flow_call_edges_add): Update.
8041 (insert_cond_bb): Update.
8042 (execute_fixup_cfg): Update.
8043 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8044 * tree-complex.c (expand_complex_div_wide): Update.
8045 * tree-eh.c (lower_resx): Update.
8046 (unsplit_eh): Update.
8047 (cleanup_empty_eh_move_lp): Update.
8048 * tree-inline.c (copy_edges_for_bb): Update.
8049 (freqs_to_counts): Update.
8050 (copy_cfg_body): Update.
8051 * tree-ssa-dce.c (remove_dead_stmt): Update.
8052 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8053 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8054 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8055 (unloop_loops): Update.
8056 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8057 * tree-ssa-loop-split.c (connect_loops): Update.
8058 (split_loop): Update.
8059 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8060 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8061 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8062 * tree-ssa-reassoc.c (branch_fixup): Update.
8063 * tree-ssa-tail-merge.c (replace_block_by): Update.
8064 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8065 (compute_path_counts): Update.
8066 (update_profile): Update.
8067 (recompute_probabilities): Update.
8068 (update_joiner_offpath_counts): Update.
8069 (estimated_freqs_path): Update.
8070 (freqs_to_counts_path): Update.
8071 (clear_counts_path): Update.
8072 (ssa_fix_duplicate_block_edges): Update.
8073 (duplicate_thread_path): Update.
8074 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8075 (case_bit_test_cmp): Update.
8076 (collect_switch_conv_info): Update.
8077 (gen_inbound_check): Update.
8078 (do_jump_if_equal): Update.
8079 (emit_cmp_and_jump_insns): Update.
8080 * tree-tailcall.c (decrease_profile): Update.
8081 (eliminate_tail_call): Update.
8082 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8083 (vect_do_peeling): Update.
8084 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8085 * ubsan.c (ubsan_expand_null_ifn): Update.
8086 (ubsan_expand_ptr_ifn): Update.
8087 * value-prof.c (gimple_divmod_fixed_value): Update.
8088 (gimple_mod_pow2): Update.
8089 (gimple_mod_subtract): Update.
8090 (gimple_ic): Update.
8091 (gimple_stringop_fixed_value): Update.
8093 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8096 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8098 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8100 PR rtl-optimization/82395
8101 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8102 based on non_spilled_static_chain_regno_p.
8104 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8106 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8107 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8108 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8110 2017-10-19 Martin Sebor <msebor@redhat.com>
8112 PR tree-optimization/82596
8113 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8115 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8117 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8118 (get_mem_refs_of_builtin_call): Likewise.
8119 * builtins.c (expand_builtin_apply): Adjust call to
8120 allocate_dynamic_stack_space.
8121 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8122 the third argument to allocate_dynamic_stack_space, otherwise -1.
8123 (expand_builtin): Deal with all alloca variants.
8124 (is_inexpensive_builtin): Likewise.
8125 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8126 * calls.c (special_function_p): Deal with all alloca variants.
8127 (initialize_argument_information): Adjust call to
8128 allocate_dynamic_stack_space.
8129 (expand_call): Likewise.
8130 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8131 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8132 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8133 use it for the stack usage computation.
8134 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8135 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8136 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8137 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8138 (in_loop_p): Remove first argument and useless check.
8139 (pass_walloca::execute): Remove useless test and adjust call to above.
8140 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8141 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8142 (gimplify_call_expr): Deal with all alloca variants.
8143 * hsa-gen.c (gen_hsa_alloca): Likewise.
8144 (gen_hsa_insns_for_call): Likewise.
8145 * ipa-pure-const.c (special_builtin_state): Likewise.
8146 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8147 * tree-object-size.c (alloc_object_size): Likewise.
8148 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8149 (call_may_clobber_ref_p_1): Likewise.
8150 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8151 (ccp_fold_stmt): Likewise.
8152 (optimize_stack_restore): Likewise.
8153 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8154 (mark_all_reaching_defs_necessary_1): Likewise.
8155 (propagate_necessity): Likewise.
8156 (eliminate_unnecessary_stmts): Likewise.
8157 * tree.c (build_common_builtin_nodes): Build
8158 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8159 (build_alloca_call_expr): New function.
8160 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8161 (CASE_BUILT_IN_ALLOCA): Likewise.
8162 (build_alloca_call_expr): Declare.
8163 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8165 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8168 * dwarf2out.c (new_die_raw): New static inline function.
8169 (new_die): Use it to create the DIE.
8170 (add_AT_external_die_ref): Likewise.
8171 (clone_die): Likewise.
8172 (clone_as_declaration): Likewise.
8173 (dwarf2out_vms_debug_main_pointer): Likewise.
8174 (base_type_die): Likewise. Remove early return for corner cases.
8175 Do not call add_pubtype on the DIE here.
8176 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8177 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8178 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8179 native order exists for base types, attach the DIE manually and call
8180 add_pubtype on it. Do not equate a reverse order DIE to the type.
8182 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8184 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8185 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8187 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8189 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8191 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8192 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8193 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8194 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8195 * opts.c (sanitizer_opts): Add builtin.
8196 * ubsan.c (instrument_builtin): New function.
8197 (pass_ubsan::execute): Call it.
8198 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8199 * doc/invoke.texi: Document -fsanitize=builtin.
8201 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8202 builtins, store max (log2 (align), 0) into uchar field instead of
8203 align into uptr field.
8204 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8205 store uchar 0 field instead of uptr 0 field.
8206 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8207 instead of passing one address of struct with 2 locations pass
8208 two addresses of structs with 1 location each.
8209 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8210 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8211 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8212 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8213 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8214 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8215 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8216 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8218 2017-10-19 Martin Liska <mliska@suse.cz>
8221 * file-find.c (remove_prefix): Remove.
8222 * file-find.h (remove_prefix): Likewise.
8223 * gcc-ar.c: Remove smartness of lookup.
8225 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8227 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8228 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8229 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8231 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8234 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8235 (setcc + and to xor + setcc): New peephole2.
8237 2017-10-19 Tom de Vries <tom@codesourcery.com>
8239 * doc/sourcebuild.texi (Test Directives, Variants of
8240 dg-require-support): Add dg-require-stack-size.
8242 2017-10-19 Martin Liska <mliska@suse.cz>
8245 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8246 that have a large alignment.
8247 (gimplify_target_expr): Likewise.
8249 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8251 PR rtl-optimization/82602
8252 * ira.c (rtx_moveable_p): Return false for volatile asm.
8254 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8257 * config/i386/i386-modes.def (CCGZ): New CC mode.
8258 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8259 * config/i386/predicates.md (ix86_comparison_operator):
8261 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8262 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8263 with double-word subtraction.
8264 (put_condition_code): Handle CCGZmode.
8266 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8268 * wide-int.cc (debug (const wide_int &)): New.
8269 (debug (const wide_int *)): New.
8270 (debug (const widest_int &)): New.
8271 (debug (const widest_int *)): New.
8273 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8276 * lra-constraints.c (curr_insn_transform): Use non-input operand
8277 instead of output one for matched reload.
8279 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8281 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8282 (tree-ssa-loop-ivopts.h): New header file.
8283 (struct builtin_info): New fields.
8284 (classify_builtin_1): Compute and record base and offset parts for
8285 memset builtin partition by calling strip_offset.
8286 (offset_cmp, fuse_memset_builtins): New functions.
8287 (finalize_partitions): Fuse adjacent memset partitions by calling
8289 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8290 Expose the interface.
8291 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8293 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8295 PR tree-optimization/82574
8296 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8297 that data reference must be executed exactly once per iteration
8298 against the outermost loop in nest.
8299 (classify_partition): Update call to above function.
8301 2017-10-18 Richard Biener <rguenther@suse.de>
8303 PR tree-optimization/82591
8304 * graphite.c (graphite_transform_loops): Move code gen message
8306 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8307 Here. Handle scop_to_isl_ast failing.
8308 (scop_to_isl_ast): Limit the number of ISL operations.
8310 2017-10-18 Richard Biener <rguenther@suse.de>
8312 * graphite-isl-ast-to-gimple.c
8313 (translate_isl_ast_to_gimple::set_rename): Simplify.
8314 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8315 (graphite_copy_stmts_from_block): ... here.
8316 (copy_bb_and_scalar_dependences): Simplify.
8317 (add_parameters_to_ivs_params): Canonicalize.
8318 (generate_entry_out_of_ssa_copies): Simplify.
8319 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8320 by passing in ISL dimension.
8321 (parameter_index_in_region_1): Rename to ...
8322 (parameter_index_in_region): ... this.
8323 (extract_affine): Adjust assert, pass down parameter index.
8324 (add_param_constraints): Use range-info when available.
8325 (build_scop_context): Adjust.
8326 * sese.c (new_sese_info): Adjust.
8327 (free_sese_info): Likewise.
8328 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8329 Remove unused typedefs.
8330 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8332 2017-10-18 Martin Liska <mliska@suse.cz>
8334 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8336 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8338 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8339 (s390_sched_init): Do not reset s390_sched_state if we entered the
8340 current basic block via a fallthru edge and all others are unlikely.
8342 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8344 * config/s390/s390.c (NUM_SIDES): New variable.
8345 (LONGRUNNING_THRESHOLD): New variable.
8346 (LATENCY_FACTOR): New variable.
8347 (s390_sched_score): Decrease score for long-running instructions on
8349 (s390_sched_variable_issue): Perform bookkeeping for long-running
8352 2017-10-18 Richard Biener <rguenther@suse.de>
8354 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8355 Simplify with removal of the parameter rename map.
8356 (set_rename): Likewise.
8357 (should_copy_to_new_region): Likewise.
8358 (graphite_copy_stmts_from_block): Likewise.
8359 (copy_bb_and_scalar_dependences): Remove initialization of
8360 unused copied_bb_map.
8362 (copy_internal_parameters): Likewise.
8363 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8364 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8365 Use INTEGRAL_TYPE_P.
8366 (parameter_index_in_region_1): Rename to ...
8367 (assign_parameter_index_in_region): ... this. Assert we have
8368 a parameter we handle.
8369 (scan_tree_for_params): Adjust.
8370 * sese.h (parameter_rename_map_t): Remove.
8371 (struct sese_info_t): Remove unused parameter_rename_map and
8372 copied_bb_map members.
8373 * sese.c (new_sese_info): Adjust.
8374 (free_sese_info): Likewise.
8376 2017-10-18 Martin Liska <mliska@suse.cz>
8379 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8380 on an abnormal edge.
8382 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8384 * doc/invoke.texi (ffunction-sections and fdata-sections):
8387 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8389 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8390 the use statement can throw internally.
8392 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8394 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8395 any RTX present on the RHS of a SET.
8396 * compare-elim.c (try_eliminate_compare): Restore comment.
8398 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8400 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8401 may be also called on tcc_type nodes.
8402 * langhooks.c (lhd_tree_size): Likewise.
8404 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8406 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8407 format_warning_at_substring.
8408 (maybe_warn): Convert source_range * param to a location_t. Pass
8409 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8410 (format_directive): Remove code to extract source_ranges and
8411 source_range * in favor of just a location_t.
8412 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8414 * substring-locations.c (format_warning_va): Convert
8415 source_range * param to a location_t.
8416 (format_warning_at_substring): Likewise.
8417 * substring-locations.h (format_warning_va): Likewise.
8418 (format_warning_at_substring): Likewise.
8420 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8422 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8424 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8425 and scatter/gather ops.
8427 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8428 vec_gather_load and vec_scatter_store.
8429 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8430 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8431 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8432 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8433 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8434 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8436 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8438 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8439 Detect FTST instruction and handle its register pops. Only pop
8440 second operand if can_pop_second_op is true.
8441 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8442 set can_pop_second_op to false in the compare_for_stack_reg call.
8444 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8445 output_fp_compare for stack register operands.
8446 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8447 instructions here. Do not emit stack register pops here. Assert
8448 that FCOMPP pops next to top stack register. Rewrite function.
8450 2017-10-17 Nathan Sidwell <nathan@acm.org>
8453 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8454 use DECL_ASSEMBLER_NAME_RAW.
8457 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8460 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8461 Wilco Dijkstra <wilco.dijkstra@arm.com>
8463 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8464 converstion to Pmode from the buf_addr.
8466 2017-10-17 Richard Biener <rguenther@suse.de>
8468 * graphite-dependences.c (scop_get_reads_and_writes): Change
8469 output parameters to references.
8471 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8473 PR 71026/tree-optimization
8474 * fold-const.c (distribute_real_division): Removed.
8475 (fold_binary_loc): Remove calls to distribute_real_divison.
8477 2017-10-17 Richard Biener <rguenther@suse.de>
8479 * graphite-scop-detection.c
8480 (scop_detection::stmt_has_simple_data_refs_p): Always use
8481 the full nest as region.
8482 (try_generate_gimple_bb): Likewise.
8483 * sese.c (scalar_evolution_in_region): Simplify now that
8484 SCEV can handle instantiation in regions.
8485 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8486 in the non-loop part of a function if requested.
8488 2017-10-17 Richard Biener <rguenther@suse.de>
8490 PR tree-optimization/82563
8491 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8493 (graphite_regenerate_ast_isl): Call it.
8494 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8496 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8498 PR tree-optimization/82549
8499 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8500 Formatting fixes. Instead of calling make_bit_field_ref with negative
8503 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8505 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8506 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8507 _mm_maskz_reduce_ss): New.
8508 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8509 __builtin_ia32_reducess_mask): Ditto..
8510 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8511 * config/i386/sse.md (reduces<mode>): Renamed to ...
8512 (reduces<mode><mask_scalar_name>): ... this.
8513 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8515 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8516 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8518 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8520 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8521 * selftest-run-tests.c (selftest::run_tests): Call
8522 selftest::unique_ptr_tests_cc_tests.
8523 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8524 * unique-ptr-tests.cc: New file.
8526 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8529 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8531 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8532 (make_hard_regno_born, make_hard_regno_dead): Update
8533 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8535 2017-10-16 Jeff Law <law@redhat.com>
8537 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8539 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8541 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8543 2017-10-16 Olivier Hainque <hainque@adacore.com>
8545 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8546 with_cpu if we were configured for an e500v2 target cpu name.
8548 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8550 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8551 * doc/invoke.texi: Document +nodsp as a valid extension for
8554 2017-10-16 Martin Liska <mliska@suse.cz>
8556 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8557 (test_set_range): Likewise.
8558 (test_range_functions): Rename to ...
8559 (test_bit_in_range): ... this.
8560 (sbitmap_c_tests): Add new test.
8562 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8564 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8566 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8567 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8569 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8571 * config/aarch64/aarch64-builtins.c
8572 (aarch64_types_quadopu_lane_qualifiers): New.
8573 (TYPES_QUADOPU_LANE): New.
8574 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8575 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8576 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8577 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8578 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8579 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8580 (Vdottype, DOTPROD): New.
8581 (sur): Add SDOT and UDOT.
8583 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8585 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8586 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8587 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8589 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8590 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8591 Enable TARGET_DOTPROD.
8592 (cortex-a75.cortex-a55): Likewise.
8593 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8595 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8597 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8598 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8600 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8602 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8603 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8604 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8605 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8606 * config/arm/types.md (neon_dot, neon_dot_q): New.
8607 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8609 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8611 * config/arm/arm.h (TARGET_DOTPROD): New.
8612 * config/arm/arm.c (arm_arch_dotprod): New.
8613 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8614 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8615 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8616 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8617 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8618 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8619 * doc/invoke.texi (armv8.2-a): Document dotprod
8621 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8623 * i386.c (ix86_vec_cost): New function.
8624 (ix86_rtx_costs): Handle vector operations better.
8625 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8626 * x86-tune-costs.h: Add new costs to all tables.
8628 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8630 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8632 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8633 divsd, sqrtss and sqrtsd
8634 * x86-tune-costs.h: Add new entries to all costs.
8635 (znver1_cost): Fix to match real instruction latencies.
8637 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8638 Michael Collison <michael.collison@arm.com>
8640 * compare-elim.c: Include emit-rtl.h.
8641 (can_merge_compare_into_arith): New function.
8642 (try_validate_parallel): Likewise.
8643 (try_merge_compare): Likewise.
8644 (try_eliminate_compare): Call the above when no previous clobber
8646 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8649 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8653 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8654 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8658 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8659 to be any operand_equal_p operands. For & (B - 1) require
8660 B to be power of 2. Recognize
8661 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8663 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8666 * optabs.c (expand_memory_blockage): Fix call of
8667 targetm.have_memory_blockage.
8669 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8672 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8673 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8674 objects to extra_objs instead of overwriting it.
8676 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8678 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8679 Use any_fp_register_operand as operand[3] predicate. Simplify
8680 equality test for operands[2] and operands[4] memory location.
8681 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8682 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8683 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8684 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8685 any_fp_register_operand as operand[1] predicate. Simplify
8686 equality test for operands[0] and operands[3] memory location.
8687 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8688 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8689 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8691 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8693 * target-insns.def: Add memory_blockage.
8694 * optabs.c (expand_memory_blockage): New function.
8695 (expand_asm_memory_barrier): Rename ...
8696 (expand_asm_memory_blockage): ... to this.
8697 (expand_mem_thread_fence): Call expand_memory_blockage
8698 instead of expand_asm_memory_barrier.
8699 (expand_mem_singnal_fence): Ditto.
8700 (expand_atomic_load): Ditto.
8701 (expand_atomic_store): Ditto.
8702 * doc/md.texi (Standard Pattern Names For Generation):
8703 Document memory_blockage instruction pattern.
8705 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8707 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8708 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8709 * config/rl78/rl78.md: New define_expand "adddi3".
8711 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8713 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8716 2017-10-13 Jeff Law <law@redhat.com>
8718 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8720 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8723 * internal-fn.c (expand_mul_overflow): If both operands have
8724 the same highpart of -1 or 0 and the topmost bit of lowpart
8725 is different, overflow is if res <= 0 rather than res < 0.
8727 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8729 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8730 TARGET_P9_VECTOR code for unaligned_load case.
8732 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8734 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8736 2017-10-13 Nathan Sidwell <nathan@acm.org>
8738 * tree-core.h (tree_contains_struct): Make bool.
8739 * tree.c (tree_contains_struct): Likewise.
8740 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8741 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8742 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8743 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8744 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8746 2017-10-13 Richard Biener <rguenther@suse.de>
8748 * graphite-isl-ast-to-gimple.c
8749 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8750 parameters and dominance check.
8751 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8752 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8753 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8754 Do not update SSA form here or do intermediate IL verification.
8755 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8756 (graphite_initialize): Remove check on the number of loops in
8757 the function and inline into graphite_transform_loops.
8758 (graphite_finalize): Inline into graphite_transform_loops.
8759 (graphite_transform_loops): Perform SSA update and IL verification
8761 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8763 2017-10-13 Richard Biener <rguenther@suse.de>
8765 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8766 graphite_expression_type_precision): Avoid global constructor
8768 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8769 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8770 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8771 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8772 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8773 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8775 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8778 * config/i386/i386.h (ix86_red_zone_size): New.
8779 * config/i386/i386.md (push peephole2s): Replace
8780 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8782 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8783 Alan Hayward <alan.hayward@arm.com>
8784 David Sherwood <david.sherwood@arm.com>
8786 * combine.c (can_change_dest_mode): Reject changes in
8787 REGMODE_NATURAL_SIZE.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8794 (expand_debug_source_expr): Likewise.
8795 * combine.c (combine_simplify_rtx): Likewise.
8796 * cse.c (fold_rtx): Likewise.
8797 * fwprop.c (canonicalize_address): Likewise.
8798 * targhooks.c (default_shift_truncation_mask): Likewise.
8800 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8801 Alan Hayward <alan.hayward@arm.com>
8802 David Sherwood <david.sherwood@arm.com>
8804 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8805 (widened_mode): Likewise.
8806 (expand_unop): Likewise.
8807 * ree.c (transform_ifelse): Likewise.
8808 (merge_def_and_ext): Likewise.
8809 (combine_reaching_defs): Likewise.
8810 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8812 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8813 Alan Hayward <alan.hayward@arm.com>
8814 David Sherwood <david.sherwood@arm.com>
8816 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8817 * combine.c (gen_lowpart_for_combine): Likewise.
8818 * dwarf2out.c (rtl_for_decl_location): Likewise.
8819 * final.c (alter_subreg): Likewise.
8820 * rtlhooks.c (gen_lowpart_general): Likewise.
8821 (gen_lowpart_if_possible): Likewise.
8823 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8824 Alan Hayward <alan.hayward@arm.com>
8825 David Sherwood <david.sherwood@arm.com>
8827 * calls.c (expand_call): Use subreg_lowpart_offset.
8828 * cse.c (cse_insn): Likewise.
8829 * regcprop.c (copy_value): Likewise.
8830 (copyprop_hardreg_forward_1): Likewise.
8832 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8835 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8836 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8837 =Q constraints to +Q and into insn condition add check
8838 that operands[0] and operands[1] are equal.
8839 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8840 =Q constraints to +Q and into insn condition add check
8841 that operands[0] is equal to either operands[1] or operands[2].
8844 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8845 instead of handling MINUS_EXPR twice (once for each argument),
8846 canonicalize operand order and handle just once, use rtype where
8847 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8850 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8851 any values of __C while still being pattern recognizable as a simple
8854 2017-10-13 Richard Biener <rguenther@suse.de>
8856 PR tree-optimization/82451
8858 2017-10-02 Richard Biener <rguenther@suse.de>
8860 PR tree-optimization/82355
8861 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8862 a mapping for the enclosing loop but avoid generating one for
8864 (copy_bb_and_scalar_dependences): Remove premature codegen
8865 error on PHIs in blocks duplicated into multiple places.
8866 * graphite-scop-detection.c
8867 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8868 in the region use it as loop and nest to analyze the DR in.
8869 (try_generate_gimple_bb): Likewise.
8870 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8871 (add_loop_constraints): For blocks in a loop not in the region
8872 create a dimension with a single iteration.
8873 * sese.h (gbb_loop_at_index): Remove assert.
8875 * cfgloop.c (loop_preheader_edge): For the loop tree root
8876 return the single successor of the entry block.
8877 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8878 Reset the SCEV hashtable and niters.
8879 * graphite-scop-detection.c
8880 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8881 assert that we only have POLYNOMIAL_CHREC that vary in loops
8882 contained in the region.
8883 (scop_detection::graphite_can_represent_expr): Adjust.
8884 (scop_detection::stmt_has_simple_data_refs_p): For loops
8885 not in the region set loop to NULL. The nest is now the
8886 entry edge to the region.
8887 (try_generate_gimple_bb): Likewise.
8888 * sese.c (scalar_evolution_in_region): Adjust for
8889 instantiate_scev change.
8890 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8891 Make nest parameter the edge into the region.
8892 (create_data_ref): Likewise.
8893 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8894 entry edge into a region and adjust instantiate_scev calls.
8895 (create_data_ref): Likewise.
8896 (graphite_find_data_references_in_stmt): Likewise.
8897 (find_data_references_in_stmt): Pass the loop preheader edge
8898 from the nest argument.
8899 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8900 parameter the edge into the region.
8901 (instantiate_parameters): Use the loop preheader edge as entry.
8902 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8904 (get_instantiated_value_entry): Make instantiate_below parameter
8905 the edge into the region.
8906 (instantiate_scev_name): Likewise. Adjust dominance checks,
8907 when we cannot use loop-based instantiation instantiate by
8908 walking use-def chains.
8909 (instantiate_scev_poly): Adjust.
8910 (instantiate_scev_binary): Likewise.
8911 (instantiate_scev_convert): Likewise.
8912 (instantiate_scev_not): Likewise.
8913 (instantiate_array_ref): Remove.
8914 (instantiate_scev_3): Likewise.
8915 (instantiate_scev_2): Likewise.
8916 (instantiate_scev_1): Likewise.
8917 (instantiate_scev_r): Do not blindly handle N-operand trees.
8918 Do not instantiate array-refs. Handle all constants and invariants.
8919 (instantiate_scev): Make instantiate_below parameter
8920 the edge into the region.
8921 (resolve_mixers): Use the loop preheader edge for the region
8922 parameter to instantiate_scev_r.
8923 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8925 2017-10-13 Richard Biener <rguenther@suse.de>
8927 PR tree-optimization/82525
8928 * graphite-isl-ast-to-gimple.c
8929 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8931 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8932 Fail code generation when we cannot represent the isl integer.
8933 (binary_op_to_tree): Elide modulo operations that are no-ops
8934 in the type we code generate. Remove now superfluous code
8937 2017-10-13 Richard Biener <rguenther@suse.de>
8939 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8940 (scop_detection::harmful_loop_in_region): Remove premature
8941 IV type restriction.
8942 (scop_detection::graphite_can_represent_scev): We can handle
8943 pointer IVs just fine.
8945 2017-10-13 Alan Modra <amodra@gmail.com>
8947 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8948 "Clobbers and Scratch Registers". Add paragraph on
8949 alternative to clobbers for scratch registers and OpenBLAS
8952 2017-10-13 Alan Modra <amodra@gmail.com>
8954 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8955 example of a memory input for a string of known length. Move
8956 commentary out of table. Add a number of new examples
8957 covering array memory inputs.
8959 2017-10-12 Martin Liska <mliska@suse.cz>
8961 PR tree-optimization/82493
8962 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8963 (test_range_functions): New function.
8964 (sbitmap_c_tests): Likewise.
8965 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8966 * selftest.h (sbitmap_c_tests): New function.
8968 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8970 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8972 * config/rs6000/amo.h: Fix spacing issue.
8974 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8977 * config/i386/i386.md (*ashl<mode>3_mask_1,
8978 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8979 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8982 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8984 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8985 (profile_probability): Set max_probability
8986 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8988 * profile-count.c (profile_probability::differs_from_p): Do not
8989 rely on max_probaiblity == 10000
8991 2017-10-12 Jeff Law <law@redhat.com>
8993 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8996 2017-10-12 Martin Sebor <msebor@redhat.com>
9000 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9001 (handle_alias_pairs): Call it.
9002 * common.opt (-Wattribute-alias): New option.
9003 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9004 * doc/invoke.texi (-Wattribute-alias): Document.
9006 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9009 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9011 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9013 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9014 (make_hard_regno_born, make_hard_regno_dead): Update
9015 bb_killed_pseudos and bb_gen_pseudos.
9017 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9019 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9021 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9023 * config/alpha/alpha.c (alpha_split_conditional_move):
9024 Use std::swap instead of manually swapping.
9025 (alpha_stdarg_optimize_hook): Ditto.
9026 (alpha_canonicalize_comparison): Ditto.
9028 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9030 * tree-loop-distribution.c (struct builtin_info): New struct.
9031 (struct partition): Refactor fields into struct builtin_info.
9032 (partition_free): Free struct builtin_info.
9033 (build_size_arg_loc, build_addr_arg_loc): Delete.
9034 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9035 information from struct builtin_info.
9036 (find_single_drs): New function refactored from classify_partition.
9037 Also moved builtin validity checks to this function.
9038 (compute_access_range, alloc_builtin): New functions.
9039 (classify_builtin_st, classify_builtin_ldst): New functions.
9040 (classify_partition): Refactor code into functions find_single_drs,
9041 classify_builtin_st and classify_builtin_ldst.
9042 (distribute_loop): Don't do runtime alias check when distributing
9044 (find_seed_stmts_for_distribution): New function.
9045 (pass_loop_distribution::execute): Refactor code finding seed
9046 stmts into above function. Support distribution for the innermost
9047 two-level loop nest. Adjust dump information.
9049 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9051 * tree-loop-distribution.c: Adjust the general comment.
9052 (NUM_PARTITION_THRESHOLD): New macro.
9053 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9054 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9055 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9056 in call to build_partition_graph.
9057 (finalize_partitions): New parameter. Make loop distribution more
9058 conservative by fusing more partitions.
9059 (distribute_loop): Don't do runtime alias check in case of loop nest
9061 (find_seed_stmts_for_distribution): New function.
9062 (prepare_perfect_loop_nest): New function.
9063 (pass_loop_distribution::execute): Refactor code finding seed stmts
9064 and loop nest into above functions. Support loop nest distribution.
9065 Adjust dump information accordingly.
9067 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9069 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9070 and set PTYPE_SEQUENTIAL for merged partition.
9072 2017-10-12 Richard Biener <rguenther@suse.de>
9074 PR tree-optimization/69728
9076 2017-09-19 Richard Biener <rguenther@suse.de>
9078 PR tree-optimization/69728
9079 * graphite-sese-to-poly.c (schedule_error): New global.
9080 (add_loop_schedule): Handle empty domain by failing the
9082 (build_original_schedule): Handle schedule_error.
9084 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9085 domain by returning an unchanged schedule.
9087 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9089 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9090 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9092 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9094 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9097 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9100 * expr.c (store_field): Don't optimize away bitsize == 0 store
9101 from CALL_EXPR with addressable return type.
9103 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9105 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9106 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9107 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9108 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9109 TARGET_ISEL instead of TARGET_ISEL<sel>.
9111 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9113 * config/rs6000/rs6000.c
9114 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9116 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9118 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9120 (reg_or_cint_operand): Fix comment.
9121 (reg_or_zero_operand): New predicate.
9122 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9123 * config/rs6000/rs6000.c (output_isel): Delete.
9124 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9125 instead of reg_or_cint_operand. Output instruction directly (not via
9127 (isel_unsigned_<mode>): Ditto.
9128 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9129 gpc_reg_operand. Add an instruction alternative for this. Output
9130 instruction directly.
9131 (*isel_reversed_unsigned_<mode>): Ditto.
9133 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9135 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9136 (TARGET_CANONICALIZE_COMPARISON): Define.
9138 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9141 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9142 Check whether the dest is REG before adding REG_EQUIV note.
9144 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9147 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9149 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9150 (make_hard_regno_born, make_hard_regno_dead): Update
9151 bb_killed_pseudos and bb_gen_pseudos.
9153 2017-10-11 Nathan Sidwell <nathan@acm.org>
9155 * incpath.h (enum incpath_kind): Name enum, prefix values.
9156 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9157 * incpath.c (heads, tails): Use INC_MAX.
9158 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9159 (merge_include_chains, split_quote_chain,
9160 register_include_chains): Update incpath_kind names.
9161 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9162 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9164 (add_framework_path, darwin_register_objc_includes): Likewise.
9165 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9167 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9169 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9170 Do not use float_operator operator predicate.
9171 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9172 * config/i386/predicates.md (float_operator): Remove predicate.
9174 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9176 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9177 (*jccxf_i387): Ditto.
9178 (*jcc<mode>_i387): Ditto.
9179 (*jccu<mode>_i387): Ditto.
9180 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9181 (*jcc_*_i387 splitters): Remove.
9182 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9183 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9184 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9187 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9189 * profile-count.h (slow_safe_scale_64bit): New function.
9190 (safe_scale_64bit): New inline.
9191 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9192 * profile-count.c: Include wide-int.h
9193 (slow_safe_scale_64bit): New.
9195 2017-10-11 Nathan Sidwell <nathan@acm.org>
9197 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9198 HAS_DECL_ASSEMBLER_NAME_P.
9199 * gimple-expr.c (gimple_decl_printable_name: Check
9200 HAS_DECL_ASSEMBLER_NAME_P too.
9201 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9202 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9203 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9204 * passes.c (rest_of_decl_compilation): Check
9205 HAS_DECL_ASSEMBLER_NAME_P too.
9206 * recog.c (verify_changes): Likewise.
9207 * tree-pretty-print.c (dump_decl_name): Likewise.
9208 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9210 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9211 (SET_DECL_ASSEMBLER_NAME): Use it.
9212 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9213 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9214 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9216 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9218 * config.gcc (i386, x86_64): Add extra objects.
9219 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9220 (ix86_min_insn_size): Declare.
9221 (ix86_issue_rate): Declare.
9222 (ix86_adjust_cost): Declare.
9223 (ia32_multipass_dfa_lookahead): Declare.
9224 (ix86_macro_fusion_p): Declare.
9225 (ix86_macro_fusion_pair_p): Declare.
9226 (ix86_bd_has_dispatch): Declare.
9227 (ix86_bd_do_dispatch): Declare.
9228 (ix86_core2i7_init_hooks): Declare.
9229 (ix86_atom_sched_reorder): Declare.
9230 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9231 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9232 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9233 (rip_relative_addr_p): Rename to ...
9234 (ix86_rip_relative_addr_p): ... this one; export.
9235 (memory_address_length): Update.
9236 (ix86_issue_rate): Move to x86-tune-sched.c.
9237 (ix86_flags_dependent): Move to x86-tune-sched.c.
9238 (ix86_agi_dependent): Move to x86-tune-sched.c.
9239 (exact_dependency_1): Move to x86-tune-sched.c.
9240 (exact_store_load_dependency): Move to x86-tune-sched.c.
9241 (ix86_adjust_cost): Move to x86-tune-sched.c.
9242 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9243 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9244 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9245 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9246 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9247 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9248 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9249 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9250 (min_insn_size): Rename to ...
9251 (ix86_min_insn_size): ... this one; export.
9252 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9253 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9254 (core2i7_first_cycle_multipass_backtrack): Move to
9255 x86-tune-sched-core.c.
9256 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9257 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9258 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9259 (ix86_avoid_jump_mispredicts): Update.
9260 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9261 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9262 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9263 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9264 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9265 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9266 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9267 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9268 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9269 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9270 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9271 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9272 (BIG): Move to ix86-tune-sched-bd.c.
9273 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9274 (enum insn_path): Move to ix86-tune-sched-bd.c.
9275 (get_mem_group): Move to ix86-tune-sched-bd.c.
9276 (is_cmp): Move to ix86-tune-sched-bd.c.
9277 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9278 (is_branch): Move to ix86-tune-sched-bd.c.
9279 (is_prefetch): Move to ix86-tune-sched-bd.c.
9280 (init_window): Move to ix86-tune-sched-bd.c.
9281 (allocate_window): Move to ix86-tune-sched-bd.c.
9282 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9283 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9284 (process_end_window): Move to ix86-tune-sched-bd.c.
9285 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9286 (find_constant): Move to ix86-tune-sched-bd.c.
9287 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9288 (has_immediate): Move to ix86-tune-sched-bd.c.
9289 (get_insn_path): Move to ix86-tune-sched-bd.c.
9290 (get_insn_group): Move to ix86-tune-sched-bd.c.
9291 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9292 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9293 (add_insn_window): Move to ix86-tune-sched-bd.c.
9294 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9295 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9296 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9297 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9298 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9299 (do_dispatch): Move to ix86-tune-sched-bd.c.
9300 (has_dispatch): Move to ix86-tune-sched-bd.c.
9301 * config/i386/t-i386: Add new object files.
9302 * config/i386/x86-tune-costs.h: New file.
9303 * config/i386/x86-tune-sched-atom.c: New file.
9304 * config/i386/x86-tune-sched-bd.c: New file.
9305 * config/i386/x86-tune-sched-core.c: New file.
9306 * config/i386/x86-tune-sched.c: New file.
9308 2017-10-11 Liu Hao <lh_mouse@126.com>
9310 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9311 the generic version below instead.
9312 (should_colorize): Recognize Windows consoles as terminals
9314 * pretty-print.c [__MINGW32__] (write_all): New function.
9315 [__MINGW32__] (find_esc_head): Likewise.
9316 [__MINGW32__] (find_esc_terminator): Likewise.
9317 [__MINGW32__] (eat_esc_sequence): Likewise.
9318 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9320 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9323 2017-10-11 Richard Biener <rguenther@suse.de>
9325 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9326 Properly call analyze_scalar_evolution with the loop of the stmt.
9328 2017-10-11 Richard Biener <rguenther@suse.de>
9330 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9331 * tree-core.h (tree_base): Add chrec_var union member.
9332 * tree.h (CHREC_VAR): Remove.
9333 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9334 * tree-chrec.h (build_polynomial_chrec): Adjust.
9335 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9336 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9338 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9340 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9341 * match.pd: ... here.
9342 ((T) X == (T) Y): Relax condition.
9344 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9346 PR tree-optimization/82472
9347 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9349 (break_alias_scc_partitions): Update postorder number.
9351 2017-10-11 Martin Liska <mliska@suse.cz>
9354 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9356 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9358 2017-10-11 Martin Liska <mliska@suse.cz>
9360 * print-rtl.c (print_insn): Move declaration of idbuf
9361 to same scope as name.
9363 2017-10-11 Martin Liska <mliska@suse.cz>
9368 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9371 2017-10-11 Martin Liska <mliska@suse.cz>
9374 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9377 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9379 * config.gcc (default_gnu_indirect_function): Default to yes for
9380 arm*-*-linux* with glibc.
9382 2017-10-11 Richard Biener <rguenther@suse.de>
9384 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9385 default-defs and types we do not want to analyze.
9386 (interpret_loop_phi): Replace unreachable code with an assert.
9387 (compute_scalar_evolution_in_loop): Remove and inline ...
9388 (analyze_scalar_evolution_1): ... here, replacing condition with
9389 what makes the intent clearer. Remove handling of cases
9390 get_scalar_evolution now handles.
9392 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9394 PR rtl-optimization/81434
9395 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9396 comment for main loop. In sched_group_found if, also add checks for
9397 pass and min_cost_group.
9399 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9401 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9402 (rs6000_insn_cost): New function.
9403 * config/rs6000/rs6000.md (cost): New attribute.
9405 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9406 H.J. Lu <hongjiu.lu@intel.com>
9410 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9411 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9412 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9413 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9414 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9415 OPTION_MASK_ISA_MMX and some other bit set require both
9416 mmx and the ISAs without the mmx bit.
9417 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9418 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9419 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9420 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9421 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9422 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9423 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9424 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9425 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9426 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9427 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9428 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9429 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9430 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9431 Add OPTION_MASK_ISA_MMX.
9433 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9435 * config.gcc (armv7*-*-freebsd*): New target.
9436 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9438 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9440 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9441 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9444 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9446 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9447 a template parameter.
9448 (WIDE_INT_REF_FOR): Update accordingly.
9449 * tree.h (wi::int_traits <const_tree>): Delete.
9450 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9451 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9452 (wi::tree_to_wide_ref): New typedef.
9453 (wi::to_wide): New function.
9454 * calls.c (get_size_range): Use wi::to_wide when operating on
9456 * cgraph.c (cgraph_node::create_thunk): Likewise.
9457 * config/i386/i386.c (ix86_data_alignment): Likewise.
9458 (ix86_local_alignment): Likewise.
9459 * dbxout.c (stabstr_O): Likewise.
9460 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9461 * expr.c (const_vector_from_tree): Likewise.
9462 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9463 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9464 (fold_negate_expr_1, int_const_binop_1, const_binop)
9465 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9466 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9467 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9468 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9469 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9470 (fold_not_const, round_up_loc): Likewise.
9471 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9472 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9473 (alloca_call_type): Likewise.
9474 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9475 * godump.c (go_output_typedef): Likewise.
9476 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9477 * internal-fn.c (get_min_precision): Likewise.
9478 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9479 * ipa-polymorphic-call.c
9480 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9481 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9482 (ipa_modify_call_arguments): Likewise.
9483 * match.pd: Likewise.
9484 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9485 * print-tree.c (print_node_brief, print_node): Likewise.
9486 * stmt.c (expand_case): Likewise.
9487 * stor-layout.c (layout_type): Likewise.
9488 * tree-affine.c (tree_to_aff_combination): Likewise.
9489 * tree-cfg.c (group_case_labels_stmt): Likewise.
9490 * tree-data-ref.c (dr_analyze_indices): Likewise.
9491 (prune_runtime_alias_test_list): Likewise.
9492 * tree-dump.c (dequeue_and_dump): Likewise.
9493 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9494 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9495 * tree-pretty-print.c (dump_generic_node): Likewise.
9496 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9497 (simple_iv_with_niters): Likewise.
9498 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9499 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9500 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9501 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9502 (refine_value_range_using_guard, number_of_iterations_ne_max)
9503 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9504 (get_cst_init_from_scev, record_nonwrapping_iv)
9505 (scev_var_range_cant_overflow): Likewise.
9506 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9507 * tree-ssa-pre.c (compute_avail): Likewise.
9508 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9509 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9510 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9511 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9512 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9513 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9514 (dump_case_nodes, try_switch_expansion): Likewise.
9515 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9516 (vect_do_peeling): Likewise.
9517 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9518 * tree-vect-stmts.c (vectorizable_load): Likewise.
9519 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9520 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9521 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9522 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9523 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9524 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9525 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9526 (evrp_dom_walker::before_dom_children): Likewise.
9527 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9528 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9529 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9530 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9531 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9532 (get_range_pos_neg): Likewise.
9533 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9534 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9535 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9536 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9537 * config/avr/avr.c (avr_fold_builtin): Likewise.
9538 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9539 * config/msp430/msp430.c (msp430_attr): Likewise.
9540 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9541 * config/powerpcspe/powerpcspe-c.c
9542 (altivec_resolve_overloaded_builtin): Likewise.
9543 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9544 (rs6000_expand_ternop_builtin): Likewise.
9545 * config/rs6000/rs6000-c.c
9546 (altivec_resolve_overloaded_builtin): Likewise.
9547 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9548 (rs6000_expand_ternop_builtin): Likewise.
9549 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9551 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9553 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9554 when copying loop nest with only one inner loop.
9556 2017-10-10 Richard Biener <rguenther@suse.de>
9558 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9559 blocks if SCEV is active.
9560 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9562 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9563 (scev_initialize): Assert we are not yet initialized.
9565 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9567 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9568 inner loop's exit stmt by making it always exit the loop, otherwise
9569 we would generate an infinite empty loop.
9571 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9573 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9574 renaming variables in new preheader if it's deleted.
9576 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9578 * tree-loop-distribution.c (struct partition): Remove unused field
9579 loops of the structure.
9580 (partition_alloc, partition_free): Ditto.
9581 (build_rdg_partition_for_vertex): Ditto.
9583 2017-10-09 Jeff Law <law@redhat.com>
9585 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9586 return type to match prototype and documentation.
9588 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9590 * config/rs6000/rs6000.c (processor_costs): Move to ...
9591 * config/rs6000/rs6000.h: ... here.
9592 (rs6000_cost): Declare.
9594 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9596 * except.c (setjmp_fn): New global variable.
9597 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9598 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9599 if DONT_USE_BUILTIN_SETJMP is defined.
9601 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9603 * target.def (insn_cost): New hook.
9604 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9605 * doc/tm.texi: Regenerate.
9606 * rtlanal.c (insn_cost): Use the new hook.
9608 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9610 * combine.c (combine_validate_cost): Compute the new insn_cost,
9611 not just pattern_cost.
9612 (try_combine): Adjust comment.
9614 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9616 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9618 * combine.c (uid_insn_cost): Adjust comment.
9619 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9621 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9622 * dse.c (find_shift_sequence): Ditto.
9623 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9624 (bb_valid_for_noce_process_p): Use pattern_cost.
9625 * rtl.h (insn_rtx_cost): Delete.
9626 (pattern_cost): New prototype.
9627 (insn_cost): New prototype.
9628 * rtlanal.c (insn_rtx_cost): Rename to...
9629 (pattern_cost): ... this.
9632 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9634 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9635 (*jcc<mode>_0_r_i387): Ditto.
9636 (*jccxf_r_i387): Ditto.
9637 (*jcc<mode>_r_i387): Ditto.
9638 (*jccu<mode>_r_i387): Ditto.
9639 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9640 (*jcc): Rename from *jcc_1.
9642 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9644 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9645 deferred rescans after the lvx/stvx recombination pre-pass.
9647 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9649 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9650 memory operation instruction support.
9651 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9652 (rs6000-ibm-aix[789]*): Likewise.
9653 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9654 Document new functions.
9656 2017-10-09 Richard Biener <rguenther@suse.de>
9658 PR tree-optimization/82397
9659 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9660 equality only for semantically equal trees.
9662 2017-10-09 Richard Biener <rguenther@suse.de>
9664 PR tree-optimization/82449
9665 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9666 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9667 allow constant addresses.
9668 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9671 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9673 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9676 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9679 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9682 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9685 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9687 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9690 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9691 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9693 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9695 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9696 (WI_BINARY_PREDICATE_RESULT): Likewise.
9697 (wi::binary_traits::operator_result): New type.
9698 (wi::binary_traits::predicate_result): Likewise.
9699 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9700 (generic_wide_int::operator==, generic_wide_int::operator!=)
9701 (generic_wide_int::operator&, generic_wide_int::and_not)
9702 (generic_wide_int::operator|, generic_wide_int::or_not)
9703 (generic_wide_int::operator^, generic_wide_int::operator+
9704 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9706 (operator~, unary operator-, operator==, operator!=, operator&)
9707 (operator|, operator^, operator+, binary operator-, operator*): New
9709 * expr.c (get_inner_reference): Use wi::bit_and_not.
9710 * fold-const.c (fold_binary_loc): Likewise.
9711 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9712 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9713 (bit_value_binop): Likewise.
9714 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9715 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9716 (extract_range_from_binary_expr_1): Likewise.
9717 (masked_increment): Likewise.
9718 (simplify_bit_ops_using_ranges): Likewise.
9720 2017-10-09 Martin Jambor <mjambor@suse.cz>
9723 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9724 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9725 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9727 (hsa_fixup_mov_insn_type): New function.
9728 (hsa_op_with_type::get_in_type): Use it.
9729 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9730 immediates in an assert.
9731 (hsa_op_with_type::extend_int_to_32bit): New method.
9732 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9733 types. Convert to dest type if necessary.
9734 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9735 (reg_for_gimple_ssa): Pass false as min32int to
9736 hsa_type_for_scalar_tree_type.
9737 (gen_hsa_addr): Fixup type when creating addresable temporary.
9738 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9739 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9740 necessary. Call hsa_fixup_mov_insn_type.
9741 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9742 extend operands and convert to dest type if necessary.
9743 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9744 to dest type if necessary.
9745 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9746 if conversion nt necessary and size matches.
9747 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9748 to dest type if necessary.
9749 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9750 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9752 (gen_hsa_clrsb): Likewise.
9753 (gen_hsa_ffs): Likewise.
9754 (gen_hsa_divmod): Extend operands and convert to dest type if
9756 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9758 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9760 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9761 Remove empty default arguments. Use a brace block as output
9763 (conditional return): Ditto.
9765 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9766 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9768 (group_ending_nop): Ditto.
9769 (doloop_end): Ditto.
9770 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9771 (splitters for those): Ditto.
9773 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9775 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9776 a conditional jump (and the compare for it) so that pc_rtx is the
9778 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9779 for the deleted and renamed ctr<mode>_internal[234] patterns.
9780 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9781 Delete second conditional return pattern.
9782 (ctr<mode>_internal2): Delete this second bdnz pattern.
9783 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9784 (ctr<mode>_internal4): Delete this second bdz pattern.
9786 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9788 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9789 (always_initialized_rtx_for_ssa_name_p): New predicate.
9790 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9791 (finish_out_of_ssa): Free new field of SA.
9792 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9793 * tree-ssa-coalesce.c: Include tree-ssa.h.
9794 (get_parm_default_def_partitions): Remove extern keyword.
9795 (get_undefined_value_partitions): New function.
9796 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9797 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9799 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9801 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9803 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9805 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9807 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9808 for targets that preffer 128bit.
9810 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9812 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9814 2017-10-08 Olivier Hainque <hainque@adacore.com>
9816 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9817 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9819 (thumb_set_return_address): Likewise.
9821 2017-10-08 Olivier Hainque <hainque@adacore.com>
9823 * common/config/arm/arm-common.c (arm_except_unwind_info):
9824 Handle DWARF2_UNWIND_INFO.
9826 2017-10-07 Michael Collison <michael.collison@arm.com>
9828 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9831 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9833 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9834 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9835 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9836 defined, force the creation of a new block for a dispatch label.
9838 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9840 * invoke.texi (Wsuggest-attribute=cold): Document.
9841 * common.opt (Wsuggest-attribute=cold): New
9842 * ipa-pure-const.c (warn_function_cold): New function.
9843 * predict.c (compute_function_frequency): Use it.
9844 * predict.h (warn_function_cold): Declare.
9846 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9848 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9851 2017-10-06 Martin Liska <mliska@suse.cz>
9853 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9854 keyword for member functions.
9855 (struct sanopt_tree_couple): New struct.
9856 (struct sanopt_tree_couple_hash): New function.
9857 (struct sanopt_ctx): Add new hash_map.
9858 (has_dominating_ubsan_ptr_check): New function.
9859 (record_ubsan_ptr_check_stmt): Likewise.
9860 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9861 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9862 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9864 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9867 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9868 aarch64_simd_valid_immediate on CONST_VECTORs.
9869 (aarch64_reg_or_bic_imm): Likewise.
9871 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9873 PR rtl-optimization/82396
9874 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9876 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9878 * graphite-dependences.c (scop_get_reads): Move code to...
9879 (scop_get_must_writes): Move code to...
9880 (scop_get_may_writes): Move code to...
9881 (scop_get_reads_and_writes): ... here.
9882 (scop_get_dependences): Call scop_get_reads_and_writes.
9884 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9886 PR tree-optimization/82434
9887 * fold-const.h (can_native_encode_type_p,
9888 can_native_encode_string_p): Remove.
9889 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9890 don't encode anything, just return what would be otherwise returned.
9891 (native_encode_fixed, native_encode_complex, native_encode_vector):
9893 (native_encode_string): Likewise. Inline by hand
9894 can_native_encode_string_p.
9895 (can_native_encode_type_p): Remove.
9896 (can_native_encode_string_p): Remove.
9897 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9898 STRING_CSTs using can_native_encode_string_p, test all
9899 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9900 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9901 argument from native_encode_expr.
9902 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9903 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9906 2017-10-06 Richard Biener <rguenther@suse.de>
9908 PR tree-optimization/82397
9909 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9910 operand_equal_p but rely on data_ref_compare_tree for detecting
9912 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9913 to match up with dr_group_sort_cmp.
9915 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9918 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9920 * config/s390/s390-builtin-types.def: Regenerate.
9922 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9925 * config/s390/s390-builtin-types.def: Regenerate.
9926 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9927 Change flag from B_VXE to B_VX.
9928 (s390_vec_min_dbl): Remove B_VXE flag.
9930 2017-10-06 Richard Biener <rguenther@suse.de>
9932 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9933 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9934 translate_isl_ast_to_gimple::is_valid_rename,
9935 translate_isl_ast_to_gimple::get_rename,
9936 translate_isl_ast_to_gimple::get_def_bb_for_const,
9937 translate_isl_ast_to_gimple::get_new_name,
9938 translate_isl_ast_to_gimple::collect_all_ssa_names,
9939 translate_isl_ast_to_gimple::copy_loop_phi_args,
9940 translate_isl_ast_to_gimple::collect_all_ssa_names,
9941 translate_isl_ast_to_gimple::copy_loop_phi_args,
9942 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9943 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9944 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9945 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9946 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9947 translate_isl_ast_to_gimple::copy_cond_phi_args,
9948 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9949 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9950 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9951 translate_isl_ast_to_gimple::rename_uses,
9952 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9953 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9954 (set_rename_for_each_def): Likewise.
9955 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9956 here. Handle rewriting SCEV analyzable uses here.
9957 (copy_bb_and_scalar_dependences): Generate code for PHI
9959 (graphite_regenerate_ast_isl): Adjust.
9960 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9961 (add_write, add_read): New functions.
9962 (build_cross_bb_scalars_def): Use it and simplify.
9963 (build_cross_bb_scalars_use): Likewise.
9964 (graphite_find_cross_bb_scalar_vars): Inline into...
9965 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9966 simulating out-of-SSA. Compute liveout and add dependencies.
9967 (build_scops): Force an empty entry block.
9968 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9970 (sese_build_liveouts): Declare.
9971 (sese_trivially_empty_bb_p): Likewise.
9972 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9973 compute liveout and debug_liveout.
9974 (sese_bad_liveouts_use): Remove.
9975 (sese_reset_debug_liveouts_bb): Likewise.
9976 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9977 (sese_build_liveouts): Build liveout and debug_liveout and store
9979 (new_sese_info): Adjust.
9980 (free_sese_info): Likewise.
9981 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9982 do not build liveout here.
9983 (move_sese_in_condition): Adjust region entry.
9984 (scev_analyzable_p): Match up with chrec_apply requirements.
9985 (sese_trivially_empty_bb_p): New.
9986 * tree-into-ssa.c (get_reaching_def): Properly support generating
9987 default-defs for incremental rewrite of anonymous names.
9989 2017-10-06 Richard Biener <rguenther@suse.de>
9991 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9992 precision do not perform modulo reduction.
9994 2017-10-06 Richard Biener <rguenther@suse.de>
9996 PR tree-optimization/82436
9997 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9998 conservatively choose the vectorization factor when checking
9999 whether we can perform the required load permutation.
10000 (vect_transform_slp_perm_load): Assert when we may not fail.
10002 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10004 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10005 message for incompatible -msdata=* and -mcall-* options.
10007 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10009 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10010 rate for post-reload scheduling.
10012 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10014 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10016 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10018 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10019 to improve monte carlo in scimark.
10021 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10023 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10024 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10025 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10026 pentium4_cost, nocona_cost): Set reassociation width to 1.
10027 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10028 width to 2 for fp operations and 1 otherwise.
10029 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10031 (atom_cost): Set reassociation width to 2.
10032 (slm_cost, generic_cost): Set fp reassociation width
10033 to 2 and 1 otherwise.
10034 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10035 (core_cost): Set fp reassociation width to 4 and vector to 2.
10036 (ix86_reassociation_width): Rewrite using cost table; special case
10037 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10038 and TARGET_AVX128_OPTIMAL.
10039 * config/i386/i386.h (processor_costs): Add
10040 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10041 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10042 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10043 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10044 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10045 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10047 2017-10-05 Nathan Sidwell <nathan@acm.org>
10049 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10051 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10053 * config/arm/arm.c (arm_test_fpu_data): New.
10054 (arm_run_selftests): Call arm_test_fpu_data.
10056 2017-10-04 Nathan Sidwell <nathan@acm.org>
10058 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10060 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10062 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10064 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10065 check type for aarch64_simd_valid_immediate.
10066 (aarch64_output_simd_mov_immediate): Update prototype.
10067 (aarch64_simd_valid_immediate): Update prototype.
10068 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10069 support for ORR-immediate.
10070 (and<mode>3): modified pattern to add support for BIC-immediate.
10071 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10072 now checks for valid immediate for BIC and ORR based on new enum
10074 (aarch64_output_simd_mov_immediate): Function now used to output
10075 BIC/ORR imm as well based on new enum argument.
10076 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10078 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10079 (aarch64_reg_or_bic_imm): Likewise.
10081 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10083 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10084 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10086 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10090 PR rtl-optimization/82396
10091 * haifa-sched.c (autopref_multipass_init): Simplify
10093 (autopref_rank_data): Simplify sort order.
10094 * sched-int.h (autopref_multipass_data_): Remove
10095 multi_mem_insn_p, min_offset and max_offset.
10097 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10099 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10101 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10103 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10104 vect_doubleint_cvt.
10106 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10108 * doc/sourcebuild.texi: Document vect_long_mult.
10110 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10112 PR tree-optimization/82413
10113 * fold-const.c (build_range_check): Use widest_int when comparing
10114 the maximum ETYPE value with HIGH.
10116 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10118 PR rtl-optimization/82396
10119 * haifa-sched.c (autopref_multipass_init): Simplify
10121 (autopref_rank_data): Simplify sort order.
10122 * sched-int.h (autopref_multipass_data_): Remove
10123 multi_mem_insn_p, min_offset and max_offset.
10125 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10127 PR tree-optimization/82381
10128 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10129 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10132 PR tree-optimization/82374
10133 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10134 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10135 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10136 current_function_decl to the new decl.
10138 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10140 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10141 helper macro for IEEE float128 hardware built-in functions.
10142 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10144 (TRUNCF128_ODD): Likewise.
10145 (ADDF128_ODD): Likewise.
10146 (SUBF128_ODD): Likewise.
10147 (MULF128_ODD): Likewise.
10148 (DIVF128_ODD): Likewise.
10149 (FMAF128_ODD): Likewise.
10150 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10151 UNSPEC_TRUNC_ROUND_TO_ODD.
10152 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10153 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10154 floating point round to odd instructions.
10155 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10156 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10157 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10158 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10159 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10160 (trunc<mode>sf2_hw): Change the truncate with round to odd
10161 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10162 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10163 to odd hardware instructions.
10164 (sub<mode>3_odd): Likewise.
10165 (mul<mode>3_odd): Likewise.
10166 (div<mode>3_odd): Likewise.
10167 (sqrt<mode>2_odd): Likewise.
10168 (fma<mode>4_odd): Likewise.
10169 (fms<mode>4_odd): Likewise.
10170 (nfma<mode>4_odd): Likewise.
10171 (nfms<mode>4_odd): Likewise.
10172 (trunc<mode>df2_odd): Change the truncate with round to odd
10173 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10175 * doc/extend.texi (PowerPC built-in functions): Update documentation
10176 for existing IEEE float128-bit built-in functions. Add built-in
10177 functions that generate the IEEE 128-bit floating point round to
10180 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10182 PR rtl-optimization/77729
10183 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10184 to (X&(C1&~C2))|C2 transformations.
10186 2017-10-03 Martin Jambor <mjambor@suse.cz>
10188 PR tree-optimization/82363
10189 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10190 mismatch, mark lacc written regardless of racc.
10192 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10194 PR tree-optimization/82381
10195 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10196 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10197 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10198 fallthrough into reassoc_stmt_dominates_stmt_p.
10201 * combine.c (combine_instructions): Don't combine in unreachable
10204 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10208 function to not use the have_cpu variable. Do not set cpu_index,
10209 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10210 or the default cpu.
10211 (rs6000_valid_attribute_p): Remove duplicate initializations of
10212 old_optimize and func_optimize.
10213 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10214 (rs6000_activate_target_options): Make global.
10215 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10218 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10220 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10221 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10222 Set *poffset to 0 and *psize and *pmax_size to -1 if
10223 *poffset + *psize overflows in HOST_WIDE_INT.
10225 PR tree-optimization/82387
10226 PR tree-optimization/82388
10227 PR tree-optimization/82389
10228 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10229 instead of live_bytes non-NULL.
10231 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10234 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10237 2017-10-02 Richard Biener <rguenther@suse.de>
10239 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10240 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10241 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10243 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10245 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10246 requested precision matches the type's.
10247 * calls.c (alloc_max_size): Calculate the new candidate size as
10248 a widest_int and use wi::to_widest when comparing it with the
10249 current candidate size.
10250 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10251 zero rather than integer_zero_node.
10252 * match.pd: Check for a no-op conversion before using wi::add
10253 rather than after. Use tree_to_uhwi when summing small shift
10254 counts into an unsigned int.
10256 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10257 Alan Hayward <alan.hayward@arm.com>
10258 David Sherwood <david.sherwood@arm.com>
10261 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10262 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10263 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10264 POINTER_AND_FP_REGS.
10266 2017-10-02 Richard Biener <rguenther@suse.de>
10268 PR tree-optimization/82355
10269 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10270 a mapping for the enclosing loop but avoid generating one for
10271 the loop tree root.
10272 (copy_bb_and_scalar_dependences): Remove premature codegen
10273 error on PHIs in blocks duplicated into multiple places.
10274 * graphite-scop-detection.c
10275 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10276 in the region use it as loop and nest to analyze the DR in.
10277 (try_generate_gimple_bb): Likewise.
10278 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10279 (add_loop_constraints): For blocks in a loop not in the region
10280 create a dimension with a single iteration.
10281 * sese.h (gbb_loop_at_index): Remove assert.
10283 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10285 * omp-expand.c (adjust_context_scope): New function.
10286 (expand_parallel_call): Call adjust_context_scope.
10288 2017-10-01 Jeff Law <law@redhat.com>
10290 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10291 dom_opt_dom_walker class with direct access to private members.
10292 Add comments. Call test_for_singularity.
10293 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10294 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10295 m_dummy_cond anymore.
10296 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10298 (pass_dominator:execute): Build the dummy_cond here and pass it
10299 to the dom_opt_dom_walker ctor.
10300 (test_for_singularity): New function.
10302 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10303 Maya Rashish <coypu@sdf.org>
10305 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10306 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10307 (alpha*-*-netbsd*) Use nbsd_tm_file.
10308 (arm*-*-netbsdelf*) Likewise.
10309 (i[34567]86-*-netbsdelf*) Likewise.
10310 (x86_64-*-netbsd*) Likewise.
10311 (mips*-*-netbsd*) Likewise.
10312 (powerpc-*-netbsd*) Likewise.
10313 (sh*-*-netbsd*) Likewise.
10314 (sparc-*-netbsdelf*) Likewise.
10315 (sparc64-*-netbsd*) Likewise.
10316 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10318 (vax-*-netbsdelf*) Likewise.
10319 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10320 (UINT_FAST8_TYPE) Likewise.
10321 (INT_FAST16_TYPE) Check CHAR_FAST16.
10322 (UINT_FAST16_TYPE) Likewise.
10324 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10327 * config/i386/i386.md
10328 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10329 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10330 *divmodsi4_zext_2): New define_insn_and_split.
10331 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10332 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10333 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10334 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10335 New define_insn_and_split.
10336 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10337 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10338 operands[1] having DImode when mode is SImode.
10340 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10341 always SImode for DIV and MOD in REG_EQUAL notes.
10343 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10345 PR middle-end/82319
10346 * match.pd: Fix handling of NaNs in pattern.
10348 2017-09-29 Jeff Law <law@redhat.com>
10350 * sbitmap.c (bitmap_bit_in_range_p): New function.
10351 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10352 * tree-ssa-dse.c (live_bytes_read): New function.
10353 (dse_classify_store): Ignore reads of dead bytes.
10355 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10356 typos and whitespace errors.
10357 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10358 * config/s390/s390.c (s390_emit_prologue): Likewise.
10360 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10363 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10364 with a symbol for LRA.
10366 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10368 PR rtl-optimization/82338
10369 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10371 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10373 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10374 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10375 (qsort_chk): Declare.
10376 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10377 (qsort_chk): New function.
10379 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10381 PR tree-optimization/82337
10382 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10383 phi definition if the PHI result appears in an abnormal PHI.
10384 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10385 basis appears in an abnormal PHI.
10387 2017-09-29 Richard Biener <rguenther@suse.de>
10389 * graphite-isl-ast-to-gimple.c
10390 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10391 (binary_op_to_tree): Use it.
10392 (get_rename_from_scev): Likewise.
10393 (copy_loop_phi_nodes): Likewise.
10394 (copy_bb_and_scalar_dependences): Likewise.
10395 (translate_pending_phi_nodes): Likewise.
10397 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10400 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10401 for movabsq $(i32 << shift), r64.
10403 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10405 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10406 index when encoding %esp as %rsp to avoid 0x67 prefix.
10408 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10410 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10411 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10413 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10415 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10416 Extensions with more than 16 double VFP registers.
10417 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10418 to_clear_mask and all code related to it. Replace the remaining
10419 entry by a sbitmap and adapt code accordingly.
10421 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10423 * brig-builtins.def: Change pure attributes to const.
10425 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10427 * config.gcc (default_gnu_indirect_function): Default to yes for
10428 sparc*-*-linux* with glibc.
10430 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10432 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10433 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10434 when creating .init_array and .fini_array sections with priority
10437 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10440 * config/aarch64/aarch64.c
10441 (aarch64_builtin_support_vector_misalignment): Always return false
10442 when misalignment is unknown.
10444 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10446 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10447 this function to return false if the definition used by the swap
10448 instruction is artificial, or if the memory address from which the
10449 constant value is loaded is not represented by a base address held
10450 in a register or if the base address register is a frame or stack
10451 pointer. Additionally, return false if the base address of the
10452 loaded constant is a SYMBOL_REF but is not considered to be a
10454 (replace_swapped_load_constant): New function.
10455 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10456 loaded constant vector with a load of a swapped constant vector.
10458 2017-09-27 Carl Love <cel@us.ibm.com>
10460 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10461 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10462 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10465 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10467 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10468 first, always call autopref_rank_data otherwise.
10470 2017-09-27 Richard Biener <rguenther@suse.de>
10472 * graphite-scop-detection.c (find_scop_parameters): Move
10473 loop bound handling ...
10474 (gather_bbs::before_dom_children): ... here, avoiding the need
10475 to build scop_info->loop_nest.
10476 (record_loop_in_sese): Remove.
10477 * sese.h (sese_info_t::loop_nest): Remove.
10478 * sese.c (new_sese_info): Do not allocate loop_nest.
10479 (free_sese_info): Do not free loop_nest.
10481 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10484 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10485 lhs from calls if the lhs has addressable type.
10487 2017-09-27 Richard Biener <rguenther@suse.de>
10489 * graphite.h (scop::max_alias_set): New member.
10490 * graphite-scop-detection.c: Remove references to non-existing
10491 --param in comments.
10492 (build_alias_sets): Record the maximum alias set used for drs.
10493 (build_scops): Support zero as unlimited for
10494 --param graphite-max-arrays-per-scop.
10495 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10496 and inline into ...
10497 (build_poly_sr_1): ... here. Compute alias set based on the
10498 maximum alias set used for drs rather than
10499 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10501 2017-09-27 Richard Biener <rguenther@suse.de>
10503 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10504 --param loop-block-tile-size=0 to disable tiling.
10506 2017-09-27 Richard Biener <rguenther@suse.de>
10508 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10509 (graphite-max-nb-scop-params): Document special value zero.
10510 * domwalk.h (dom_walker::STOP): New symbolical constant.
10511 (dom_walker::dom_walker): Add optional parameter for bb to
10513 (dom_walker::~dom_walker): Declare.
10514 (dom_walker::before_dom_children): Document STOP return value.
10515 (dom_walker::m_user_bb_to_rpo): New member.
10516 (dom_walker::m_bb_to_rpo): Likewise.
10517 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10518 mapping here if not provided by the user.
10519 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10520 provided by the user.
10521 (dom_walker::STOP): Define.
10522 (dom_walker::walk): Do not compute bb to RPO mapping here.
10523 Support STOP return value from before_dom_children to stop
10525 * graphite-optimize-isl.c (optimize_isl): If the schedule
10526 is the same still generate code if -fgraphite-identity
10527 or -floop-parallelize-all are given.
10528 * graphite-scop-detection.c: Include cfganal.h.
10529 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10531 (gather_bbs::before_dom_children): Return STOP for BBs
10533 (build_scops): Compute bb to RPO mapping and pass it to
10534 the domwalk. Treat --param graphite-max-nb-scop-params=0
10535 as not limiting the number of params.
10536 * graphite.c (graphite_initialize): Remove limit on the
10537 number of basic-blocks in a function.
10538 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10539 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10540 default value of 10.
10542 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10544 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10545 Adjust code to eliminate needing to do the shift right 32-bits
10546 operation after XSCVDPSPN.
10548 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10550 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10551 ((X / Y) != 0 -> X >= Y): Likewise.
10553 2017-09-26 Carl Love <cel@us.ibm.com>
10555 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10556 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10557 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10558 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10559 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10560 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10561 definitions and overloading.
10562 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10563 statement for P9V_BUILTIN_XST_LEN_R.
10564 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10565 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10566 define_expand and define_insn for the instructions and builtins.
10567 * doc/extend.texi: Update the built-in documentation file for the new
10568 built-in functions.
10569 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10570 define_insn for the instructions
10572 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10575 * gcc/config/netbsd-protos.h: New file.
10576 * gcc/config/netbsd.c: New file.
10577 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10578 * gcc/config/t-netbsd: New file.
10579 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10580 (tmake_file) Add t-netbsd.
10581 (extra_objs) Add netbsd.o.
10583 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10587 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10589 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10591 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10592 sign extension from a vector register to a GPR by doing a 32-bit
10593 direct move and then an EXTSW.
10594 (extendsi<mode>2 splitter): Likewise.
10595 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10596 right or vector extract after doing XSCVDPSPN. Use
10597 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10599 (movdi_from_sf_zero_ext): Likewise.
10600 (reload_gpr_from_vsxsf): Likewise.
10601 (p8_mfvsrd_4_disf): Delete, no longer used.
10602 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10603 then needing to move the SFmode to a GPR to use the XSCVDPSP
10604 instruction instead of FRSP and XSCVDPSPN.
10605 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10606 it is adjacent to the other XSCVSPDP insns.
10607 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10608 SFmode to be in traditional Altivec registers.
10609 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10610 (vsx_xscvspdpn): Likewise.
10611 (vsx_xscvspdpn_scalar): Likewise.
10613 2017-09-26 Martin Jambor <mjambor@suse.cz>
10615 * tree-sra.c (compare_access_positions): Put integral types first,
10616 stabilize sorting of integral types, remove conditions putting
10617 non-full-precision integers last.
10618 (sort_and_splice_var_accesses): Disable scalarization if a
10619 non-integert would be represented by a non-full-precision integer.
10621 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10623 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10624 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10625 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10626 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10627 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10628 conditionals inside the function instead of around it. Call
10629 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10630 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10632 2017-09-26 Richard Biener <rguenther@suse.de>
10634 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10636 (scop_detection::build_scop_breadth): ... this. Removed.
10637 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10638 (scop_detection::harmful_stmt_in_bb): Likewise.
10639 (scop_detection::graphite_can_represent_stmt): Likewise.
10640 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10641 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10642 (scop_detection::can_represent_loop_1): ... this. Removed.
10643 (scop_detection::harmful_loop_in_region): Simplify after inlining
10644 the above and remove more quadraticness.
10645 (build_scops): Adjust.
10646 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10649 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10652 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10653 REGNO (base) == SP_REG if base is a REG.
10655 PR middle-end/35691
10656 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10657 if it is different SSA_NAME.
10658 (optimize_range_tests_cmp_bitwise): New function.
10659 (optimize_range_tests): Call it.
10661 2017-09-26 Richard Biener <rguenther@suse.de>
10663 PR tree-optimization/82321
10664 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10665 for the def being inside the loop.
10667 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10669 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10671 * config/s390/s390-builtins.def: Fix constraint on op4.
10673 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10675 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10676 independent expanders.
10677 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10678 ("vec_ordered", "vec_unordered"): New expanders.
10680 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10682 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10685 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10687 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10688 vec_unpacks_lo_v16qi.
10689 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10691 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10693 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10694 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10695 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10697 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10699 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10701 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10703 ("*vec_slb<mode>"): New insn pattern.
10704 ("vec_shr_<mode>"): New expander.
10705 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10706 and force the shift count operand to V16QImode.
10707 ("vec_srb<mode>"): Set shift count mode to V16QI.
10709 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10711 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10712 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10713 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10715 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10718 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10720 2017-09-26 Richard Biener <rguenther@suse.de>
10722 PR tree-optimization/82320
10723 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10726 2017-09-25 Jeff Law <law@redhat.com>
10728 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10729 prototype for new argument.
10730 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10731 mostly extracted from rs6000_emit_allocate_stack.
10732 (rs6000_emit_probe_stack_range_stack_clash): New function.
10733 (rs6000_emit_allocate_stack): Call
10734 rs6000_emit_probe_stack_range_stack_clash as needed.
10735 (rs6000_emit_probe_stack_range): Add additional argument
10736 to call to gen_probe_stack_range{si,di}.
10737 (output_probe_stack_range): New.
10738 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10739 (output_probe_stack_range_stack_clash): New.
10740 (rs6000_emit_prologue): Emit notes into dump file as requested.
10741 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10742 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10743 Add additional operand and pass it to output_probe_stack_range.
10745 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10747 PR tree-optimization/82163
10748 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10749 (checking_verify_loop_closed_ssa): New parameter.
10750 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10751 (check_loop_closed_ssa_stmt): Delete.
10752 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10753 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10754 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10757 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10759 * brig-builtins.def: Treat HSAIL barrier builtins as
10760 setjmp/longjump style functions.
10762 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10764 * target.def (constant_alignment): New hook.
10765 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10766 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10767 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10768 * doc/tm.texi: Regenerate.
10769 * targhooks.h (default_constant_alignment): Declare.
10770 (constant_alignment_word_strings): Likewise.
10771 * targhooks.c (default_constant_alignment): New function.
10772 (constant_alignment_word_strings): Likewise.
10773 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10774 instead of CONSTANT_ALIGNMENT.
10775 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10776 (force_const_mem): Likewise.
10777 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10778 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10779 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10780 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10781 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10783 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10784 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10785 constant_alignment_word_strings.
10786 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10787 (CONSTANT_ALIGNMENT): Likewise.
10788 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10789 (arm_constant_alignment): New function.
10790 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10791 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10792 constant_alignment_word_strings.
10793 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10794 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10795 constant_alignment_word_strings.
10796 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10797 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10798 (cris_constant_alignment): New function.
10799 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10800 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10801 (epiphany_constant_alignment): New function.
10802 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10803 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10804 constant_alignment_word_strings.
10805 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10806 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10807 constant_alignment_word_strings.
10808 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10809 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10810 constant_alignment_word_strings.
10811 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10812 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10813 * config/i386/i386.c (ix86_constant_alignment): Make static.
10814 Use the same interface as the target hook.
10815 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10816 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10817 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10818 constant_alignment_word_strings.
10819 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10820 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10821 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10822 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10823 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10824 constant_alignment_word_strings.
10825 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10826 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10827 constant_alignment_word_strings.
10828 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10829 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10830 constant_alignment_word_strings.
10831 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10832 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10834 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10835 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10836 * config/mips/mips.c (mips_constant_alignment): New function.
10837 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10838 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10839 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10840 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10841 (mmix_constant_alignment): Make static. Use the same interface
10842 as the target hook.
10843 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10844 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10845 constant_alignment_word_strings.
10846 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10847 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10848 constant_alignment_word_strings.
10849 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10850 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10851 constant_alignment_word_strings.
10852 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10853 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10854 (rs6000_constant_alignment): New function.
10855 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10856 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10857 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10858 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10859 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10860 (rs6000_constant_alignment): New function.
10861 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10862 * config/s390/s390.c (s390_constant_alignment): New function.
10863 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10864 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10865 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10866 constant_alignment_word_strings.
10867 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10868 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10869 (sparc_constant_alignment): New function.
10870 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10871 * config/spu/spu.c (spu_constant_alignment): New function.
10872 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10873 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10874 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10875 constant_alignment_word_strings.
10876 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10877 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10878 constant_alignment_word_strings.
10879 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10880 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10881 constant_alignment_word_strings.
10882 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10883 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10884 (visium_constant_alignment): New function.
10885 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10886 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10887 (xtensa_constant_alignment): New function.
10888 * system.h (CONSTANT_ALIGNMENT): Poison.
10890 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10892 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10893 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10894 (rs6000_builtin_valid_without_lhs): New helper function.
10895 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10896 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10898 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10900 * target.h (vec_perm_indices): Use unsigned short rather than
10902 (auto_vec_perm_indices): Likewise.
10903 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10904 Use unsigned int rather than unsigned char.
10905 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10907 2017-09-25 Richard Biener <rguenther@suse.de>
10909 * cfgloop.h (sort_sibling_loops): Declare.
10910 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10911 (sort_sibling_loops): New function sorting the sibling loop list
10913 * graphite.c (graphite_transform_loops): Sort sibling loops.
10915 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10917 * target.def (vec_perm_const_ok): Change sel parameter to
10919 * optabs-query.c (can_vec_perm_p): Update accordingly.
10920 * doc/tm.texi: Regenerate.
10921 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10922 auto_vec_perm_indices and remove separate nelt field.
10923 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10924 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10925 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10926 (aarch64_expand_vec_perm_const): Update accordingly.
10927 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10928 to vec_perm_indices.
10929 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10930 auto_vec_perm_indices and remove separate nelt field.
10931 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10932 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10933 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10935 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10936 to vec_perm_indices.
10937 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10938 sel to vec_perm_indices.
10939 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10940 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10941 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10943 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10946 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10949 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10950 on the FUNCTION_DECL function context if it has a DIE that is a
10953 2017-09-25 Richard Biener <rguenther@suse.de>
10955 PR tree-optimization/82285
10956 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10959 2017-09-25 Tom de Vries <tom@codesourcery.com>
10963 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10966 2017-09-25 Richard Biener <rguenther@suse.de>
10968 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10969 ISL errors other than isl_error_quota happen. Dump if the
10970 schedule is the same.
10971 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10972 errors instead of aborting inside ISL.
10974 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10977 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10978 of libgcc_eh for m64.
10979 * config/i386/darwin64.h: Likewise.
10981 2017-09-25 Richard Biener <rguenther@suse.de>
10983 PR middle-end/82144
10984 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10985 attribute for incomplete types nor twice for complete ones.
10987 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10990 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10991 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10994 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10997 * config/i386/i386.opt (mprefer-avx256): Use
10998 ix86_target_flags variable.
10999 * config/i386/i386.c (ix86_target_string): Move
11000 -mprefer-avx256 to flag2_opts.
11002 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11004 PR middle-end/35691
11005 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11006 and x != -1 | y != -1 into (x & y) != -1.
11008 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11010 * config.gcc: Add new case statement to set
11011 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11012 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11013 s390x-*-linux* case statements. Added aarch64 to the list of
11014 supported architectures.
11016 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11018 PR tree-optimization/82289
11019 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11020 STMT_VINFO_RELEVANT_P.
11022 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11023 Alan Hayward <alan.hayward@arm.com>
11024 David Sherwood <david.sherwood@arm.com>
11026 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11027 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11028 (extract_range_from_binary_expr_1): Don't call
11029 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11031 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11032 Alan Hayward <alan.hayward@arm.com>
11033 David Sherwood <david.sherwood@arm.com>
11035 * target.def (preferred_vector_alignment): New hook.
11036 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11038 * doc/tm.texi: Regenerate.
11039 * targhooks.h (default_preferred_vector_alignment): Declare.
11040 * targhooks.c (default_preferred_vector_alignment): New function.
11041 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11043 (DR_TARGET_ALIGNMENT): New macro.
11044 (aligned_access_p): Update commentary.
11045 (vect_known_alignment_in_bytes): New function.
11046 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11048 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11049 Calculate the misalignment based on the target alignment rather than
11051 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11052 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11053 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11054 the target alignment, rather than masking the element misalignment
11055 with the number of elements in a vector. Also use the target
11056 alignment when calculating the maximum number of peels.
11057 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11058 instead of TYPE_ALIGN_UNIT.
11059 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11060 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11061 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11062 (vect_create_data_ref_ptr): Likewise.
11063 (vect_setup_realignment): Realign by ANDing with
11064 -DR_TARGET_MISALIGNMENT.
11065 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11066 the number of peels based on DR_TARGET_ALIGNMENT.
11067 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11068 with the guaranteed alignment boundary when deciding whether
11070 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11071 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11072 (ensure_base_align): Remove stmt_info parameter. Get the
11073 target base alignment from DR_TARGET_ALIGNMENT.
11074 (vectorizable_store): Update call accordingly. Interpret
11075 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11077 (vectorizable_load): Likewise.
11079 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11080 Alan Hayward <alan.hayward@arm.com>
11081 David Sherwood <david.sherwood@arm.com>
11083 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11084 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11085 (vect_enhance_data_refs_alignment): Likewise.
11087 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11089 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11090 error. Only quit immediately if parsing is complete.
11091 (BEGIN): Initialize fatal_err and parse_done.
11092 (begin fpu, end fpu): Check number of arguments.
11093 (begin arch, end arch): Likewise.
11094 (begin cpu, end cpu): Likewise.
11095 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11096 (optalias): Likewise.
11098 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11100 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11101 * config/arm/arm-isa.h: Delete. Move definitions to ...
11102 * arm-cpus.in: ... here. Use new feature and fgroup values.
11103 * config/arm/arm.c (arm_option_override): Use lower case for feature
11105 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11106 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11107 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11109 (print_isa_bits_for): New function.
11110 (gen_isa): New function.
11111 (gen_comm_data): Use print_isa_bits_for.
11112 (define feature): New keyword.
11113 (define fgroup): New keyword.
11114 * config/arm/t-arm (TM_H): Remove.
11115 (GTM_H): Add arm-isa.h.
11116 (arm-isa.h): Add rule to generate file.
11117 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11118 case for feature bit names.
11120 2017-09-22 Richard Biener <rguenther@suse.de>
11122 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11124 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11125 print of no dependency loops ...
11126 * graphite.c (graphite_transform_loops): ... here.
11127 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11129 (same_close_phi_node, remove_duplicate_close_phi,
11130 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11131 (canonicalize_loop_closed_ssa): ... here and simplify.
11132 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11133 (optimize_isl): Use dump_printf_loc to tell when we stopped
11134 optimizing because of an ISL timeout.
11136 2017-09-22 Richard Biener <rguenther@suse.de>
11138 PR tree-optimization/82291
11139 * tree-if-conv.c (predicate_mem_writes): Make sure to
11140 remove writes in blocks predicated with false.
11142 2017-09-22 Richard Biener <rguenther@suse.de>
11144 * sese.c: Include cfganal.h.
11145 (if_region_set_false_region): Remove.
11146 (create_if_region_on_edge): Likewise.
11147 (move_sese_in_condition): Re-implement without destroying
11150 2017-09-22 Richard Biener <rguenther@suse.de>
11152 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11153 Verify both BBs contain loop PHI nodes before dispatching to
11154 copy_loop_phi_args.
11155 (graphite_regenerate_ast_isl): Do not recompute dominators,
11156 do not verify three times. Restructure for clarity.
11157 * graphite-scop-detection.c (same_close_phi_node,
11158 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11159 defined_in_loop_p, canonicalize_loop_closed_ssa,
11160 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11161 checking and SSA rewrite, move to ...
11162 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11163 (graphite_initialize): Do not pass in ctx, do not reset the
11164 SCEV cache, compute only dominators.
11165 (graphite_transform_loops): Allocate ISL ctx after
11166 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11167 Maintain post-dominators only around build_scops.
11168 * sese.c (if_region_set_false_region): Make static. Free
11169 and recompute dominators.
11170 (move_sese_in_condition): Assert we don't get called with
11171 post-dominators computed.
11172 * sese.h (if_region_set_false_region): Remove.
11174 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11176 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11177 mode attribute for TARGET_AVX512VL.
11179 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11181 * config/i386/i386.opt (mprefer-avx256): New option.
11182 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11184 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11185 for TARGET_PREFER_AVX256.
11186 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11188 2017-09-21 Jeff Law <law@redhat.com>
11190 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11191 Fix dump output if the only stack space is for pushed registers.
11193 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11195 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11198 2017-09-21 Martin Sebor <msebor@redhat.com>
11201 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11202 code (in C++) or code that triggers warnings.
11204 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11206 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11208 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11210 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11211 * sched-rgn.c: Ditto.
11212 * sel-sched-ir.c: Ditto.
11214 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11216 * toplev.h (set_random_seed): Adjust return type.
11217 * toplev.c (init_local_tick): Move eager initialization of random_seed
11218 to get_random_seed. Adjust comment.
11219 (init_random_seed): Inline to get_random_seed, delete.
11220 (get_random_seed): Initialize random_seed lazily.
11221 (set_random_seed): Do not return previous value.
11222 (print_switch_value): Do not call get_random_seed.
11224 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11226 * cgraph.c (delete_function_version): New, broken out from...
11227 (cgraph_node::delete_function_version): ...here. Rename to
11228 cgraph_node::delete_function_version_by_decl. Update all uses.
11229 (cgraph_node::remove): Call delete_function_version.
11231 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11234 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11235 VAR_DECLs to which addressable non-volatile parameters are mapped
11236 and for id->retvar after the return value assignment. Clear
11237 id->retval and id->retbnd after inlining.
11239 2017-09-21 Richard Biener <rguenther@suse.de>
11241 PR tree-optimization/82276
11242 PR tree-optimization/82244
11243 * tree-vrp.c (build_assert_expr_for): Set
11244 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11246 (remove_range_assertions): Revert earlier change.
11248 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11251 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11253 2017-09-21 Richard Biener <rguenther@suse.de>
11255 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11256 Restore valid IL after code generation errors.
11257 * graphite.c (graphite_transform_loops): Diagnose code
11258 generation issues as MSG_MISSED_OPTIMIZATION and continue
11259 with processing SCOPs.
11261 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11262 Alan Hayward <alan.hayward@arm.com>
11263 David Sherwood <david.sherwood@arm.com>
11265 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11266 rather than choosing between plus_constant and gen_rtx_<CODE>.
11267 * expr.c (emit_push_insn): Likewise.
11268 (expand_expr_real_2): Likewise.
11270 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11271 Alan Hayward <alan.hayward@arm.com>
11272 David Sherwood <david.sherwood@arm.com>
11274 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11276 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11277 Alan Hayward <alan.hayward@arm.com>
11278 David Sherwood <david.sherwood@arm.com>
11280 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11281 calling tree_to_uhwi.
11283 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11284 Alan Hayward <alan.hayward@arm.com>
11285 David Sherwood <david.sherwood@arm.com>
11287 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11288 INTEGER_CST rather than a negative test for ADDR_EXPR.
11290 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11291 Alan Hayward <alan.hayward@arm.com>
11292 David Sherwood <david.sherwood@arm.com>
11294 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11295 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11297 2017-09-21 Richard Biener <rguenther@suse.de>
11299 PR tree-optimization/71351
11300 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11301 graphite_create_new_loop_guard): Remove, fold remaining parts
11303 (translate_isl_ast_node_for): ... here and simplify.
11305 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11308 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11309 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11310 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11311 alternative always use QI mode, for -Os imov (=R,R) alternative
11312 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11315 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11316 Jeff Law <law@redhat.com>
11318 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11319 (allocate_stack_space): New function, partially extracted from
11320 s390_emit_prologue.
11321 (s390_emit_prologue): Track offset to most recent stack probe.
11322 Code to allocate space moved into allocate_stack_space.
11323 Dump actions when no stack is allocated.
11324 (s390_prologue_plus_offset): New function.
11325 (s390_emit_stack_probe): Likewise.
11327 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11329 * common.opt (Wa, Wl, Wp, g, gz=): Add
11331 (gno-column-info): Remove.
11332 (gcolumn-info): Drop RejectNegative.
11333 (gno-): New prefix.
11334 (gno-record-gcc-switches): Remove.
11335 (grecord-gcc-switches): Drop RejectNegative.
11336 (gno-split-dwarf): Remove.
11337 (gsplit-dwarf): Drop RejectNegative.
11338 (gno-strict-dwarf): Remove.
11339 (gstrict-dwarf): Drop RejectNegative.
11340 * config/darwin.opt (gfull, gused): Add RejectNegative.
11341 * dwarf2out.c (gen_producer_string): Drop
11342 gno-record-gcc-switches handler.
11343 * optc-gen.awk: Add g to prefixes with negative forms.
11344 * opts-common.c (remapping_prefix_p): New.
11345 (find_opt): Check it.
11346 (generate_canonical_option): Test g prefix.
11347 (option_map): Add -gno- mapping.
11348 (add_misspelling_candidates): Check remapping_prefix_p.
11350 2017-09-20 Jeff Law <law@redhat.com>
11352 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11353 thinko in stack clash protection support.
11355 * explow.c (compute_stack_clash_protection_loop_data): Use
11356 CONST_INT_P instead of explicit test. Verify object is a
11357 CONST_INT_P before looking at INTVAL.
11358 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11359 instead of explicit test.
11361 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11364 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11365 address instead of to r1 and r11.
11367 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11369 * config.gcc: Support "knm".
11370 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11371 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11373 * config/i386/i386.c (m_KNM): Define.
11374 (processor_target_table): Add "knm".
11376 (ix86_option_override_internal): Add "knm".
11377 (ix86_issue_rate): Add PROCESSOR_KNM.
11378 (ix86_adjust_cost): Ditto.
11379 (ia32_multipass_dfa_lookahead): Ditto.
11380 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11381 (fold_builtin_cpu): Add M_INTEL_KNM.
11382 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11383 (processor_type): Add PROCESSOR_KNM.
11384 * config/i386/x86-tune.def: Add m_KNM.
11385 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11387 2017-09-20 Richard Biener <rguenther@suse.de>
11389 PR tree-optimization/80213
11390 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11391 are allowed in empty BBs as well.
11392 (canonicalize_loop_closed_ssa): Also look for other complex
11394 (scop_detection::get_sese): Include the loop-closed PHI block
11396 (scop_detection::merge_sese): Remove code adding extra blocks.
11397 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11398 (build_scops): Assert the final returned scop is invalid.
11400 2017-09-20 Richard Biener <rguenther@suse.de>
11402 PR tree-optimization/82264
11403 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11405 (vn_phi_lookup): Likewise.
11406 (vn_phi_insert): Likewise.
11408 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11410 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11411 that fits into uhwi or shwi, add DW_AT_const_value regardless
11412 of early_dwarf without going through RTL, using add_AT_unsigned
11415 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11416 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11417 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11418 (DEBUG_MACRO_SECTION): Likewise.
11419 (DEBUG_LTO_MACRO_SECTION): Likewise.
11420 (DEBUG_STR_DWO_SECTION): Likewise.
11421 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11422 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11423 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11424 (DEBUG_LTO_LINE_STR_SECTION): Define.
11425 (init_sections_and_labels): Initialize debug_line_str_section
11426 variable. Initialize debug_loc_section for -gdwarf-5 to
11427 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11429 2017-09-20 Richard Biener <rguenther@suse.de>
11431 * graphite-sese-to-poly.c (extract_affine): Properly handle
11432 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11434 2017-09-20 Richard Biener <rguenther@suse.de>
11436 PR tree-optimization/81373
11437 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11438 Force SESE live-out defs to be handled even if they are
11441 2017-09-19 Jeff Law <law@redhat.com>
11443 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11444 nothing for stack adjustments with REG_STACK_CHECK.
11445 * sched-deps.c (parse_add_or_inc): Reject insns with
11446 REG_STACK_CHECK from dependency breaking.
11447 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11448 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11449 * reg-notes.def (STACK_CHECK): New note.
11451 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11452 (ix86_expand_prologue): Dump stack clash info as needed.
11453 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11455 * function.c (dump_stack_clash_frame_info): New function.
11456 * function.h (dump_stack_clash_frame_info): Prototype.
11457 (enum stack_clash_probes): New enum.
11459 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11460 flag_stack_clash_protection.
11461 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11462 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11463 (arm_frame_pointer_required): Likewise.
11464 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11465 (ia64_expand_prologue): Likewise.
11466 * config/mips/mips.c (mips_expand_prologue): Likewise.
11467 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11468 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11469 (sparc_flat_expand_prologue): Likewise.
11470 * config/spu/spu.c (spu_expand_prologue): Likewise.
11472 * explow.c: Include "params.h".
11473 (anti_adjust_stack_and_probe_stack_clash): New function.
11474 (get_stack_check_protect): Likewise.
11475 (compute_stack_clash_protection_loop_data): Likewise.
11476 (emit_stack_clash_protection_loop_start): Likewise.
11477 (emit_stack_clash_protection_loop_end): Likewise.
11478 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11479 Use anti_adjust_stack_and_probe_stack_clash.
11480 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11481 (emit_stack_clash_protection_loop_start): Likewise.
11482 (emit_stack_clash_protection_loop_end): Likewise.
11483 * rtl.h (get_stack_check_protect): Prototype.
11484 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11485 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11486 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11488 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11490 * doc/tm.texi: Rebuilt.
11491 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11492 get_stack_check_protect.
11493 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11494 * config/arm/arm.c (arm_expand_prologue): Likewise.
11495 (arm_frame_pointer_required): Likewise.
11496 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11497 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11498 * config/mips/mips.c (mips_expand_prologue): Likewise.
11499 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11500 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11501 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11502 (sparc_flat_expand_prologue): Likewise.
11504 * common.opt (-fstack-clash-protection): New option.
11505 * flag-types.h (enum stack_check_type): Note difference between
11506 -fstack-check= and -fstack-clash-protection.
11507 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11508 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11509 * toplev.c (process_options): Issue warnings/errors for cases
11510 not handled with -fstack-clash-protection.
11511 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11512 (-fstack-check): Note additional problem with -fstack-check=generic.
11513 Note that -fstack-check is primarily for Ada and refer users
11514 to -fstack-clash-protection for stack-clash-protection.
11515 Document new params for stack clash protection.
11517 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11519 * config/i386/i386.c (ix86_split_long_move): Do not handle
11520 address used for LEA in a special way.
11522 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11524 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11526 2017-09-19 Martin Sebor <msebor@redhat.com>
11529 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11530 of incompatible types.
11532 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11534 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11535 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11536 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11537 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11539 2017-09-19 Richard Biener <rguenther@suse.de>
11541 PR tree-optimization/82244
11542 * tree-vrp.c (remove_range_assertions): Do not propagate
11543 a constant to abnormals but replace the assert with a copy.
11545 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11547 PR rtl-optimization/57878
11548 PR rtl-optimization/68988
11549 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11550 avoidance test involving non_reload_pseudos. Move frequency test
11551 below the general fragmentation avoidance test.
11553 2017-09-19 Richard Biener <rguenther@suse.de>
11555 PR tree-optimization/69728
11556 * graphite-sese-to-poly.c (schedule_error): New global.
11557 (add_loop_schedule): Handle empty domain by failing the
11559 (build_original_schedule): Handle schedule_error.
11561 2017-09-19 Richard Biener <rguenther@suse.de>
11563 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11564 Do not iterate to sibling loops but only to siblings of inner
11567 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11570 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11571 that the registers are different.
11573 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11575 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11576 to processor_model and "amdfam17h" to arch_names_table.
11577 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11579 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11582 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11584 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11585 Alan Hayward <alan.hayward@arm.com>
11586 David Sherwood <david.sherwood@arm.com>
11588 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11590 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11592 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11593 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11594 vect_schedule_slp_instance.
11595 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11596 Update call to vect_slp_analyze_node_operations. Simplify return
11598 (vect_slp_analyze_bb_1): Update call accordingly.
11599 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11600 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11601 (vect_schedule_slp): Update call accordingly.
11603 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11604 Alan Hayward <alan.hayward@arm.com>
11605 David Sherwood <david.sherwood@arm.com>
11607 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11608 with types that aren't in fact scalar.
11610 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11612 * tree-vect-slp.c (vect_record_max_nunits): New function,
11614 (vect_build_slp_tree_1): ...here.
11615 (vect_build_slp_tree_2): Call it for phis too.
11617 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11619 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11620 to vect_get_vec_def_for_operand when getting the mask operand.
11622 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11623 Alan Hayward <alan.hayward@arm.com>
11624 David Sherwood <david.sherwood@arm.com>
11626 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11629 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11630 Alan Hayward <alan.hayward@arm.com>
11631 David Sherwood <david.sherwood@arm.com>
11633 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11634 calculation for vector booleans.
11636 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11637 Alan Hayward <alan.hayward@arm.com>
11638 David Sherwood <david.sherwood@arm.com>
11640 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11642 (vect_transform_stmt): ...here.
11643 (vect_analyze_stmt): Use it instead of calling
11644 vectorizable_live_operation directly.
11646 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11648 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11649 non-SIMT targets in acc vector loops.
11651 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11653 * configure.ac: Add arc and check if assembler supports gdwarf2.
11654 * configure: Regenerate.
11656 2017-09-18 Richard Biener <rguenther@suse.de>
11658 PR tree-optimization/82220
11659 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11660 epilogue niters from the min_profitable_iters compute.
11662 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11665 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11666 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11667 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11669 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11672 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11673 switching to a new text section.
11675 2017-09-18 Richard Biener <rguenther@suse.de>
11677 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11679 (build_alias_set): Reject aliases with no access function.
11681 2017-09-18 Richard Biener <rguenther@suse.de>
11683 PR tree-optimization/79622
11684 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11686 (build_cross_bb_scalars_use): Likewise.
11688 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11690 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11692 2017-09-18 Alan Modra <amodra@gmail.com>
11695 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11696 stack_pointer_rtx for count 0. Update comments. Break up
11697 large rtl expression.
11699 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11701 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11702 Increase to 20 bytes.
11703 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11704 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11705 or avx version of the stub.
11707 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11710 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11711 compute the minimum stack alignment. Also update preferred stack
11712 boundary for leaf functions.
11714 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11716 PR tree-optimization/82228
11717 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11720 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11722 * common/config/nds32/nds32-common.c
11723 (nds32_option_optimization_table): Refine formatting.
11724 (nds32_option_optimization_table): Use -fsched-pressure and
11725 -fomit-frame-pointer for specific optimization level.
11727 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11729 * config/nds32/nds32.c: Refine formatting and comments.
11730 * config/nds32/nds32.h: Likewise.
11731 * config/nds32/nds32.md: Likewise.
11732 * config/nds32/nds32-cost.c: Likewise.
11733 * config/nds32/nds32-isr.c: Likewise.
11734 * config/nds32/nds32-md-auxiliary.c: Likewise.
11735 * config/nds32/nds32-multiple.md: Likewise.
11736 * config/nds32/nds32-predicates.c: Likewise.
11738 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11739 Jakub Jelinek <jakub@redhat.com>
11741 Add support for -std=c++2a.
11742 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11744 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11746 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11749 * doc/extend.texi (Common Function Attributes): Add
11750 references to ARM, AArch64, and S/390 specific attributes.
11751 (Function Specific Option Pragmas): Add AArch64 and S/390
11752 to list of back ends that support the target pragma.
11754 2017-09-15 Nathan Sidwell <nathan@acm.org>
11756 * doc/standards.texi: Fix C++17 description. Update URLs for
11759 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11761 * common.opt (Wcast-align=strict): New warning option.
11762 * doc/invoke.texi: Document -Wcast-align=strict.
11764 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11766 * cgraph.h (cgraph_thunk_info): Add comments.
11767 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11768 assert for VIRTUAL_* arguments stricter.
11770 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11772 PR tree-optimization/71026
11773 * match.pd: Move RDIV patterns from fold-const.c
11774 * fold-const.c (distribute_real_division): Removed.
11775 (fold_binary_loc): Remove calls to distribute_real_divison.
11777 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11779 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11780 c++1z and gnu++1z as deprecated. Change other references to
11781 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11782 Change -Wc++1z-compat to -Wc++17-compat.
11783 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11784 * dwarf2out.c (highest_c_language): Handle C++17.
11785 (gen_compile_unit_die): Likewise.
11787 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11789 PR rtl-optimization/82192
11790 * combine.c (make_extraction): Don't look through non-paradoxical
11791 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11794 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11795 Alan Hayward <alan.hayward@arm.com>
11796 David Sherwood <david.sherwood@arm.com>
11798 * target.def (function_arg_offset): New hook.
11799 * targhooks.h (default_function_arg_offset): Declare.
11800 * targhooks.c (default_function_arg_offset): New function.
11801 * function.c (locate_and_pad_parm): Use
11802 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11803 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11804 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11805 * doc/tm.texi: Regenerate.
11806 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11807 * config/spu/spu.c (spu_function_arg_offset): New function.
11808 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11809 * system.h (FUNCTION_ARG_OFFSET): Poison.
11811 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11812 Alan Hayard <alan.hayward@arm.com>
11813 David Sherwood <david.sherwood@arm.com>
11815 * target.def (truly_noop_truncation): New hook.
11816 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11817 than TRULY_NOOP_TRUNCATION.
11818 * hooks.h (hook_bool_uint_uint_true): Declare.
11819 * hooks.c (hook_bool_uint_uint_true): New function.
11820 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11821 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11822 * doc/tm.texi: Regenerate.
11823 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11824 rather than TRULY_NOOP_TRUNCATION in comments.
11825 (simplify_comparison): Likewise.
11826 (record_truncated_value): Likewise.
11827 * expmed.c (extract_bit_field_1): Likewise.
11828 (extract_split_bit_field): Likewise.
11829 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11830 instead of TRULY_NOOP_TRUNCATION.
11831 * function.c (assign_parm_setup_block): Likewise.
11832 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11833 * rtlhooks.c: Include target.h.
11834 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11835 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11836 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11837 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11838 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11839 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11840 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11841 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11842 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11843 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11844 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11845 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11846 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11847 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11848 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11849 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11850 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11851 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11852 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11853 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11854 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11855 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11856 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11857 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11858 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11859 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11860 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11861 rather than TRULY_NOOP_TRUNCATION in comments.
11862 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11863 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11864 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11865 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11866 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11867 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11868 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11869 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11870 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11871 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11872 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11873 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11874 rather than TRULY_NOOP_TRUNCATION in comments.
11875 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11876 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11877 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11878 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11879 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11880 TRULY_NOOP_TRUNCATION condition.
11881 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11882 (TRULY_NOOP_TRUNCATION): Delete.
11883 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11884 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11885 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11886 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11887 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11888 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11889 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11890 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11891 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11892 rather than TRULY_NOOP_TRUNCATION in comments.
11893 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11894 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11895 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11896 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11897 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11898 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11900 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11903 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11904 (*cmp_ior): Likewise.
11905 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11906 (*ior_scc_scc_cmp): Likewise.
11907 (*and_scc_scc): Likewise.
11908 (*and_scc_scc_cmp): Likewise.
11910 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11911 Alan Hayard <alan.hayward@arm.com>
11912 David Sherwood <david.sherwood@arm.com>
11914 * target.def (can_change_mode_class): New hook.
11915 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11916 (hard_regno_nregs): Likewise.
11917 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11918 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11919 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11920 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11921 (LOAD_EXTEND_OP): Update accordingly.
11922 * doc/tm.texi: Regenerate.
11923 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11924 CANNOT_CHANGE_MODE_CLASS.
11925 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11926 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11927 * combine.c (simplify_set): Update accordingly.
11928 * emit-rtl.c (validate_subreg): Likewise.
11929 * recog.c (general_operand): Likewise.
11930 * regcprop.c (mode_change_ok): Likewise.
11931 * reload1.c (choose_reload_regs): Likewise.
11932 (inherit_piecemeal_p): Likewise.
11933 * rtlanal.c (simplify_subreg_regno): Likewise.
11934 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11935 instead of CANNOT_CHANGE_MODE_CLASS.
11936 (reload_cse_simplify_operands): Likewise.
11937 * reload.c (push_reload): Use targetm.can_change_mode_class
11938 instead of CANNOT_CHANGE_MODE_CLASS.
11939 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11940 REG_CANNOT_CHANGE_MODE_P.
11941 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11942 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11943 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11944 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11945 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11946 (arm_can_change_mode_class): New function.
11947 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11948 than CANNOT_CHANGE_MODE_CLASS in comments.
11949 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11950 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11951 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11952 (ix86_can_change_mode_class): ...this new function, inverting the
11953 sense of the return value.
11954 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11955 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11956 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11957 (ia64_can_change_mode_class): New function.
11958 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11959 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11960 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11961 (m32c_can_change_mode_class): ...this new function, inverting the
11962 sense of the return value.
11963 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11964 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11965 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11966 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11967 (mips_can_change_mode_class): ...this new function, inverting the
11968 sense of the return value.
11969 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11970 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11971 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11972 (msp430_can_change_mode_class): New function.
11973 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11974 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11975 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11976 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11977 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11978 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11979 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11980 (pa_cannot_change_mode_class): Replace with...
11981 (pa_can_change_mode_class): ...this new function, inverting the
11982 sense of the return value.
11983 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11984 than CANNOT_CHANGE_MODE_CLASS in comments.
11985 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11986 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11987 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11988 (pdp11_cannot_change_mode_class): Replace with...
11989 (pdp11_can_change_mode_class): ...this new function, inverting the
11990 sense of the return value.
11991 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11992 * config/powerpcspe/powerpcspe-protos.h
11993 (rs6000_cannot_change_mode_class_ptr): Delete.
11994 * config/powerpcspe/powerpcspe.c
11995 (rs6000_cannot_change_mode_class_ptr): Delete.
11996 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11997 (rs6000_option_override_internal): Assign to
11998 targetm.can_change_mode_class instead of
11999 rs6000_cannot_change_mode_class_ptr.
12000 (rs6000_cannot_change_mode_class): Replace with...
12001 (rs6000_can_change_mode_class): ...this new function, inverting the
12002 sense of the return value.
12003 (rs6000_debug_cannot_change_mode_class): Replace with...
12004 (rs6000_debug_can_change_mode_class): ...this new function.
12005 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12006 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12007 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12008 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12009 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12011 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12012 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12013 (rs6000_option_override_internal): Assign to
12014 targetm.can_change_mode_class instead of
12015 rs6000_cannot_change_mode_class_ptr.
12016 (rs6000_cannot_change_mode_class): Replace with...
12017 (rs6000_can_change_mode_class): ...this new function, inverting the
12018 sense of the return value.
12019 (rs6000_debug_cannot_change_mode_class): Replace with...
12020 (rs6000_debug_can_change_mode_class): ...this new function.
12021 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12022 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12023 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12024 (s390_can_change_mode_class): ...this new function, inverting the
12025 sense of the return value.
12026 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12027 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12028 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12029 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12030 (sh_cannot_change_mode_class): Replace with...
12031 (sh_can_change_mode_class): ...this new function, inverting the
12032 sense of the return value.
12033 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12034 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12035 (sparc_can_change_mode_class): New function.
12036 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12037 * config/spu/spu.c (spu_can_change_mode_class): New function.
12038 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12039 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12040 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12041 (visium_can_change_mode_class): New function.
12042 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12044 2017-09-15 Richard Biener <rguenther@suse.de>
12046 PR tree-optimization/82217
12047 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12048 but not undefined case.
12050 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12053 * postreload.c (reload_cse_simplify_operands): Skip
12054 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12056 2017-09-15 Richard Biener <rguenther@suse.de>
12058 PR tree-optimization/68823
12059 * graphite-scop-detection.c (build_alias_set): If we have a
12060 possible dependence check whether we can handle them by just
12061 looking at the DRs DR_ACCESS_FNs.
12062 (build_scops): If build_alias_set fails, fail the SCOP.
12064 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12066 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12067 to support float128 built-in functions that require the ISA 3.0
12069 (BU_FLOAT128_3_HW): Likewise.
12070 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12071 built-in functions.
12072 (FMAF128): Likewise.
12074 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12075 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12076 floating point instructions.
12077 (rs6000_invalid_builtin): Likewise.
12078 (rs6000_builtin_mask_names): Likewise.
12079 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12080 (RS6000_BTM_FLOAT128_HW): Likewise.
12081 (RS6000_BTM_COMMON): Likewise.
12082 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12084 * doc/extend.texi (RS/6000 built-in functions): Document the
12085 IEEE 128-bit floating point square root and fused multiply-add
12086 built-in functions.
12088 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12090 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12091 reg (r2) isn't in the set of registers defined in the prologue.
12093 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12094 Alan Hayward <alan.hayward@arm.com>
12095 David Sherwood <david.sherwood@arm.com>
12097 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12098 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12099 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12100 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12101 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12103 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12104 max_vectorization_factor.
12105 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12107 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12108 Alan Hayward <alan.hayward@arm.com>
12109 David Sherwood <david.sherwood@arm.com>
12111 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12112 (vect_worthwhile_without_simd_p): Declare.
12113 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12114 (vectorizable_reduction): Use it.
12115 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12116 (vectorizable_operation): Likewise.
12118 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12119 Alan Hayward <alan.hayward@arm.com>
12120 David Sherwood <david.sherwood@arm.com>
12122 * tree-vectorizer.h (vect_get_num_copies): New function.
12123 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12124 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12125 (vectorizable_induction): Likewise.
12126 (vectorizable_live_operation): Likewise.
12127 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12128 (vectorizable_bswap): Likewise.
12129 (vectorizable_call): Likewise.
12130 (vectorizable_conversion): Likewise.
12131 (vectorizable_assignment): Likewise.
12132 (vectorizable_shift): Likewise.
12133 (vectorizable_operation): Likewise.
12134 (vectorizable_store): Likewise.
12135 (vectorizable_load): Likewise.
12136 (vectorizable_condition): Likewise.
12137 (vectorizable_comparison): Likewise.
12138 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12140 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12141 Alan Hayward <alan.hayward@arm.com>
12142 David Sherwood <david.sherwood@arm.com>
12144 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12145 of vect_init_vector.
12147 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12148 Alan Hayward <alan.hayward@arm.com>
12149 David Sherwood <david.sherwood@arm.com>
12151 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12152 an inline wrapper that provides a location.
12153 (gimple_build_vector): Likewise.
12154 * gimple-fold.c (gimple_build_vector_from_val): New function.
12155 (gimple_build_vector): Likewise.
12156 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12157 functions to build the initial value. Always return a gimple value.
12158 (get_initial_defs_for_reduction): Likewise. Only compute
12160 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12161 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12162 (vectorizable_induction): Use gimple_build_vector rather than
12165 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12166 Alan Hayward <alan.hayward@arm.com>
12167 David Sherwood <david.sherwood@arm.com>
12169 * target.h (vec_perm_indices): New typedef.
12170 (auto_vec_perm_indices): Likewise.
12171 * optabs-query.h: Include target.h
12172 (can_vec_perm_p): Take a vec_perm_indices *.
12173 * optabs-query.c (can_vec_perm_p): Likewise.
12174 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12175 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12176 * tree-vect-generic.c (lower_vec_perm): Likewise.
12177 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12178 (vect_grouped_load_supported): Likewise.
12179 (vect_shift_permute_load_chain): Likewise.
12180 (vect_permute_store_chain): Use auto_vec_perm_indices.
12181 (vect_permute_load_chain): Likewise.
12182 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12183 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12184 Update uses of can_vec_perm_p.
12185 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12186 mode with a number of elements. Take a vec_perm_indices *.
12187 (vect_create_epilog_for_reduction): Update accordingly.
12188 Use auto_vec_perm_indices.
12189 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12190 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12191 (vect_transform_slp_perm_load): Likewise.
12192 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12193 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12194 (vect_gen_perm_mask_checked): Likewise.
12195 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12196 (vect_gen_perm_mask_checked): Likewise.
12197 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12198 (vectorizable_store): Likewise.
12199 (vectorizable_load): Likewise.
12200 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12201 (vectorizable_bswap): Likewise.
12203 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12204 Alan Hayward <alan.hayward@arm.com>
12205 David Sherwood <david.sherwood@arm.com>
12207 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12208 * tree.c (build_vector): Likewise.
12209 (build_vector_from_ctor): Update accordingly.
12210 (build_vector_from_val): Likewise.
12211 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12212 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12213 * tree-vect-generic.c (add_rshift): Likewise.
12214 (expand_vector_divmod): Likewise.
12215 (optimize_vector_constructor): Likewise.
12216 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12217 (vect_transform_slp_perm_load): Likewise.
12218 (vect_schedule_slp_instance): Likewise.
12219 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12220 (vectorizable_call): Likewise.
12221 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12222 * expmed.c (make_tree): Likewise.
12223 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12224 a vector passed to build_vector.
12225 (fold_convert_const): Likewise.
12226 (exact_inverse): Likewise.
12227 (fold_ternary_loc): Likewise.
12228 (fold_relational_const): Likewise.
12229 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12230 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12231 (const_unop): Likewise. Store the reduction accumulator in a
12232 variable rather than an array.
12233 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12234 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12235 the new vector, rather than constructing it after the input arrays.
12236 (native_interpret_vector): Use auto_vec<tree> when building
12237 a vector passed to build_vector. Add elements in order.
12238 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12239 auto_vec<tree> when building a vector passed to build_vector.
12240 (vect_create_epilog_for_reduction): Likewise.
12241 (vectorizable_induction): Likewise.
12242 (get_initial_def_for_reduction): Likewise. Fix indentation of
12244 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12246 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12247 passed to build_vector.
12249 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12250 Alan Hayward <alan.hayward@arm.com>
12251 David Sherwood <david.sherwood@arm.com>
12253 * tree-core.h (tree_base::u): Add an "nelts" field.
12254 (tree_vector): Use VECTOR_CST_NELTS as the length.
12255 * tree.c (tree_size): Likewise.
12256 (make_vector): Initialize VECTOR_CST_NELTS.
12257 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12258 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12259 TYPE_VECTOR_SUBPARTS.
12260 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12261 as the number of units, setting it from VECTOR_CST_NELTS.
12262 (const_vector_from_tree): Likewise.
12263 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12264 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12265 (fold_negate_expr_1): Likewise.
12266 (fold_convert_const): Likewise.
12267 (const_binop): Likewise. Differentiate the number of output and
12269 (const_unop): Likewise.
12270 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12271 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12272 in cases that did the opposite.
12274 2017-09-14 Richard Biener <rguenther@suse.de>
12276 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12279 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12281 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12283 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12286 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12287 if and where to split a bb, except for splitting before debug insn
12288 sequences followed by non-label real insn. Delete debug insns
12289 in between basic blocks.
12291 * combine.c (make_compound_operation_int): Formatting fixes.
12293 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12294 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12295 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12296 * config/sol2.h (LINK_EH_SPEC): Likewise.
12297 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12298 * config/s390/linux.h (LINK_SPEC): Likewise.
12299 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12300 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12301 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12302 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12303 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12304 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12305 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12306 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12308 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12310 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12312 (ENDFILE_LINUX_SPEC): Likewise.
12313 (LINK_EH_SPEC): Likewise.
12314 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12315 (LINK_OS_LINUX_SPEC32): Likewise.
12316 (LINK_OS_LINUX_SPEC64): Likewise.
12317 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12318 (LINK_OS_LINUX_SPEC): Likewise.
12320 2017-09-13 Martin Liska <mliska@suse.cz>
12322 PR middle-end/82154
12323 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12324 CASE_HIGH is NULL_TREE.
12326 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12327 Alan Hayward <alan.hayward@arm.com>
12328 David Sherwood <david.sherwood@arm.com>
12330 * target.def (secondary_memory_needed): New hook.
12331 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12332 instead of SECONDARY_MEMORY_NEEDED.
12333 (secondary_memory_needed_mode): Likewise.
12334 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12335 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12336 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12337 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12338 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12339 * doc/tm.texi: Regenerate.
12340 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12341 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12342 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12343 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12344 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12345 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12346 mode argument first and change the reg_class arguments to reg_class_t.
12347 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12348 Make static. Update the call to inline_secondary_memory_needed.
12349 (ix86_register_move_cost): Update the call to
12350 inline_secondary_memory_needed.
12351 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12352 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12354 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12355 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12357 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12358 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12359 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12360 and match hook interface. Add comment from mips.h.
12361 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12362 * config/mmix/mmix.md (truncdfsf2): Refer to
12363 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12365 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12366 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12367 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12368 (pa_secondary_memory_needed): New function.
12369 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12370 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12371 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12372 (pdp11_secondary_memory_needed): Make static and match hook interface.
12373 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12374 * config/powerpcspe/powerpcspe-protos.h
12375 (rs6000_secondary_memory_needed_ptr): Delete.
12376 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12378 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12379 (rs6000_option_override_internal): Assign to
12380 targetm.secondary_memory_needed rather than
12381 rs6000_secondary_memory_needed_ptr.
12382 (rs6000_secondary_memory_needed): Match hook interface.
12383 (rs6000_debug_secondary_memory_needed): Likewise.
12384 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12385 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12386 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12387 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12388 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12389 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12391 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12392 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12393 (rs6000_option_override_internal): Assign to
12394 targetm.secondary_memory_needed rather than
12395 rs6000_secondary_memory_needed_ptr.
12396 (rs6000_secondary_memory_needed): Match hook interface.
12397 (rs6000_debug_secondary_memory_needed): Likewise.
12398 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12399 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12400 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12401 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12402 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12403 (sparc_secondary_memory_needed): New function.
12404 * lra-constraints.c (check_and_process_move): Refer to
12405 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12407 (curr_insn_transform): Likewise.
12408 (process_alt_operands): Use targetm.secondary_memory_needed
12409 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12410 (check_secondary_memory_needed_p): Likewise.
12411 (choose_split_class): Likewise.
12412 * reload.c: Unconditionally include code that was previously
12413 conditional on SECONDARY_MEMORY_NEEDED.
12414 (push_secondary_reload): Use targetm.secondary_memory_needed
12415 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12416 (push_reload): Likewise.
12417 * reload1.c: Unconditionally include code that was previously
12418 conditional on SECONDARY_MEMORY_NEEDED.
12419 (choose_reload_regs): Use targetm.secondary_memory_needed
12420 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12421 (gen_reload): Likewise.
12422 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12424 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12425 Alan Hayward <alan.hayward@arm.com>
12426 David Sherwood <david.sherwood@arm.com>
12428 * target.def (secondary_memory_needed_mode): New hook:
12429 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12430 * targhooks.h (default_secondary_memory_needed_mode): New function.
12431 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12432 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12433 * doc/tm.texi: Regenerate.
12434 * lra-constraints.c (check_and_process_move): Use
12435 targetm.secondary_memory_needed_mode instead of
12436 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12437 (curr_insn_transform): Likewise.
12438 * reload.c (get_secondary_mem): Likewise.
12439 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12440 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12442 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12443 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12444 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12445 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12446 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12448 * config/powerpcspe/powerpcspe-protos.h
12449 (rs6000_secondary_memory_needed_mode): Delete.
12450 * config/powerpcspe/powerpcspe.c
12451 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12452 (rs6000_secondary_memory_needed_mode): Make static.
12453 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12454 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12456 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12458 (rs6000_secondary_memory_needed_mode): Make static.
12459 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12460 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12461 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12462 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12463 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12465 (sparc_secondary_memory_needed_mode): New function.
12466 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12468 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12470 * config/aarch64/constraints.md (Umq): New constraint.
12471 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12473 (mov<mode>): Update condition.
12475 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12477 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12478 when bitposition is the same.
12480 2017-09-13 Richard Biener <rguenther@suse.de>
12482 * dwarf2out.c (output_die_symbol): Remove.
12483 (output_die): Do not output a DIEs symbol.
12485 2017-09-13 Richard Biener <rguenther@suse.de>
12487 PR middle-end/82128
12488 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12489 default-def to avoid breaking iterator update with the weird
12490 interaction with cgraph_update_edges_for_call_stmt_node.
12492 2017-09-13 Richard Biener <rguenther@suse.de>
12494 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12495 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12496 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12497 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12499 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12501 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12502 Disable pc relative literal load irrespective of
12503 TARGET_FIX_ERR_A53_84341 for default.
12505 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12507 * config/sparc/sparc.c (output_return): Output the source location of
12508 the insn in the delay slot, if any.
12509 (output_sibcall): Likewise.
12511 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12514 * common.opt (-static-pie): New alias.
12515 (shared): Negate static-pie.
12516 (-no-pie): Update help text.
12518 (static-pie): New option.
12519 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12520 -static-pie support.
12521 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12522 (LINK_EH_SPEC): Likewise.
12523 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12524 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12525 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12526 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12527 (init_gcc_specs): Likewise.
12528 (init_spec): Likewise.
12529 (display_help): Update help message for -pie.
12530 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12533 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12535 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12536 (movdi_aarch64): Likewise.
12537 (movti_aarch64): Likewise.
12539 2017-09-12 Simon Wright <simon@pushface.org>
12542 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12543 calculation of the minor version, always output as 0.
12545 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12548 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12549 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12550 on it early, rather than manual conversion late. For
12551 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12552 instead of performing manual conversion.
12554 2017-09-12 Carl Love <cel@us.ibm.com>
12556 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12557 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12559 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12560 VMULOSW): Add definitions.
12561 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12562 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12563 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12564 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12565 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12567 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12569 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12571 (movti_aarch64): Likewise.
12572 (movdf_aarch64): Likewise.
12573 (movtf_aarch64): Likewise.
12574 (load_pairdi): Likewise.
12575 (store_pairdi): Likewise.
12576 (load_pairdf): Likewise.
12577 (store_pairdf): Likewise.
12578 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12579 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12580 (ldr_got_small_<mode>): Likewise.
12581 (ldr_got_small_28k_<mode>): Likewise.
12582 (ldr_got_tiny): Likewise.
12583 * config/aarch64/iterators.md (ldst_sz): New.
12584 (ldpstp_sz): Likewise.
12585 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12587 (thunderx_load): Split load_8 to load_16.
12588 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12590 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12591 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12592 (xgene1_store_pair): Split store_8 to store_16.
12593 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12594 (falkor_st_0_st_sd): Split store_8 to store_16.
12596 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12598 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12599 and store1/2/3/4 to store_4/8/12/16.
12600 * config/aarch64/aarch64.md: Update for rename.
12601 * config/arm/arm.md: Likewise.: Likewise.
12602 * config/arm/arm.c: Likewise.
12603 * config/arm/thumb1.md: Likewise.
12604 * config/arm/thumb2.md: Likewise.
12605 * config/arm/vfp.md: Likewise.
12606 * config/arm/arm-generic.md: Likewise.
12607 * config/arm/arm1020e.md: Likewise.
12608 * config/arm/arm1026ejs.md: Likewise.
12609 * config/arm/arm1136jfs.md: Likewise.
12610 * config/arm/arm926ejs.md: Likewise.
12611 * config/arm/cortex-a15.md: Likewise.
12612 * config/arm/cortex-a17.md: Likewise.
12613 * config/arm/cortex-a5.md: Likewise.
12614 * config/arm/cortex-a53.md: Likewise.
12615 * config/arm/cortex-a57.md: Likewise.
12616 * config/arm/cortex-a7.md: Likewise.
12617 * config/arm/cortex-a8.md: Likewise.
12618 * config/arm/cortex-a9.md: Likewise.
12619 * config/arm/cortex-m4.md: Likewise.
12620 * config/arm/cortex-m7.md: Likewise.
12621 * config/arm/cortex-r4.md: Likewise.
12622 * config/arm/exynos-m1.md: Likewise.
12623 * config/arm/fa526.md: Likewise.
12624 * config/arm/fa606te.md: Likewise.
12625 * config/arm/fa626te.md: Likewise.
12626 * config/arm/fa726te.md: Likewise.
12627 * config/arm/fmp626.md: Likewise.
12628 * config/arm/iwmmxt.md: Likewise.
12629 * config/arm/ldmstm.md: Likewise.
12630 * config/arm/marvell-pj4.md: Likewise.
12631 * config/arm/xgene1.md: Likewise.
12632 * config/aarch64/thunderx.md: Likewise.
12633 * config/aarch64/thunderx2t99.md: Likewise.
12634 * config/aarch64/falkor.md: Likewise.
12636 2017-09-12 Martin Liska <mliska@suse.cz>
12638 * attribs.c (private_lookup_attribute): New function.
12639 * attribs.h (private_lookup_attribute): Declared here.
12640 (lookup_attribute): Called from this place.
12642 2017-09-12 Richard Biener <rguenther@suse.de>
12644 PR tree-optimization/82157
12645 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12646 stmts with side-effects.
12648 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12649 Alan Hayward <alan.hayward@arm.com>
12650 David Sherwood <david.sherwood@arm.com>
12652 * target.def (hard_regno_nregs): New hook.
12653 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12654 * targhooks.h (default_hard_regno_nregs): Declare.
12655 * targhooks.c (default_hard_regno_nregs): New function.
12656 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12657 (TARGET_HARD_REGNO_NREGS): ...this hook.
12658 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12659 (CLASS_MAX_NREGS): Likewise.
12660 * doc/tm.texi: Regenerate.
12661 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12662 instead of HARD_REGNO_NREGS.
12663 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12664 HARD_REGNO_NREGS in the comment.
12665 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12666 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12667 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12668 Return an unsigned int.
12669 (TARGET_HARD_REGNO_NREGS): Redefine.
12670 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12671 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12672 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12673 (arc_hard_regno_nregs): New function.
12674 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12675 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12676 (arm_hard_regno_nregs): New function.
12677 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12678 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12679 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12680 (TARGET_HARD_REGNO_NREGS): Redefine.
12681 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12682 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12683 (HARD_REGNO_NREGS): Delete.
12684 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12685 (cr16_hard_regno_nregs): New function.
12686 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12687 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12688 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12689 (cris_hard_regno_nregs): New function.
12690 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12691 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12692 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12693 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12694 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12695 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12696 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12697 (frv_hard_regno_nregs): Make static. Take and return an
12699 (frv_class_max_nregs): Remove outdated copy of documentation.
12700 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12701 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12702 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12703 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12704 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12705 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12706 (TARGET_HARD_REGNO_NREGS): Redefine.
12707 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12708 (CLASS_MAX_NREGS): Update comment.
12709 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12710 (ia64_hard_regno_nregs): New function.
12711 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12712 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12713 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12714 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12715 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12717 (m32c_hard_regno_nregs): Likewise. Make static.
12718 (TARGET_HARD_REGNO_NREGS): Redefine.
12719 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12720 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12721 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12722 (m68k_hard_regno_nregs): New function.
12723 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12724 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12725 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12726 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12727 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12728 Take and return an unsigned int.
12729 (TARGET_HARD_REGNO_NREGS): Redefine.
12730 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12731 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12732 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12733 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12734 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12735 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12736 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12737 (msp430_hard_regno_nregs): Make static. Take and return an
12739 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12740 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12741 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12742 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12743 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12744 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12745 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12746 (TARGET_HARD_REGNO_NREGS): Redefine.
12747 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12748 (PA_HARD_REGNO_NREGS): ...this.
12749 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12750 (PA_HARD_REGNO_NREGS): ...this.
12751 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12752 (pa_hard_regno_nregs): New function.
12753 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12754 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12755 (pdp11_hard_regno_nregs): New function.
12756 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12757 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12758 (rs6000_hard_regno_nregs_hook): New function.
12759 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12760 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12761 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12762 Take and return an unsigned int. Move earlier in file.
12763 (TARGET_HARD_REGNO_NREGS): Redefine.
12764 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12765 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12766 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12767 (rl78_hard_regno_nregs): Make static. Take and return an
12769 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12770 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12771 (rs6000_hard_regno_nregs_hook): New function.
12772 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12773 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12774 (TARGET_HARD_REGNO_NREGS): Redefine.
12775 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12776 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12777 instead of HARD_REGNO_NREGS.
12778 (s390_hard_regno_nregs): New function.
12779 (s390_hard_regno_mode_ok): Add comment from s390.h.
12780 (TARGET_HARD_REGNO_NREGS): Redefine.
12781 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12782 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12783 (sh_hard_regno_nregs): New function.
12784 (sh_pass_in_reg_p): Use it.
12785 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12786 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12787 (sparc_hard_regno_nregs): New function.
12788 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12789 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12790 (spu_function_arg_advance): Use it, supplying a valid register number.
12791 (TARGET_HARD_REGNO_NREGS): Redefine.
12792 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12793 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12794 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12795 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12796 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12797 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12798 (CLASS_MAX_NREGS): Remove copy of old documentation.
12799 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12800 (visium_hard_regno_nregs): New function.
12801 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12802 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12803 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12804 xtensa_hard_regno_nregs): New function.
12805 * system.h (HARD_REGNO_NREGS): Poison.
12807 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12809 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12810 hard_regno_nregs instead of HARD_REGNO_NREGS.
12811 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12812 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12813 (c6x_expand_epilogue): Likewise.
12814 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12815 (frv_read_iacc_argument): Likewise.
12816 * config/sh/sh.c: Include regs.h.
12817 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12818 (regs_used): Likewise.
12819 (output_stack_adjust): Likewise.
12820 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12821 * expmed.c: Include regs.h.
12822 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12823 * ree.c: Include regs.h.
12824 (combine_reaching_defs): Use hard_regno_nregs instead of
12826 (add_removable_extension): Likewise.
12828 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12830 * regs.h (hard_regno_nregs): Turn into a function.
12831 (end_hard_regno): Update accordingly.
12832 * caller-save.c (setup_save_areas): Likewise.
12833 (save_call_clobbered_regs): Likewise.
12834 (replace_reg_with_saved_mem): Likewise.
12835 (insert_restore): Likewise.
12836 (insert_save): Likewise.
12837 * combine.c (can_change_dest_mode): Likewise.
12838 (move_deaths): Likewise.
12839 (distribute_notes): Likewise.
12840 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12841 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12842 (rs6000_split_multireg_move): Likewise.
12843 (rs6000_register_move_cost): Likewise.
12844 (rs6000_memory_move_cost): Likewise.
12845 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12846 (rs6000_split_multireg_move): Likewise.
12847 (rs6000_register_move_cost): Likewise.
12848 (rs6000_memory_move_cost): Likewise.
12849 * cselib.c (cselib_reset_table): Likewise.
12850 (cselib_lookup_1): Likewise.
12851 * emit-rtl.c (set_mode_and_regno): Likewise.
12852 * function.c (aggregate_value_p): Likewise.
12853 * ira-color.c (setup_profitable_hard_regs): Likewise.
12854 (check_hard_reg_p): Likewise.
12855 (calculate_saved_nregs): Likewise.
12856 (assign_hard_reg): Likewise.
12857 (improve_allocation): Likewise.
12858 (calculate_spill_cost): Likewise.
12859 * ira-emit.c (modify_move_list): Likewise.
12860 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12861 (ira_hard_reg_in_set_p): Likewise.
12862 * ira.c (setup_reg_mode_hard_regset): Likewise.
12863 (clarify_prohibited_class_mode_regs): Likewise.
12864 (check_allocation): Likewise.
12865 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12866 (lra_setup_reg_renumber): Likewise.
12867 (setup_try_hard_regno_pseudos): Likewise.
12868 (spill_for): Likewise.
12869 (assign_hard_regno): Likewise.
12870 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12871 * lra-constraints.c (in_class_p): Likewise.
12872 (lra_constraint_offset): Likewise.
12873 (simplify_operand_subreg): Likewise.
12874 (lra_constraints): Likewise.
12875 (split_reg): Likewise.
12876 (split_if_necessary): Likewise.
12877 (invariant_p): Likewise.
12878 (inherit_in_ebb): Likewise.
12879 * lra-lives.c (process_bb_lives): Likewise.
12880 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12881 (get_hard_regs): Likewise.
12882 (do_remat): Likewise.
12883 * lra-spills.c (assign_spill_hard_regs): Likewise.
12884 * mode-switching.c (create_pre_exit): Likewise.
12885 * postreload.c (reload_combine_recognize_pattern): Likewise.
12886 * recog.c (peep2_find_free_register): Likewise.
12887 * regcprop.c (kill_value_regno): Likewise.
12888 (set_value_regno): Likewise.
12889 (copy_value): Likewise.
12890 (maybe_mode_change): Likewise.
12891 (find_oldest_value_reg): Likewise.
12892 (copyprop_hardreg_forward_1): Likewise.
12893 * regrename.c (check_new_reg_p): Likewise.
12894 (regrename_do_replace): Likewise.
12895 * reload.c (push_reload): Likewise.
12896 (combine_reloads): Likewise.
12897 (find_dummy_reload): Likewise.
12898 (operands_match_p): Likewise.
12899 (find_reloads): Likewise.
12900 (find_equiv_reg): Likewise.
12901 (reload_adjust_reg_for_mode): Likewise.
12902 * reload1.c (count_pseudo): Likewise.
12903 (count_spilled_pseudo): Likewise.
12904 (find_reg): Likewise.
12905 (clear_reload_reg_in_use): Likewise.
12906 (free_for_value_p): Likewise.
12907 (allocate_reload_reg): Likewise.
12908 (choose_reload_regs): Likewise.
12909 (reload_adjust_reg_for_temp): Likewise.
12910 (emit_reload_insns): Likewise.
12911 (delete_output_reload): Likewise.
12912 * rtlanal.c (subreg_get_info): Likewise.
12913 * sched-deps.c (sched_analyze_reg): Likewise.
12914 * sel-sched.c (init_regs_for_mode): Likewise.
12915 (mark_unavailable_hard_regs): Likewise.
12916 (choose_best_reg_1): Likewise.
12917 (verify_target_availability): Likewise.
12918 * valtrack.c (dead_debug_insert_temp): Likewise.
12919 * var-tracking.c (track_loc_p): Likewise.
12920 (emit_note_insn_var_location): Likewise.
12921 * varasm.c (make_decl_rtl): Likewise.
12922 * reginfo.c (choose_hard_reg_mode): Likewise.
12923 (init_reg_modes_target): Refer directly to
12924 this_target_regs->x_hard_regno_nregs.
12926 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12928 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12929 instead of hard_regno_nregs.
12931 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12933 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12934 end_hard_regno instead of hard_regno_nregs.
12935 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12936 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12937 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12938 * ira-color.c (improve_allocation): Likewise.
12939 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12940 * lra-lives.c (mark_regno_live): Likewise.
12941 (mark_regno_dead): Likewise.
12942 * lra-remat.c (operand_to_remat): Likewise.
12943 * lra.c (collect_non_operand_hard_regs): Likewise.
12944 * postreload.c (reload_combine_note_store): Likewise.
12945 (move2add_valid_value_p): Likewise.
12946 * reload.c (regno_clobbered_p): Likewise.
12948 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12950 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12952 * config/v850/v850.c (v850_reorg): Likewise.
12953 * reload.c (refers_to_regno_for_reload_p): Likewise.
12954 (find_equiv_reg): Likewise.
12955 * reload1.c (reload_reg_reaches_end_p): Likewise.
12957 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12959 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12961 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12962 * config/arm/arm.c (output_move_neon): Likewise.
12963 (arm_attr_length_move_neon): Likewise.
12964 (neon_split_vcombine): Likewise.
12965 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12966 (c6x_mark_reg_written): Likewise.
12967 (c6x_dwarf_register_span): Likewise.
12968 * config/i386/i386.c (ix86_save_reg): Likewise.
12969 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12970 (rws_access_reg): Likewise.
12971 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12972 * mode-switching.c (create_pre_exit): Likewise.
12973 * ree.c (combine_reaching_defs): Likewise.
12974 (add_removable_extension): Likewise.
12975 * regcprop.c (find_oldest_value_reg): Likewise.
12976 (copyprop_hardreg_forward_1): Likewise.
12977 * reload.c (reload_inner_reg_of_subreg): Likewise.
12978 (push_reload): Likewise.
12979 (combine_reloads): Likewise.
12980 (find_dummy_reload): Likewise.
12981 (reload_adjust_reg_for_mode): Likewise.
12982 * reload1.c (find_reload_regs): Likewise.
12983 (forget_old_reloads_1): Likewise.
12984 (reload_reg_free_for_value_p): Likewise.
12985 (reload_adjust_reg_for_temp): Likewise.
12986 (emit_reload_insns): Likewise.
12987 (delete_output_reload): Likewise.
12988 * sel-sched.c (choose_best_reg_1): Likewise.
12989 (choose_best_pseudo_reg): Likewise.
12991 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12992 Alan Hayward <alan.hayward@arm.com>
12993 David Sherwood <david.sherwood@arm.com>
12995 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12996 * target.def (slow_unaligned_access): New hook.
12997 * targhooks.h (default_slow_unaligned_access): Declare.
12998 * targhooks.c (default_slow_unaligned_access): New function.
12999 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13000 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13001 * doc/tm.texi: Regenerate.
13002 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13003 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13004 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13006 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13007 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13009 (rs6000_slow_unaligned_access): New function.
13010 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13011 (expand_block_compare): Likewise.
13012 (expand_strn_compare): Likewise.
13013 (rs6000_rtx_costs): Likewise.
13014 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13015 (riscv_slow_unaligned_access): Likewise.
13016 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13017 (riscv_slow_unaligned_access_p): ...this and make static.
13018 (riscv_option_override): Update accordingly.
13019 (riscv_slow_unaligned_access): New function.
13020 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13021 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13022 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13023 (rs6000_slow_unaligned_access): New function.
13024 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13025 (rs6000_rtx_costs): Likewise.
13026 * config/rs6000/rs6000-string.c (expand_block_compare)
13027 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13028 of SLOW_UNALIGNED_ACCESS.
13029 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13030 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13031 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13032 of SLOW_UNALIGNED_ACCESS.
13033 * expmed.c (simple_mem_bitfield_p): Likewise.
13034 * expr.c (alignment_for_piecewise_move): Likewise.
13035 (emit_group_load_1): Likewise.
13036 (emit_group_store): Likewise.
13037 (copy_blkmode_from_reg): Likewise.
13038 (emit_push_insn): Likewise.
13039 (expand_assignment): Likewise.
13040 (store_field): Likewise.
13041 (expand_expr_real_1): Likewise.
13042 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13043 * lra-constraints.c (simplify_operand_subreg): Likewise.
13044 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13045 * gimple-ssa-store-merging.c: Likewise in block comment at start
13047 * tree-ssa-strlen.c: Include target.h.
13048 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13049 of SLOW_UNALIGNED_ACCESS.
13050 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13052 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13054 PR rtl-optimization/82185
13055 * expmed.c (emit_store_flag_int): Only test tem if it has been
13058 2017-09-12 Richard Biener <rguenther@suse.de>
13060 PR middle-end/82149
13061 * match.pd ((FTYPE) N CMP CST): Fix typo.
13063 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13065 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13067 (mips_near_type_p): Add 'short_call' attribute as a synonym
13069 * doc/extend.texi (short_call): Document new function attribute.
13071 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13074 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13075 assertion check that in the condition.
13076 (get_atomic_generic_size): Likewise. Before testing if parameter
13077 has pointer type, if it has array type, call for C++
13078 default_conversion to perform array-to-pointer conversion.
13080 2017-09-12 Richard Biener <rguenther@suse.de>
13082 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13083 for operations we cannot scalarize.
13085 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13087 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13088 vectors heap vectors. Clean up comments.
13089 Make visited_bbs a reference.
13090 (profitable_jump_thread_path): Make GC
13091 vectors heap vectors. Clean up comments.
13093 (convert_and_register_jump_thread_path): Make GC vectors heap
13095 (check_subpath_and_update_thread_path): Same. Clean up comments.
13096 Make visited_bbs a reference.
13097 (handle_phi): Abstract common code to to
13098 register_jump_thread_path_if_profitable.
13099 Rename VAR_BB to DEF_BB.
13101 Make GC vectors heap vectors.
13102 Make visited_bbs a reference.
13103 (handle_assignment): Same.
13104 (register_jump_thread_path_if_profitable): New.
13105 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13107 Make GC vectors heap vectors. Clean up comments.
13108 Make visited_bbs a reference.
13109 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13110 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13113 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13116 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13117 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13119 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13121 Revert r251800 and r251799.
13123 2017-09-11 Martin Jambor <mjambor@suse.cz>
13126 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13127 arguments in advance.
13128 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13129 use it to find predecessor edges.
13130 (naive_outof_ssa): Collect vector of predecessors.
13132 2017-09-08 Jason Merrill <jason@redhat.com>
13134 PR c++/70029 - ICE with ref-qualifier and -flto
13135 * langhooks.h (struct lang_hooks_for_types): Add
13136 copy_lang_qualifiers.
13137 * attribs.c (build_type_attribute_qual_variant): Use it.
13138 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13140 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13141 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13143 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13146 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13147 (*mulsi3_sp64): New instruction.
13148 (mulsi3): New expander.
13150 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13152 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13154 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13156 * sancov.c: Include memmodel.h.
13158 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13161 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13164 2017-09-07 Carl Love <cel@us.ibm.com>
13166 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13167 the sldi instruction.
13169 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13171 * sancov.c: Include tm_p.h.
13173 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13176 * output.h (switch_to_other_text_partition): New declaration.
13177 * varasm.c (switch_to_other_text_partition): New function.
13178 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13179 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13180 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13181 to the other text partition before emitting LCL label and switch back
13182 after emitting the word after it.
13184 2017-09-07 Richard Biener <rguenther@suse.de>
13186 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13187 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13188 critical edges here, after loop init.
13189 (pass_data_pre): Remove PROP_no_crit_edges flags.
13190 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13191 for valueization of call args to avoid leaking VN_TOP.
13192 (visit_use): Assert we do not visit default defs.
13193 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13194 Use error_mark_node to more easily detect leaking VN_TOP.
13195 All default-defs are varying, not VN_TOP. Mark them visited.
13196 (run_scc_vn): Make code match comment.
13198 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13200 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13201 OPTION_MASK_FLOAT128_KEYWORD.
13202 (POWERPC_MASKS): Likewise.
13203 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13204 support for the -mfloat128-type option, and make -mfloat128
13205 default on PowerPC Linux systems. Define or undefine
13206 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13207 Define __float128 to be __ieee128 if IEEE 128-bit support is
13208 enabled, or undefine it.
13209 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13210 Delete defining __FLOAT128_TYPE__.
13211 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13212 -mfloat128-type option and make -mfloat128 default on PowerPC
13214 (TARGET_FLOAT128_TYPE): Likewise.
13215 (-mfloat128-type): Likewise.
13216 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13217 Delete the -mfloat128-type option and make -mfloat128 default on
13218 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13219 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13220 128-bit floating point is enabled. Change tests from using
13221 -mfloat128-type to -mfloat128.
13222 (rs6000_mangle_type): Use the correct mangling for the __float128
13223 type even if normal long double is restricted to 64-bits.
13224 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13226 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13227 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13228 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13230 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13231 documentation for -mfloat128.
13233 2017-09-06 Olivier Hainque <hainque@adacore.com>
13235 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13237 2017-09-06 Wish Wu <wishwu007@gmail.com>
13238 Jakub Jelinek <jakub@redhat.com>
13240 * asan.c (initialize_sanitizer_builtins): Add
13241 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13242 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13243 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13244 BT_FN_VOID_UINT64_PTR variables.
13245 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13246 (BT_FN_VOID_UINT16_UINT16): Likewise.
13247 (BT_FN_VOID_UINT32_UINT32): Likewise.
13248 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13249 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13250 (BT_FN_VOID_UINT64_PTR): Likewise.
13251 * common.opt (flag_sanitize_coverage): New variable.
13252 (fsanitize-coverage=trace-pc): Remove.
13253 (fsanitize-coverage=): Add.
13254 * flag-types.h (enum sanitize_coverage_code): New enum.
13255 * fold-const.c (fold_range_test): Disable non-short-circuit
13256 optimization if flag_sanitize_coverage.
13257 (fold_truth_andor): Likewise.
13258 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13259 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13260 (coverage_sanitizer_opts): New array.
13261 (get_closest_sanitizer_option): Add OPTS argument, handle also
13262 OPT_fsanitize_coverage_.
13263 (parse_sanitizer_options): Adjusted to also handle
13264 OPT_fsanitize_coverage_.
13265 (common_handle_option): Add OPT_fsanitize_coverage_.
13266 * sancov.c (instrument_comparison, instrument_switch): New function.
13267 (sancov_pass): Add trace-cmp support.
13268 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13269 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13270 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13271 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13272 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13273 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13274 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13275 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13276 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13277 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13279 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13281 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13282 error. Only quit immediately if parsing is complete.
13283 (BEGIN): Initialize fatal_err and parse_done.
13284 (begin fpu, end fpu): Check number of arguments.
13285 (begin arch, end arch): Likewise.
13286 (begin cpu, end cpu): Likewise.
13287 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13288 (optalias): Likewise.
13290 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13292 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13293 * config/arm/arm-isa.h: Delete. Move definitions to ...
13294 * arm-cpus.in: ... here. Use new feature and fgroup values.
13295 * config/arm/arm.c (arm_option_override): Use lower case for feature
13297 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13298 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13299 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13301 (print_isa_bits_for): New function.
13302 (gen_isa): New function.
13303 (gen_comm_data): Use print_isa_bits_for.
13304 (define feature): New keyword.
13305 (define fgroup): New keyword.
13306 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13307 (arm-isa.h): Add rule to generate file.
13308 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13309 case for feature bit names.
13311 2017-09-06 Richard Biener <rguenther@suse.de>
13313 * tree-ssa-pre.c (NECESSARY): Remove.
13314 (create_expression_by_pieces): Do not touch pass-local flags.
13315 (insert_into_preds_of_block): Likewise.
13316 (do_pre_regular_insertion): Likewise.
13317 (eliminate_insert): Likewise.
13318 (eliminate_dom_walker::before_dom_children): Likewise.
13319 (fini_eliminate): Do not look at inserted_exprs.
13320 (mark_operand_necessary): Remove.
13321 (remove_dead_inserted_code): Replace with simple work-list
13322 algorithm based on inserted_exprs and SSA uses.
13323 (pass_pre::execute): Re-order fini_eliminate and
13324 remove_dead_inserted_code.
13326 2017-09-06 Olivier Hainque <hainque@adacore.com>
13328 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13329 for VxWorks 7. Adjust surrounding comments.
13331 2017-09-06 Richard Biener <rguenther@suse.de>
13333 * gimple-ssa-strength-reduction.c
13334 (find_candidates_dom_walker::before_dom_children): Also allow
13337 2017-09-06 Richard Biener <rguenther@suse.de>
13339 PR tree-optimization/82108
13340 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13341 for gap in the non-permutation SLP case.
13343 2017-09-06 Martin Jambor <mjambor@suse.cz>
13345 PR tree-optimization/82078
13346 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13347 add_access_to_work_queue...
13348 (build_accesses_from_assign): ...here.
13349 (propagate_all_subaccesses): Make sure racc is the group
13350 representative, if there is one.
13352 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13354 PR middle-end/82095
13355 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13358 2017-09-06 Richard Biener <rguenther@suse.de>
13360 * gimple-ssa-strength-reduction.c
13361 (find_candidates_dom_walker::before_doom_children): Use a
13362 type and not a mode check.
13364 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13367 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13369 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13371 2017-09-05 Jeff Law <law@redhat.com>
13373 PR tree-optimization/64910
13374 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13375 cases where we have 3 or more operands.
13377 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13379 PR middle-end/81768
13380 * omp-low.c (lower_omp_for): Recompute tree invariant if
13381 gimple_omp_for_initial/final is ADDR_EXPR.
13383 PR middle-end/81768
13384 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13385 into gimple val before gimplification fo the COND_EXPR.
13387 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13389 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13390 REGION_COPY argument.
13391 (thread_through_all_blocks): Remove unused argument to
13392 duplicate_thread_path.
13394 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13395 Alan Hayward <alan.hayward@arm.com>
13396 David Sherwood <david.sherwood@arm.com>
13398 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13399 Take a scalar_mode rather than a machine_mode.
13400 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13401 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13402 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13403 (aarch64_gen_adjusted_ldpstp): Likewise.
13404 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13406 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13407 Alan Hayward <alan.hayward@arm.com>
13408 David Sherwood <david.sherwood@arm.com>
13410 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13411 Take a scalar_int_mode instead of a machine_mode.
13412 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13413 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13414 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13415 (aarch64_simd_attr_length_rglist): Delete.
13416 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13417 a scalar_int_mode instead of a machine_mode.
13418 (aarch64_add_offset): Likewise.
13419 (aarch64_internal_mov_immediate): Likewise
13420 (aarch64_add_constant_internal): Likewise.
13421 (aarch64_add_constant): Likewise.
13422 (aarch64_movw_imm): Likewise.
13423 (aarch64_rtx_arith_op_extract_p): Likewise.
13424 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13425 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13426 Remove assert that the mode isn't a vector.
13427 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13428 (aarch64_expand_mov_immediate): Update calls after above changes.
13429 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13430 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13431 (aarch64_move_imm): Likewise.
13432 (aarch64_can_const_movi_rtx_p): Likewise.
13433 (aarch64_strip_extend): Likewise.
13434 (aarch64_extr_rtx_p): Likewise.
13435 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13436 a CONST_INT when the mode parameter is VOIDmode.
13437 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13439 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13441 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13442 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13443 (bitwise_type_for_mode): Update accordingly.
13445 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13447 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13448 * stor-layout.c (mode_for_size_tree): Likewise.
13449 (mode_for_array): Update accordingly.
13450 (layout_decl): Likewise.
13451 (compute_record_mode): Likewise. Only set the mode once.
13453 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13455 * target.def (get_mask_mode): Change return type to opt_mode.
13457 * doc/tm.texi: Regenerate.
13458 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13459 * targhooks.c (default_get_mask_mode): Likewise.
13460 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13461 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13462 targetm.get_mask_mode.
13463 * tree.c (build_truth_vector_type): Likewise.
13465 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13467 * machmode.h (mode_for_vector): Return an opt_mode.
13468 * stor-layout.c (mode_for_vector): Likewise.
13469 (mode_for_int_vector): Update accordingly.
13470 (layout_type): Likewise.
13471 * config/i386/i386.c (emit_memmov): Likewise.
13472 (ix86_expand_set_or_movmem): Likewise.
13473 (ix86_expand_vector_init): Likewise.
13474 (ix86_get_mask_mode): Likewise.
13475 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13477 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13478 * expmed.c (extract_bit_field_1): Likewise.
13479 * expr.c (expand_expr_real_2): Likewise.
13480 * optabs-query.c (can_vec_perm_p): Likewise.
13481 (can_vec_mask_load_store_p): Likewise.
13482 * optabs.c (expand_vec_perm): Likewise.
13483 * targhooks.c (default_get_mask_mode): Likewise.
13484 * tree-vect-stmts.c (vectorizable_store): Likewise.
13485 (vectorizable_load): Likewise.
13486 (get_vectype_for_scalar_type_and_size): Likewise.
13488 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13490 * machmode.h (mode_for_int_vector): New function.
13491 * stor-layout.c (mode_for_int_vector): Likewise.
13492 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13493 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13494 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13495 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13496 (s390_expand_vcond): Likewise.
13498 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13500 * machmode.h (opt_machine_mode): New type.
13501 (opt_mode<T>): Allow construction from anything that can be
13503 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13504 (mode_for_size): Return an opt_machine_mode.
13505 * stor-layout.c (mode_for_size): Likewise.
13506 (mode_for_size_tree): Update call accordingly.
13507 (bitwise_mode_for_mode): Likewise.
13508 (make_fract_type): Likewise.
13509 (make_accum_type): Likewise.
13510 * caller-save.c (replace_reg_with_saved_mem): Update call
13512 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13513 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13514 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13515 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13516 * expmed.c (extract_bit_field_1): Likewise.
13517 * reload.c (get_secondary_mem): Likewise.
13518 * varasm.c (assemble_integer): Likewise.
13519 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13522 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13524 * machmode.h (decimal_float_mode_for_size): New function.
13525 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13526 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13527 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13528 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13529 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13530 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13532 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13534 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13535 (get_builtin_sync_mode): Likewise.
13536 (expand_ifn_atomic_compare_exchange): Likewise.
13537 (expand_builtin_atomic_clear): Likewise.
13538 (expand_builtin_atomic_test_and_set): Likewise.
13539 (fold_builtin_atomic_always_lock_free): Likewise.
13540 * calls.c (compute_argument_addresses): Likewise.
13541 (emit_library_call_value_1): Likewise.
13542 (store_one_arg): Likewise.
13543 * combine.c (combine_instructions): Likewise.
13544 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13545 * config/arm/arm.c (arm_function_value): Likewise.
13546 (aapcs_allocate_return_reg): Likewise.
13547 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13548 * config/i386/i386.c (construct_container): Likewise.
13549 (ix86_gimplify_va_arg): Likewise.
13550 (ix86_expand_sse_cmp): Likewise.
13551 (emit_memmov): Likewise.
13552 (emit_memset): Likewise.
13553 (expand_small_movmem_or_setmem): Likewise.
13554 (ix86_expand_pextr): Likewise.
13555 (ix86_expand_pinsr): Likewise.
13556 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13557 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13559 * config/mips/mips.c (mips_function_value_1) Likewise.
13560 (mips_block_move_straight): Likewise.
13561 (mips_expand_ins_as_unaligned_store): Likewise.
13562 * config/powerpcspe/powerpcspe.c
13563 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13564 (rs6000_darwin64_record_arg_flush): Likewise.
13565 * config/rs6000/rs6000.c
13566 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13567 (rs6000_darwin64_record_arg_flush): Likewise.
13568 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13569 (sparc_function_value_1): Likewise.
13570 * config/spu/spu.c (adjust_operand): Likewise.
13571 (spu_emit_branch_or_set): Likewise.
13572 (arith_immediate_p): Likewise.
13573 * emit-rtl.c (gen_lowpart_common): Likewise.
13574 * expr.c (expand_expr_real_1): Likewise.
13575 * function.c (assign_parm_setup_block): Likewise.
13576 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13577 * reload1.c (alter_reg): Likewise.
13578 * stor-layout.c (mode_for_vector): Likewise.
13579 (layout_type): Likewise.
13581 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13583 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13584 (spu_convert_move): Likewise.
13585 * lower-subreg.c (resolve_simple_move): Likewise.
13587 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13590 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13591 define_insn to a define_expand.
13592 (altivec_vsum2sws_direct): New define_insn.
13593 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13595 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13597 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13600 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13605 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13606 realign stack if stack alignment needed is less than incoming
13609 2017-09-05 Marek Polacek <polacek@redhat.com>
13612 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13615 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13617 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13619 2017-09-05 Richard Biener <rguenther@suse.de>
13621 PR tree-optimization/82084
13622 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13624 2017-09-05 Richard Biener <rguenther@suse.de>
13626 PR tree-optimization/82102
13627 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13629 2017-09-05 Martin Liska <mliska@suse.cz>
13631 PR tree-optimization/82032
13632 * tree-cfg.c (generate_range_test): New function.
13633 * tree-cfg.h (generate_range_test): Declared here.
13634 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13635 (cleanup_control_expr_graph): Use it.
13636 * tree-switch-conversion.c (try_switch_expansion): Remove
13638 (emit_case_nodes): Use generate_range_test.
13640 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13643 * config/i386/i386.md (*<btsc><mode>_mask): Add
13644 TARGET_USE_BT to insn constraint.
13645 (*btr<mode>_mask): Ditto.
13647 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13649 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13650 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13652 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13655 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13656 TARGET_NEON and TARGET_IWMMXT.
13657 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13658 TARGET_NEON and TARGET_IWMMXT.
13659 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13661 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13663 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13664 (ix86_rewrite_tls_address): Ditto.
13665 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13666 (ix86_rewrite_tls_address_1): Ditto.
13667 (ix86_rewrite_tls_address): Ditto.
13668 * config/i386/predicates.md (tls_address_pattern): New predicate.
13669 * config/i386/i386.md (TLS address splitter): New splitter.
13671 2017-09-04 Richard Biener <rguenther@suse.de>
13673 PR tree-optimization/82084
13674 * fold-const.h (can_native_encode_string_p): Declare.
13675 * fold-const.c (can_native_encode_string_p): Factor out from ...
13676 (native_encode_string): ... here.
13677 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13678 vectorizing stores from constants we later cannot handle.
13680 2017-09-04 Marek Polacek <polacek@redhat.com>
13683 * doc/invoke.texi: Update -Wtautological-compare documentation.
13685 2017-09-04 Jeff Law <law@redhat.com>
13687 PR tree-optimization/64910
13688 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13689 swap the first and last operand if the last is a constant.
13691 2017-09-04 Marek Polacek <polacek@redhat.com>
13694 * convert.c (do_narrow): When sanitizing signed integer overflows,
13695 bail out for signed types.
13696 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13698 2017-09-04 Richard Biener <rguenther@suse.de>
13700 PR tree-optimization/82060
13701 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13702 Move devirtualization after stmt folding and before EH/AB/noreturn
13703 cleanup to get the stmt refs canonicalized. Use a bool instead
13704 of gimple_modified_p since that doesn't work for NOPs. Schedule
13705 NOPs generated by folding for removal.
13707 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13708 Alan Hayward <alan.hayward@arm.com>
13709 David Sherwood <david.sherwood@arm.com>
13711 * coretypes.h (pad_direction): New enum.
13712 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13713 (FUNCTION_ARG_PADDING): Likewise.
13714 * target.def (function_arg_padding): New hook.
13715 * targhooks.h (default_function_arg_padding): Declare.
13716 * targhooks.c (default_function_arg_padding): New function.
13717 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13718 (TARGET_FUNCTION_ARG_PADDING): ...this.
13719 * doc/tm.texi: Regenerate.
13720 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13721 instead of direction.
13722 (compute_argument_addresses): Likewise.
13723 (load_register_parameters): Likewise.
13724 (emit_library_call_value_1): Likewise.
13725 (store_one_arg): Use targetm.calls.function_arg_padding instead
13726 of FUNCTION_ARG_PADDING.
13727 (must_pass_in_stack_var_size_or_pad): Likewise.
13728 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13729 (emit_group_store): Likewise.
13730 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13731 instead of FUNCTION_ARG_PADDING.
13732 (emit_push_insn): Likewise, and propagate enum change throughout
13734 * function.h (direction): Delete.
13735 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13737 * function.c (assign_parm_find_stack_rtl): Likewise.
13738 (assign_parm_setup_block_p): Likewise.
13739 (assign_parm_setup_block): Likewise.
13740 (gimplify_parameters): Likewise.
13741 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13742 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13744 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13745 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13746 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13747 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13748 (aarch64_function_arg_padding): ...this new function.
13749 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13750 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13751 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13752 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13753 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13754 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13755 (arm_pad_arg_upward): Replace with...
13756 (arm_function_arg_padding): ...this new function.
13757 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13759 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13760 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13761 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13762 (ia64_hpux_function_arg_padding): Replace with...
13763 (ia64_function_arg_padding): ...this new function. Use pad_direction
13764 instead of direction. Check for TARGET_HPUX.
13765 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13766 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13767 (iq2000_function_arg_padding): New function.
13768 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13769 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13770 (mips_function_arg_padding): ...this new function.
13771 (mips_pad_reg_upward): Update accordingly.
13772 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13773 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13774 targetm.calls.function_arg_padding.
13775 (FUNCTION_ARG_PADDING): Delete.
13776 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13777 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13778 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13779 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13780 (nios2_block_reg_padding): Return pad_direction instead of direction.
13781 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13782 instead of direction.
13783 (nios2_function_arg_padding): Likewise. Make static.
13784 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13785 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13786 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13787 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13788 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13789 (pa_function_arg_padding): Make static. Return pad_direction instead
13791 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13792 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13793 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13794 instead of direction. Use targetm.calls.function_arg_padding.
13795 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13796 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13797 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13798 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13799 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13801 (function_arg_padding): Rename to...
13802 (rs6000_function_arg_padding): ...this. Make static. Return
13803 pad_direction instead of direction.
13804 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13805 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13806 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13807 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13808 instead of direction. Use targetm.calls.function_arg_padding.
13809 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13810 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13811 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13812 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13813 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13814 (function_arg_padding): Rename to...
13815 (rs6000_function_arg_padding): ...this. Make static. Return
13816 pad_direction instead of direction.
13817 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13818 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13819 * config/s390/s390.c (s390_function_arg_padding): New function.
13820 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13821 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13822 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13823 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13824 (function_arg_padding): Rename to...
13825 (sparc_function_arg_padding): ...this. Make static. Return
13826 pad_direction instead of direction.
13827 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13828 * config/spu/spu.c (spu_function_arg_padding): New function.
13829 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13830 * system.h (FUNCTION_ARG_PADDING): Poison.
13832 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13833 Alan Hayward <alan.hayward@arm.com>
13834 David Sherwood <david.sherwood@arm.com>
13836 * target.def (modes_tieable_p): New hook.
13837 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13838 (TARGET_MODES_TIEABLE_P): ...this.
13839 * doc/tm.texi.in: Regenerate.
13840 * hooks.h (hook_bool_mode_mode_true): Declare.
13841 * hooks.c (hook_bool_mode_mode_true): New function.
13842 * combine.c (subst): Use targetm.modes_tieable_p instead of
13844 * dse.c (find_shift_sequence): Likewise.
13845 * expmed.c (extract_low_bits): Likewise.
13846 * lower-subreg.c: Include target.h.
13847 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13849 * rtlanal.c (rtx_cost): Likewise.
13850 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13851 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13852 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13853 (TARGET_MODES_TIEABLE_P): Redefine.
13854 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13855 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13856 (TARGET_MODES_TIEABLE_P): Redefine.
13857 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13858 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13859 (arc_modes_tieable_p): New function.
13860 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13861 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13862 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13863 (arm_modes_tieable_p): Make static.
13864 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13865 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13866 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13867 (TARGET_MODES_TIEABLE_P): Redefine.
13868 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13869 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13870 (TARGET_MODES_TIEABLE_P): Redefine.
13871 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13872 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13873 (cr16_modes_tieable_p): New function.
13874 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13875 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13876 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13877 (TRULY_NOOP_TRUNCATION): Update comment.
13878 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13879 (TRULY_NOOP_TRUNCATION): Update comment.
13880 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13881 (frv_modes_tieable_p): New function.
13882 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13883 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13884 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13885 (TARGET_MODES_TIEABLE_P): Redefine.
13886 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13887 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13888 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13889 (TARGET_MODES_TIEABLE_P): Redefine.
13890 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13891 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13892 (ia64_modes_tieable_p): New function.
13893 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13894 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13895 (iq2000_modes_tieable_p): New function.
13896 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13897 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13898 (lm32_modes_tieable_p): New function.
13899 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13900 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13901 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13902 (TARGET_MODES_TIEABLE_P): Redefine.
13903 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13904 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13905 (m32r_modes_tieable_p): New function.
13906 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13907 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13908 (m68k_modes_tieable_p): New function.
13909 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13910 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13911 (mcore_modes_tieable_p): New function.
13912 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13913 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13915 (TARGET_MODES_TIEABLE_P): Redefine.
13916 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13917 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13918 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13919 (TARGET_MODES_TIEABLE_P): Redefine.
13920 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13921 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13922 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13923 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13924 (mn10300_modes_tieable_p): ...this and make static.
13925 (TARGET_MODES_TIEABLE_P): Redefine.
13926 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13927 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13928 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13929 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13930 (msp430_modes_tieable_p): Make static.
13931 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13932 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13933 (TARGET_MODES_TIEABLE_P): Redefine.
13934 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13935 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13936 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13937 (TARGET_MODES_TIEABLE_P): Redefine.
13938 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13939 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13940 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13941 (TARGET_MODES_TIEABLE_P): Redefine.
13942 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13943 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13944 (pdp11_modes_tieable_p): New function.
13945 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13946 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13947 (rs6000_modes_tieable_p): New function.
13948 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13949 * config/powerpcspe/powerpcspe.md: Update comment.
13950 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13951 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13952 (TARGET_MODES_TIEABLE_P): Redefine.
13953 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13954 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13955 (rl78_modes_tieable_p): New function.
13956 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13957 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13958 (rs6000_modes_tieable_p): New function.
13959 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13960 * config/rs6000/rs6000.md: Update comment.
13961 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13962 * config/rx/rx.c (rx_modes_tieable_p): New function.
13963 (TARGET_MODES_TIEABLE_P): Redefine.
13964 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13965 * config/s390/s390.c (s390_modes_tieable_p): New function.
13966 (TARGET_MODES_TIEABLE_P): Redefine.
13967 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13968 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13969 (sh_modes_tieable_p): New function.
13970 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13971 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13972 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13973 (sparc_modes_tieable_p): Make static.
13974 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13975 * config/spu/spu.c (spu_modes_tieable_p): New function.
13976 (TARGET_MODES_TIEABLE_P): Redefine.
13977 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13978 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13979 (TARGET_MODES_TIEABLE_P): Redefine.
13980 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13981 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13982 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13983 * config/v850/v850.c (v850_modes_tieable_p): New function.
13984 (TARGET_MODES_TIEABLE_P): Redefine.
13985 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13986 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13987 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13988 (visium_modes_tieable_p): New function.
13989 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13990 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13991 (xtensa_modes_tieable_p): New function.
13992 * system.h (MODES_TIEABLE_P): Poison.
13994 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13995 Alan Hayward <alan.hayward@arm.com>
13996 David Sherwood <david.sherwood@arm.com>
13998 * target.def (hard_regno_mode_ok): New hook.
13999 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14000 (TARGET_HARD_REGNO_MODE_OK): ...this.
14001 * doc/tm.texi.in: Regenerate.
14002 * hooks.h (hook_bool_uint_mode_true): Declare.
14003 * hooks.c (hook_bool_uint_mode_true): New function.
14004 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14005 HARD_REGNO_MODE_OK.
14006 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14007 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14008 instead of HARD_REGNO_MODE_OK.
14009 * caller-save.c: Include target.h.
14010 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14011 HARD_REGNO_MODE_OK.
14012 * combine.c (can_combine_p): Likewise.
14013 (combinable_i3pat): Likewise.
14014 (can_change_dest_mode): Likewise.
14015 * expr.c (init_expr_target): Likewise.
14016 (convert_move): Likewise.
14017 (convert_modes): Likewise.
14018 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14019 (setup_prohibited_mode_move_regs): Likewise.
14020 * ira.h (target_ira): Likewise.
14021 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14022 * lra-constraints.c (process_alt_operands): Likewise.
14023 (split_reg): Likewise.
14024 * recog.c (peep2_find_free_register): Likewise.
14025 * ree.c (combine_reaching_defs): Likewise.
14026 * regcprop.c (maybe_mode_change): Likewise.
14027 * reginfo.c (init_reg_sets_1): Likewise.
14028 (choose_hard_reg_mode): Likewise.
14029 (simplifiable_subregs): Likewise.
14030 * regrename.c (check_new_reg_p): Likewise.
14031 * reload.c (find_valid_class): Likewise.
14032 (find_valid_class_1): Likewise.
14033 (reload_inner_reg_of_subreg): Likewise.
14034 (push_reload): Likewise.
14035 (combine_reloads): Likewise.
14036 (find_dummy_reload): Likewise.
14037 (find_reloads): Likewise.
14038 * reload1.c (find_reg): Likewise.
14039 (set_reload_reg): Likewise.
14040 (allocate_reload_reg): Likewise.
14041 (choose_reload_regs): Likewise.
14042 (reload_adjust_reg_for_temp): Likewise.
14043 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14044 (simplify_subreg_regno): Likewise.
14045 * sel-sched.c (init_regs_for_mode): Likewise.
14046 * varasm.c (make_decl_rtl): Likewise.
14047 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14048 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14049 HARD_REGNO_MODE_OK.
14050 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14051 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14052 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14053 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14054 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14055 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14056 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14057 (arc_mode_class): Delete.
14058 (HARD_REGNO_MODE_OK): Delete.
14059 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14060 (arc_hard_regno_mode_ok): Rename old array to...
14061 (arc_hard_regno_mode_ok_modes): ...this.
14062 (arc_conditional_register_usage): Update accordingly.
14063 (arc_mode_class): Make static.
14064 (arc_hard_regno_mode_ok): New function.
14065 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14066 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14067 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14068 (arm_hard_regno_mode_ok): Make static.
14069 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14070 HARD_REGNO_MODE_OK.
14071 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14072 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14073 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14075 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14076 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14077 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14078 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14079 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14080 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14081 * config/bfin/predicates.md (valid_reg_operand): Use
14082 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14083 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14084 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14085 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14086 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14087 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14088 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14089 (cr16_hard_regno_mode_ok): Make static and return a bool.
14090 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14091 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14092 (cris_hard_regno_mode_ok): New function.
14093 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14094 (epiphany_mode_class): Delete.
14095 (HARD_REGNO_MODE_OK): Delete.
14096 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14097 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14098 (hard_regno_mode_ok): Rename to...
14099 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14100 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14101 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14102 HARD_REGNO_MODE_OK.
14103 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14104 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14105 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14106 (frv_hard_regno_mode_ok): Make static and return a bool.
14107 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14108 HARD_REGNO_MODE_OK.
14109 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14110 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14111 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14112 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14114 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14115 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14116 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14117 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14119 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14120 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14121 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14122 (ia64_hard_regno_mode_ok): New function.
14123 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14124 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14125 (iq2000_hard_regno_mode_ok): New function.
14126 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14127 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14128 (lm32_hard_regno_mode_ok): New function.
14129 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14130 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14131 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14132 instead of HARD_REGNO_MODE_OK.
14133 (m32c_hard_regno_ok): Rename to...
14134 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14135 (m32c_cannot_change_mode_class): Update accordingly.
14136 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14137 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14138 (m32r_mode_class): Delete.
14139 (HARD_REGNO_MODE_OK): Delete.
14140 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14141 (m32r_hard_regno_mode_ok): Rename to...
14142 (m32r_hard_regno_modes): ...this.
14143 (m32r_mode_class): Make static.
14144 (m32r_hard_regno_mode_ok): New function.
14145 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14146 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14147 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14148 (m68k_hard_regno_mode_ok): Make static.
14149 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14150 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14151 (mcore_hard_regno_mode_ok): New function.
14152 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14153 (HARD_REGNO_MODE_OK): Delete.
14154 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14156 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14157 (microblaze_hard_regno_mode_ok): New function.
14158 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14159 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14160 (mips_hard_regno_mode_ok): Delete.
14161 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14162 (mips_hard_regno_mode_ok_p): ...this and make static.
14163 (mips_hard_regno_mode_ok_p): Rename to...
14164 (mips_hard_regno_mode_ok_uncached): ...this.
14165 (mips_hard_regno_mode_ok): New function.
14166 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14167 of HARD_REGNO_MODE_OK.
14168 (mips_option_override): Update after above name changes.
14169 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14170 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14171 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14172 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14173 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14174 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14175 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14176 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14177 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14178 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14179 (msp430_hard_regno_mode_ok): Make static and return a bool.
14180 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14181 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14182 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14184 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14185 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14186 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14187 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14188 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14189 (PA_HARD_REGNO_MODE_OK): ...this
14190 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14191 (PA_HARD_REGNO_MODE_OK): ...this.
14192 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14193 (pa_hard_regno_mode_ok): New function.
14194 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14195 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14196 (pdp11_hard_regno_mode_ok): New function.
14197 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14198 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14200 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14202 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14203 (rs6000_hard_regno_mode_ok): Rename to...
14204 (rs6000_hard_regno_mode_ok_uncached): ...this.
14205 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14206 (rs6000_hard_regno_mode_ok): New function.
14207 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14208 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14209 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14210 (riscv_hard_regno_mode_ok): ...this and make static.
14211 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14212 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14213 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14214 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14215 (rl78_hard_regno_mode_ok): Make static and return bool.
14216 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14217 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14219 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14220 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14221 (rs6000_hard_regno_mode_ok): Rename to...
14222 (rs6000_hard_regno_mode_ok_uncached): ...this.
14223 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14224 (rs6000_hard_regno_mode_ok): New function.
14225 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14226 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14227 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14228 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14229 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14230 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14231 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14232 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14233 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14234 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14235 (sh_hard_regno_mode_ok): Make static.
14236 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14237 instead of HARD_REGNO_MODE_OK.
14238 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14239 (sparc_mode_class): Delete.
14240 (HARD_REGNO_MODE_OK): Delete.
14241 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14242 (hard_regno_mode_classes): Make static.
14243 (sparc_mode_class): Likewise.
14244 (sparc_hard_regno_mode_ok): New function.
14245 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14246 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14247 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14249 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14250 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14251 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14252 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14253 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14254 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14255 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14256 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14257 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14258 (visium_hard_regno_mode_ok): New function.
14259 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14260 instead of HARD_REGNO_MODE_OK.
14261 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14262 (HARD_REGNO_MODE_OK): Delete.
14263 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14264 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14265 (xtensa_option_override): Update accordingly.
14266 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14267 (xtensa_hard_regno_mode_ok): New function.
14268 * system.h (HARD_REGNO_MODE_OK): Poison.
14270 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14271 Alan Hayward <alan.hayward@arm.com>
14272 David Sherwood <david.sherwood@arm.com>
14274 * target.def (hard_regno_call_part_clobbered): New hook.
14275 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14276 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14277 * doc/tm.texi: Regenerate.
14278 * hooks.h (hook_bool_uint_mode_false): Declare.
14279 * hooks.c (hook_bool_uint_mode_false): New function.
14280 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14281 * cselib.c (cselib_process_insn): Use
14282 targetm.hard_regno_call_part_clobbered instead of
14283 HARD_REGNO_CALL_PART_CLOBBERED.
14284 * ira-conflicts.c (ira_build_conflicts): Likewise.
14285 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14286 * lra-constraints.c (need_for_call_save_p): Likewise.
14287 * lra-lives.c: Include target.h.
14288 (check_pseudos_live_through_calls): Use
14289 targetm.hard_regno_call_part_clobbered instead of
14290 HARD_REGNO_CALL_PART_CLOBBERED.
14291 * regcprop.c: Include target.h.
14292 (copyprop_hardreg_forward_1): Use
14293 targetm.hard_regno_call_part_clobbered instead of
14294 HARD_REGNO_CALL_PART_CLOBBERED.
14295 * reginfo.c (choose_hard_reg_mode): Likewise.
14296 * regrename.c (check_new_reg_p): Likewise.
14297 * reload.c (find_equiv_reg): Likewise.
14298 * reload1.c (emit_reload_insns): Likewise.
14299 * sched-deps.c (deps_analyze_insn): Likewise.
14300 * sel-sched.c (init_regs_for_mode): Likewise.
14301 (mark_unavailable_hard_regs): Likewise.
14302 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14303 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14304 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14306 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14307 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14308 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14310 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14312 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14313 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14314 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14316 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14317 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14318 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14320 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14321 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14323 * config/powerpcspe/powerpcspe.c
14324 (rs6000_hard_regno_call_part_clobbered): New function.
14325 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14326 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14327 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14329 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14330 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14331 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14333 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14334 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14335 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14337 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14338 Alan Hayward <alan.hayward@arm.com>
14339 David Sherwood <david.sherwood@arm.com>
14341 * rtl.h (subreg_memory_offset): Declare.
14342 * emit-rtl.c (subreg_memory_offset): New function.
14343 * expmed.c (store_bit_field_1): Use it.
14344 * expr.c (undefined_operand_subword_p): Likewise.
14345 * simplify-rtx.c (simplify_subreg): Likewise.
14347 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14349 PR rtl-optimization/57448
14352 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14353 using atomic_load pattern.
14354 (expand_atomic_store): Likewise.
14356 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14359 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14360 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14361 handling. Use replace_call_with_value with NULL instead of
14362 gsi_replace, unlink_stmt_vdef and release_defs.
14364 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14367 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14369 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14372 * rtl.h (emit_library_call_value_1): Declare.
14373 (emit_library_call): Replace declaration with a series of overloads.
14374 Remove the parameter count argument.
14375 (emit_library_call_value): Likewise.
14376 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14377 with an "rtx_mode_t *".
14378 (emit_library_call_value): Delete.
14379 (emit_library_call): Likewise.
14380 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14381 (asan_emit_allocas_unpoison): Likewise.
14382 * builtins.c (expand_builtin_powi): Likewise.
14383 (expand_asan_emit_allocas_unpoison): Likewise.
14384 * cfgexpand.c (expand_main_function): Likewise.
14385 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14386 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14387 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14388 * config/arm/arm.c (arm_trampoline_init): Likewise.
14389 (arm_call_tls_get_addr): Likewise.
14390 (arm_expand_divmod_libfunc): Likewise.
14391 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14392 (smulsi3_highpart): Likewise.
14393 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14394 (c6x_expand_compare): Likewise.
14395 (c6x_expand_movmem): Likewise.
14396 * config/frv/frv.c (frv_trampoline_init): Likewise.
14397 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14398 (ix86_expand_divmod_libfunc): Likewise.
14399 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14400 (ia64_expand_compare): Likewise.
14401 (ia64_profile_hook): Likewise.
14402 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14403 (nonlocal_goto): Likewise.
14404 (restore_stack_nonlocal): Likewise.
14405 * config/m32r/m32r.c (block_move_call): Likewise.
14406 (m32r_trampoline_init): Likewise.
14407 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14408 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14409 (m68k_call_m68k_read_tp): Likewise.
14410 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14411 (microblaze_expand_divide): Likewise.
14412 * config/mips/mips.h (mips_args): Likewise.
14413 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14414 (MIPS_ICACHE_SYNC): Likewise.
14415 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14416 (nios2_trampoline_init): Likewise.
14417 * config/pa/pa.c (hppa_tls_call): Likewise.
14418 (pa_trampoline_init): Likewise.
14419 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14420 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14421 (expand_strn_compare): Likewise.
14422 (rs6000_generate_compare): Likewise.
14423 (rs6000_expand_float128_convert): Likewise.
14424 (output_profile_hook): Likewise.
14425 (rs6000_trampoline_init): Likewise.
14426 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14427 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14428 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14429 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14430 (rs6000_generate_compare): Likewise.
14431 (rs6000_expand_float128_convert): Likewise.
14432 (output_profile_hook): Likewise.
14433 (rs6000_trampoline_init): Likewise.
14434 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14435 * config/sh/sh.c (sh_trampoline_init): Likewise.
14436 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14437 (sparc_emit_float_lib_cmp): Likewise.
14438 (sparc32_initialize_trampoline): Likewise.
14439 (sparc64_initialize_trampoline): Likewise.
14440 (sparc_profile_hook): Likewise.
14441 * config/spu/spu.c (ea_load_store): Likewise.
14442 * config/spu/spu.md (floatunssidf2): Likewise.
14443 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14444 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14445 * config/visium/visium.c (expand_block_move_4): Likewise.
14446 (expand_block_move_2): Likewise.
14447 (expand_block_move_1): Likewise.
14448 (expand_block_set_4): Likewise.
14449 (expand_block_set_2): Likewise.
14450 (expand_block_set_1): Likewise.
14451 (visium_trampoline_init): Likewise.
14452 (visium_profile_hook): Likewise.
14453 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14454 (xtensa_setup_frame_addresses): Likewise.
14455 (xtensa_trampoline_init): Likewise.
14456 * except.c (sjlj_emit_function_enter): Likewise.
14457 (sjlj_emit_function_exit): Likewise.
14458 * explow.c (allocate_dynamic_stack_space): Likewise.
14459 (probe_stack_range): Likewise.
14460 * expr.c (convert_mode_scalar): Likewise.
14461 * optabs.c (expand_binop): Likewise.
14462 (expand_twoval_binop_libfunc): Likewise.
14463 (expand_unop): Likewise.
14464 (prepare_cmp_insn): Likewise.
14465 (prepare_float_lib_cmp): Likewise.
14466 (expand_float): Likewise.
14467 (expand_fix): Likewise.
14468 (expand_fixed_convert): Likewise.
14469 (maybe_emit_sync_lock_test_and_set): Likewise.
14470 (expand_atomic_compare_and_swap): Likewise.
14471 (expand_mem_thread_fence): Likewise.
14472 (expand_atomic_fetch_op): Likewise.
14474 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14476 * doc/generic.texi (OpenACC): Adjust URL.
14477 * doc/invoke.texi (C Dialect Options): Ditto.
14479 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14481 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14482 predicate for operand 1. Add (m,<S>) constraint.
14483 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14484 Prevent memory operand 1 with register operand 2.
14486 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14488 PR rtl-optimization/82024
14489 * combine.c (try_combine): If the combination result is a PARALLEL,
14490 and we only need to retain the SET in there that would be placed
14491 at I2, check that we can place that at I3 instead, before doing so.
14493 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14496 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14498 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14499 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14502 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14503 Jeff Law <law@redhat.com>
14505 * varasm.c (bss_initializer_p): Do not put constants into .bss
14506 (categorize_decl_for_section): Handle bss_initializer_p returning
14507 false when DECL_INITIAL is NULL.
14509 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14512 * config/s390/s390.c (s390_can_inline_p): New function.
14514 2017-09-01 Jeff Law <law@redhat.com>
14516 PR tree-optimization/82052
14517 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14518 Always initialize the returned slot after a hash table miss
14519 when INSERT is true.
14521 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14523 * config/s390/s390.md (mem_signal_fence): Remove.
14524 * doc/md.texi (mem_signal_fence): Remove.
14525 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14527 * target-insns.def (mem_signal_fence): Remove.
14529 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14532 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14535 * asan.c (create_odr_indicator): Strip name encoding from assembler
14536 name before appending it after __odr_asan_.
14538 2017-09-01 Martin Liska <mliska@suse.cz>
14540 PR tree-optimization/82059
14541 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14542 frequency only when an edge is redirected.
14544 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14546 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14547 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14548 (arc_conditional_register_usage): Remove ARC600 lp_count
14550 (arc_file_start): Emit Tag_ARC_CPU_variation.
14551 (arc_can_use_doloop_p): New conditions to use ZOLs.
14552 (hwloop_fail): New function.
14553 (hwloop_optimize): Likewise.
14554 (hwloop_pattern_reg): Likewise.
14555 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14556 (arc_reorg_loops): New function, calls reorg_loops.
14557 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14558 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14560 (arc_loop_hazard): Remove function, functionality moved into
14562 (arc_hazard): Remove arc_loop_hazard call.
14563 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14564 into hwloop_optimize.
14565 (arc_label_align): Remove ZOL handling.
14566 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14567 * config/arc/arc.md (doloop_begin): Remove pattern.
14568 (doloop_begin_i): Likewise.
14569 (doloop_end_i): Likewise.
14570 (doloop_fallback): Likewise.
14571 (doloop_fallback_m): Likewise.
14572 (doloop_end): Reimplement expand.
14573 (arc_lp): New pattern for LP instruction.
14574 (loop_end): New pattern.
14575 (loop_fail): Likewise.
14576 (decrement_and_branch_until_zero): Likewise.
14577 * config/arc/arc.opt (mlpc-width): New option.
14578 * doc/invoke.texi (mlpc-width): Document option.
14580 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14582 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14583 (arc_ccfsm_advance): Fix checking for delay slots.
14584 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14586 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14588 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14590 (movhi_insn): Update store instruction constraint which are saving
14591 6-bit short immediates.
14592 (movsi_insn): Consider also short scaled load operations.
14593 (zero_extendhisi2_i): Use Usd constraint instead of T.
14594 (extendhisi2_i): Add q constraint.
14595 (arc_clzsi2): Add type and length attributes.
14596 (arc_ctzsi2): Likewise.
14597 * config/arc/constraints.md (Usc): Update constraint, the
14598 assembler can parse two relocations for a single instruction.
14600 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14602 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14603 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14605 2017-08-31 Olivier Hainque <hainque@adacore.com>
14607 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14608 match as powerpc-wrs-vxworks*.
14610 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14612 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14613 register constraint for by-element operand.
14614 (aarch64_mls_elt_merge<mode>): Likewise.
14616 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14618 * config/arc/arc.c (arc_can_follow_jump): Check for short
14621 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14623 * config.gcc: Use g.opt for arc.
14624 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14625 functionality moved to ...
14626 (legitimate_scaled_address_p): New function, ...here.
14627 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14628 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14629 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14631 (arc_override_options): Handle G option.
14632 (arc_output_pic_addr_const): Correct function definition.
14633 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14634 (arc_decl_anon_ns_mem_p): Delete.
14635 (arc_in_small_data_p): Overhaul this function to take into
14636 consideration the value given via G option.
14637 (arc_rewrite_small_data_1): Renamed and corrected old
14638 arc_rewrite_small_data function.
14639 (arc_rewrite_small_data): New function.
14640 (small_data_pattern): Don't use pic_offset_table_rtx.
14641 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14642 * config/arc/simdext.md (movmisalignv2hi): Use
14643 prepare_move_operands function.
14645 (movmisalign*): Likewise.
14646 * doc/invoke.texi (ARC options): Document -G option.
14648 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14650 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14652 * config/arc/arc.c (arc_print_operand): Output scalled address for
14653 sdata whenever is possible.
14654 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14655 load/stores are available.
14656 (compact_sda_memory_operand): Check for the alignment required by
14657 code density instructions.
14658 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14660 * config/arc/constraints.md (Usd): Update constraint.
14661 (Us0): New constraint.
14662 (Usc): Update constraint.
14664 2017-08-31 Richard Biener <rguenther@suse.de>
14666 PR middle-end/82054
14667 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14668 function only once.
14670 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14672 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14673 Resize type_signature.
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.c (aarch64_base_register_rtx_p): Only allow
14680 subregs whose inner modes can be stored in GPRs.
14681 (aarch64_classify_index): Likewise.
14683 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14684 Alan Hayward <alan.hayward@arm.com>
14685 David Sherwood <david.sherwood@arm.com>
14687 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14688 (V_INT_EQUIV): ...this.
14689 (v_cmp_result): Rename to...
14690 (v_int_equiv): ...this.
14691 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14692 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14693 (copysign<mode>3): Likewise.
14694 (aarch64_simd_bsl<mode>_internal): Likewise.
14695 (aarch64_simd_bsl<mode>): Likewise.
14696 (vec_cmp<mode><mode>): Likewise.
14697 (vcond<mode><mode>): Likewise.
14698 (vcond<v_cmp_mixed><mode>): Likewise.
14699 (vcondu<mode><v_cmp_mixed>): Likewise.
14700 (aarch64_cm<optab><mode>): Likewise.
14701 (aarch64_cmtst<mode>): Likewise.
14702 (aarch64_fac<optab><mode>): Likewise.
14703 (vec_perm_const<mode>): Likewise.
14704 (vcond_mask_<mode><v_cmp_result>): Rename to...
14705 (vcond_mask_<mode><v_int_equiv>): ...this.
14706 (vec_cmp<mode><v_cmp_result>): Rename to...
14707 (vec_cmp<mode><v_int_equiv>): ...this.
14709 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14710 Alan Hayward <alan.hayward@arm.com>
14711 David Sherwood <david.sherwood@arm.com>
14713 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14715 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14716 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14717 (UNSPEC_LD4_DREG): New unspecs.
14718 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14719 (aarch64_ld2<mode>_dreg_be): Replace with...
14720 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14722 (aarch64_ld3<mode>_dreg_le)
14723 (aarch64_ld3<mode>_dreg_be): Replace with...
14724 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14726 (aarch64_ld4<mode>_dreg_le)
14727 (aarch64_ld4<mode>_dreg_be): Replace with...
14728 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14731 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14733 PR tree-optimization/81987
14734 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14735 insert an initializer in a location not dominated by the stride
14738 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14740 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14741 on the entire header of the finally block in the fallthru case.
14743 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14745 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14747 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14749 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14750 rs6000_emit_move_from_cr and call renamed function.
14751 (rs6000_emit_prologue): Call renamed functions.
14752 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14753 movesi_from_cr, remove volatile CRs.
14755 2017-08-30 Jon Beniston <jon@beniston.com>
14756 Richard Biener <rguenther@suse.de>
14758 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14759 of VECTOR_MODE_P check.
14760 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14761 element vector types.
14763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14765 * df.h (df_read_modify_subreg_p): Remove in favor of...
14766 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14767 const_rtx instead of an rtx.
14768 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14769 * df-problems.c (df_word_lr_mark_ref): Likewise.
14770 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14771 (mark_pseudo_reg_dead): Likewise.
14772 (mark_ref_dead): Likewise.
14773 * reginfo.c (init_subregs_of_mode): Likewise.
14774 * sched-deps.c (sched_analyze_1): Likewise.
14775 * df-scan.c (df_def_record_1): Likewise.
14776 (df_uses_record): Likewise.
14777 (df_read_modify_subreg_p): Remove in favor of...
14778 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14779 const_rtx instead of an rtx.
14781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14782 Alan Hayward <alan.hayward@arm.com>
14783 David Sherwood <david.sherwood@arm.com>
14785 * rtl.h (partial_subreg_p): New function.
14786 * caller-save.c (save_call_clobbered_regs): Use it.
14787 * calls.c (expand_call): Likewise.
14788 * combine.c (combinable_i3pat): Likewise.
14789 (simplify_set): Likewise.
14790 (make_extraction): Likewise.
14791 (make_compound_operation_int): Likewise.
14792 (gen_lowpart_or_truncate): Likewise.
14793 (force_to_mode): Likewise.
14794 (make_field_assignment): Likewise.
14795 (reg_truncated_to_mode): Likewise.
14796 (record_truncated_value): Likewise.
14797 (move_deaths): Likewise.
14798 * cse.c (record_jump_cond): Likewise.
14799 (cse_insn): Likewise.
14800 * cselib.c (cselib_lookup_1): Likewise.
14801 * expmed.c (extract_bit_field_using_extv): Likewise.
14802 * function.c (assign_parm_setup_reg): Likewise.
14803 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14804 * ira-build.c (create_insn_allocnos): Likewise.
14805 * lra-coalesce.c (merge_pseudos): Likewise.
14806 * lra-constraints.c (match_reload): Likewise.
14807 (simplify_operand_subreg): Likewise.
14808 (curr_insn_transform): Likewise.
14809 * lra-lives.c (process_bb_lives): Likewise.
14810 * lra.c (new_insn_reg): Likewise.
14811 (lra_substitute_pseudo): Likewise.
14812 * regcprop.c (mode_change_ok): Likewise.
14813 (maybe_mode_change): Likewise.
14814 (copyprop_hardreg_forward_1): Likewise.
14815 * reload.c (push_reload): Likewise.
14816 (find_reloads): Likewise.
14817 (find_reloads_subreg_address): Likewise.
14818 * reload1.c (alter_reg): Likewise.
14819 (eliminate_regs_1): Likewise.
14820 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14822 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14824 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14825 back to if statements, including unpack.
14827 2017-08-30 Martin Liska <mliska@suse.cz>
14829 PR inline-asm/82001
14830 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14832 (func_checker::compare_asm_inputs_outputs): ... this function.
14833 (func_checker::compare_gimple_asm): Use the function to compare
14834 also ASM constrains.
14835 * ipa-icf-gimple.h: Rename the function.
14837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14838 Alan Hayward <alan.hayward@arm.com>
14839 David Sherwood <david.sherwood@arm.com>
14841 * coretypes.h (complex_mode): New type.
14842 * gdbhooks.py (build_pretty_printer): Handle it.
14843 * machmode.h (complex_mode): New class.
14844 (complex_mode::includes_p): New function.
14845 (is_complex_int_mode): Likewise.
14846 (is_complex_float_mode): Likewise.
14847 * genmodes.c (get_mode_class): Handle complex mode classes.
14848 * function.c (expand_function_end): Use is_complex_int_mode.
14850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14851 Alan Hayward <alan.hayward@arm.com>
14852 David Sherwood <david.sherwood@arm.com>
14854 * coretypes.h (scalar_mode_pod): New typedef.
14855 * gdbhooks.py (build_pretty_printer): Handle it.
14856 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14857 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14858 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14859 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14860 as_a <scalar_mode>.
14862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14863 Alan Hayward <alan.hayward@arm.com>
14864 David Sherwood <david.sherwood@arm.com>
14866 * machmode.h (mode_for_vector): Take a scalar_mode instead
14868 * stor-layout.c (mode_for_vector): Likewise.
14869 * explow.c (promote_mode): Use as_a <scalar_mode>.
14870 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14873 Alan Hayward <alan.hayward@arm.com>
14874 David Sherwood <david.sherwood@arm.com>
14876 * target.def (preferred_simd_mode): Take a scalar_mode
14877 instead of a machine_mode.
14878 * targhooks.h (default_preferred_simd_mode): Likewise.
14879 * targhooks.c (default_preferred_simd_mode): Likewise.
14880 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14881 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14882 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14883 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14884 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14885 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14886 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14887 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14889 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14890 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14891 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14892 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14893 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14894 * doc/tm.texi: Regenerate.
14895 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14899 Alan Hayward <alan.hayward@arm.com>
14900 David Sherwood <david.sherwood@arm.com>
14902 * target.def (scalar_mode_supported_p): Take a scalar_mode
14903 instead of a machine_mode.
14904 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14905 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14906 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14907 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14908 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14909 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14910 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14911 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14912 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14913 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14914 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14915 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14916 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14917 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14919 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14920 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14921 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14922 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14923 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14925 * doc/tm.texi: Regenerate.
14927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14928 Alan Hayward <alan.hayward@arm.com>
14929 David Sherwood <david.sherwood@arm.com>
14931 * coretypes.h (opt_scalar_mode): New typedef.
14932 * gdbhooks.py (build_pretty_printers): Handle it.
14933 * machmode.h (mode_iterator::get_2xwider): Add overload for
14935 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14937 * expr.c (convert_mode_scalar): Likewise.
14938 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14939 * optabs.c (expand_float): Likewise.
14940 (expand_fix): Likewise.
14941 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14944 Alan Hayward <alan.hayward@arm.com>
14945 David Sherwood <david.sherwood@arm.com>
14947 * optabs.c (expand_float): Explicitly check for scalars before
14948 using a branching expansion.
14949 (expand_fix): Likewise.
14951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14952 Alan Hayward <alan.hayward@arm.com>
14953 David Sherwood <david.sherwood@arm.com>
14955 * expr.c (convert_mode): Split scalar handling out into...
14956 (convert_mode_scalar): ...this new function. Treat the modes
14959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14960 Alan Hayward <alan.hayward@arm.com>
14961 David Sherwood <david.sherwood@arm.com>
14963 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14965 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14968 Alan Hayward <alan.hayward@arm.com>
14969 David Sherwood <david.sherwood@arm.com>
14971 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14972 rather than a machine_mode.
14973 (fixed_from_string): Likewise.
14974 (fixed_convert): Likewise.
14975 (fixed_convert_from_int): Likewise.
14976 (fixed_convert_from_real): Likewise.
14977 (real_convert_from_fixed): Likewise.
14978 * fixed-value.c (fixed_from_double_int): Likewise.
14979 (fixed_from_string): Likewise.
14980 (fixed_convert): Likewise.
14981 (fixed_convert_from_int): Likewise.
14982 (fixed_convert_from_real): Likewise.
14983 (real_convert_from_fixed): Likewise.
14984 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14987 Alan Hayward <alan.hayward@arm.com>
14988 David Sherwood <david.sherwood@arm.com>
14990 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14991 of separate mode class checks. Do not allow vector modes here.
14992 (immed_wide_int_const): Use as_a <scalar_mode>.
14993 * explow.c (trunc_int_for_mode): Likewise.
14994 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14995 (wi::shwi): Likewise.
14996 (wi::min_value): Likewise.
14997 (wi::max_value): Likewise.
14998 * dwarf2out.c (loc_descriptor): Likewise.
14999 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15000 for CONST_WIDE_INT.
15002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15003 Alan Hayward <alan.hayward@arm.com>
15004 David Sherwood <david.sherwood@arm.com>
15006 * tree.h (SCALAR_TYPE_MODE): New macro.
15007 * expr.c (expand_expr_addr_expr_1): Use it.
15008 (expand_expr_real_2): Likewise.
15009 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15010 (fold_convert_const_fixed_from_int): Likewise.
15011 (fold_convert_const_fixed_from_real): Likewise.
15012 (native_encode_fixed): Likewise
15013 (native_encode_complex): Likewise
15014 (native_encode_vector): Likewise.
15015 (native_interpret_fixed): Likewise.
15016 (native_interpret_real): Likewise.
15017 (native_interpret_complex): Likewise.
15018 (native_interpret_vector): Likewise.
15019 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15020 (simd_clone_adjust_argument_types): Likewise.
15021 (simd_clone_init_simd_arrays): Likewise.
15022 (simd_clone_adjust): Likewise.
15023 * stor-layout.c (layout_type): Likewise.
15024 * tree.c (build_minus_one_cst): Likewise.
15025 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15026 * tree-inline.c (estimate_move_cost): Likewise.
15027 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15028 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15029 (vectorizable_reduction): Likewise.
15030 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15031 (vect_recog_mixed_size_cond_pattern): Likewise.
15032 (check_bool_pattern): Likewise.
15033 (adjust_bool_pattern): Likewise.
15034 (search_type_for_mask_1): Likewise.
15035 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15036 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15037 (vectorizable_load): Likewise.
15038 (vectorizable_store): Likewise.
15039 * ubsan.c (ubsan_encode_value): Likewise.
15040 * varasm.c (output_constant): Likewise.
15042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15043 Alan Hayward <alan.hayward@arm.com>
15044 David Sherwood <david.sherwood@arm.com>
15046 * coretypes.h (scalar_mode): New class.
15047 * machmode.h (scalar_mode): Likewise.
15048 (scalar_mode::includes_p): New function.
15049 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15050 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15051 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15052 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15053 * expmed.c (store_bit_field_1): Likewise.
15054 (extract_bit_field_1): Likewise.
15055 * expr.c (write_complex_part): Likewise.
15056 (read_complex_part): Likewise.
15057 (emit_move_complex_push): Likewise.
15058 (expand_expr_real_2): Likewise.
15059 * function.c (assign_parm_setup_reg): Likewise.
15060 (assign_parms_unsplit_complex): Likewise.
15061 * optabs.c (expand_binop): Likewise.
15062 * rtlanal.c (subreg_get_info): Likewise.
15063 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15064 * varasm.c (output_constant_pool_2): Likewise.
15066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15067 Alan Hayward <alan.hayward@arm.com>
15068 David Sherwood <david.sherwood@arm.com>
15070 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15072 (expmed_mult_highpart_optab): Likewise.
15073 (expmed_mult_highpart): Likewise.
15075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15076 Alan Hayward <alan.hayward@arm.com>
15077 David Sherwood <david.sherwood@arm.com>
15079 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15080 instead of a machine_mode.
15081 (builtin_memset_read_str): Likewise.
15082 * builtins.c (c_readstr): Likewise.
15083 (builtin_memcpy_read_str): Likewise.
15084 (builtin_strncpy_read_str): Likewise.
15085 (builtin_memset_read_str): Likewise.
15086 (builtin_memset_gen_str): Likewise.
15087 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15088 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15089 instead of a machine_mode.
15090 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15092 (make_extraction): Likewise.
15093 (try_widen_shift_mode): Take and return scalar_int_modes instead
15095 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15096 a scalar_int_mode instead of a machine_mode.
15097 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15098 (avr_addr_space_pointer_mode): Likewise.
15099 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15100 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15101 (msp430_unwind_word_mode): Likewise.
15102 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15103 (spu_addr_space_pointer_mode): Likewise.
15104 (spu_addr_space_address_mode): Likewise.
15105 (spu_libgcc_cmp_return_mode): Likewise.
15106 (spu_libgcc_shift_count_mode): Likewise.
15107 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15108 (rl78_addr_space_pointer_mode): Likewise.
15109 (fl78_unwind_word_mode): Likewise.
15110 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15112 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15113 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15114 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15115 (mips_valid_pointer_mode): Likewise.
15116 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15117 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15118 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15120 (ft32_addr_space_address_mode): Likewise.
15121 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15122 scalar_int_mode instead of a machine_mode.
15123 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15125 (m32c_addr_space_address_mode): Likewise.
15126 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15127 (rs6000_eh_return_filter_mode): Likewise.
15128 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15129 (rs6000_eh_return_filter_mode): Likewise.
15130 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15131 (s390_libgcc_shift_count_mode): Likewise.
15132 (s390_unwind_word_mode): Likewise.
15133 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15135 * target.def (mode_rep_extended): Likewise.
15136 (valid_pointer_mode): Likewise.
15137 (addr_space.valid_pointer_mode): Likewise.
15138 (eh_return_filter_mode): Return a scalar_int_mode rather than
15140 (libgcc_cmp_return_mode): Likewise.
15141 (libgcc_shift_count_mode): Likewise.
15142 (unwind_word_mode): Likewise.
15143 (addr_space.pointer_mode): Likewise.
15144 (addr_space.address_mode): Likewise.
15145 * doc/tm.texi: Regenerate.
15146 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15148 (do_jump): Use scalar_int_mode for local variables.
15149 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15150 rather than a machine_mode.
15151 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15152 (scompare_loc_descriptor_wide): Likewise.
15153 (scompare_loc_descriptor_narrow): Likewise.
15154 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15156 * except.c (sjlj_emit_dispatch_table): Likewise.
15157 (expand_builtin_eh_copy_values): Likewise.
15158 * explow.c (convert_memory_address_addr_space_1): Likewise.
15159 Take a scalar_int_mode rather than a machine_mode.
15160 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15161 than a machine_mode.
15162 (memory_address_addr_space): Use scalar_int_mode for local variables.
15163 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15164 rather than a machine_mode.
15165 * expmed.c (mask_rtx): Likewise.
15166 (init_expmed_one_conv): Likewise.
15167 (expand_mult_highpart_adjust): Likewise.
15168 (extract_high_half): Likewise.
15169 (expmed_mult_highpart_optab): Likewise.
15170 (expmed_mult_highpart): Likewise.
15171 (expand_smod_pow2): Likewise.
15172 (expand_sdiv_pow2): Likewise.
15173 (emit_store_flag_int): Likewise.
15174 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15176 (extract_low_bits): Likewise.
15177 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15179 * expr.c (pieces_addr::adjust): Likewise.
15180 (can_store_by_pieces): Likewise.
15181 (store_by_pieces): Likewise.
15182 (clear_by_pieces_1): Likewise.
15183 (expand_expr_addr_expr_1): Likewise.
15184 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15185 (expand_expr_real_1): Likewise.
15186 (try_casesi): Likewise.
15187 * final.c (shorten_branches): Likewise.
15188 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15189 type of "mode" to machine_mode.
15190 * internal-fn.c (expand_arith_overflow_result_store): Take a
15191 scalar_int_mode rather than a machine_mode.
15192 (expand_mul_overflow): Use scalar_int_mode for local variables.
15193 * loop-doloop.c (doloop_modify): Likewise.
15194 (doloop_optimize): Likewise.
15195 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15196 than a machine_mode.
15197 (expand_doubleword_shift_condmove): Likewise.
15198 (expand_doubleword_shift): Likewise.
15199 (expand_doubleword_clz): Likewise.
15200 (expand_doubleword_popcount): Likewise.
15201 (expand_doubleword_parity): Likewise.
15202 (expand_absneg_bit): Use scalar_int_mode for local variables.
15203 (prepare_float_lib_cmp): Likewise.
15204 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15205 rather than a machine_mode.
15206 (convert_memory_address_addr_space): Likewise.
15207 (get_mode_bounds): Likewise.
15208 (get_address_mode): Return a scalar_int_mode rather than a
15210 * rtlanal.c (get_address_mode): Likewise.
15211 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15212 than a machine_mode.
15213 * targhooks.c (default_mode_rep_extended): Likewise.
15214 (default_valid_pointer_mode): Likewise.
15215 (default_addr_space_valid_pointer_mode): Likewise.
15216 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15217 than a machine_mode.
15218 (default_libgcc_cmp_return_mode): Likewise.
15219 (default_libgcc_shift_count_mode): Likewise.
15220 (default_unwind_word_mode): Likewise.
15221 (default_addr_space_pointer_mode): Likewise.
15222 (default_addr_space_address_mode): Likewise.
15223 * targhooks.h (default_eh_return_filter_mode): Likewise.
15224 (default_libgcc_cmp_return_mode): Likewise.
15225 (default_libgcc_shift_count_mode): Likewise.
15226 (default_unwind_word_mode): Likewise.
15227 (default_addr_space_pointer_mode): Likewise.
15228 (default_addr_space_address_mode): Likewise.
15229 (default_mode_rep_extended): Take a scalar_int_mode rather than
15231 (default_valid_pointer_mode): Likewise.
15232 (default_addr_space_valid_pointer_mode): Likewise.
15233 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15235 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15236 rather than a machine_mode.
15237 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15238 for local variables.
15239 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15240 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15241 than a machine_mode.
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 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15248 the mode argument to scalar_int_mode.
15249 (do_jump_by_parts_zero_rtx): Likewise.
15250 (do_jump_by_parts_equality_rtx): Likewise.
15251 (do_jump_by_parts_greater): Take a mode argument.
15252 (do_jump_by_parts_equality): Likewise.
15253 (do_jump_1): Update calls accordingly.
15255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15256 Alan Hayward <alan.hayward@arm.com>
15257 David Sherwood <david.sherwood@arm.com>
15259 * is-a.h (safe_dyn_cast): New function.
15260 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15261 (jump_table_for_label): Likewise.
15262 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15263 instead of an rtx_insn *.
15264 (shorten_branches): Use dyn_cast instead of LABEL_P and
15265 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15266 rtx_jump_table_data::get_data_mode.
15267 (final_scan_insn): Likewise.
15269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15270 Alan Hayward <alan.hayward@arm.com>
15271 David Sherwood <david.sherwood@arm.com>
15273 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15274 trying to combine a full-register integer set with a subreg
15277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15278 Alan Hayward <alan.hayward@arm.com>
15279 David Sherwood <david.sherwood@arm.com>
15281 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15282 that is always either address_mode or pointer_mode.
15284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15285 Alan Hayward <alan.hayward@arm.com>
15286 David Sherwood <david.sherwood@arm.com>
15288 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15289 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
15296 is_a <scalar_int_mode> instead of checking for a nonzero
15297 precision. Forcibly convert op_mode to a scalar_int_mode
15298 in that case. More clearly differentiate the operand and
15299 result modes and use the former when deciding what the value
15300 of a count-bits operation should be. Use is_int_mode instead
15301 of checking for a MODE_INT. Remove redundant check for whether
15302 this mode has a zero precision.
15304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15305 Alan Hayward <alan.hayward@arm.com>
15306 David Sherwood <david.sherwood@arm.com>
15308 * optabs.c (widen_leading): Change the type of the mode argument
15309 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15310 (widen_bswap): Likewise.
15311 (expand_parity): Likewise.
15312 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15313 (expand_ffs): Likewise.
15314 (epand_unop): Check for scalar integer modes before calling the
15317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15318 Alan Hayward <alan.hayward@arm.com>
15319 David Sherwood <david.sherwood@arm.com>
15321 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15323 (expand_expr_real_1): Update call accordingly.
15325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15326 Alan Hayward <alan.hayward@arm.com>
15327 David Sherwood <david.sherwood@arm.com>
15329 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15330 value_mode arguments. Use scalar_int_mode internally.
15331 (store_bit_field_1): Rename the new integer mode from imode
15332 to op0_mode and use it instead of GET_MODE (op0). Update calls
15333 to store_split_bit_field, store_bit_field_using_insv and
15334 store_fixed_bit_field.
15335 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15336 Use scalar_int_mode internally. Use a bit count rather than a mode
15337 when calculating the largest bit size for get_best_mode.
15338 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15339 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15340 Remove assertion that OP0 has a scalar integer mode.
15341 (store_split_bit_field): Add op0_mode and value_mode arguments.
15342 Update calls to extract_fixed_bit_field.
15343 (extract_bit_field_using_extv): Add an op0_mode argument.
15344 Use scalar_int_mode internally.
15345 (extract_bit_field_1): Rename the new integer mode from imode to
15346 op0_mode and use it instead of GET_MODE (op0). Update calls to
15347 extract_split_bit_field, extract_bit_field_using_extv and
15348 extract_fixed_bit_field.
15349 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15350 to extract_split_bit_field and extract_fixed_bit_field_1.
15351 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15352 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15353 on the target mode.
15354 (extract_split_bit_field): Add an op0_mode argument. Update call
15355 to extract_fixed_bit_field.
15357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15358 Alan Hayward <alan.hayward@arm.com>
15359 David Sherwood <david.sherwood@arm.com>
15361 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15362 * explow.c (hard_function_value): Likewise.
15363 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15364 convert_to_mode call outside the loop.
15365 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15366 for the mode iterator. Require the mode specified by max_pieces
15368 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15370 (copy_blkmode_to_reg): Likewise.
15371 (set_storage_via_setmem): Likewise.
15372 * optabs.c (prepare_cmp_insn): Likewise.
15373 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15374 * stor-layout.c (finish_bitfield_representative): Likewise.
15376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15377 Alan Hayward <alan.hayward@arm.com>
15378 David Sherwood <david.sherwood@arm.com>
15380 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15381 * expr.c (convert_move): Use them.
15382 (convert_modes): Likewise.
15383 (store_expr_with_bounds): Likewise.
15385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15386 Alan Hayward <alan.hayward@arm.com>
15387 David Sherwood <david.sherwood@arm.com>
15389 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15390 parameter for the mode of "x". Remove the "known_x", "known_mode"
15391 and "known_ret" arguments. Change the type of the mode argument
15392 to scalar_int_mode.
15393 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15394 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15395 (reg_num_sign_bit_copies_for_combine): Likewise.
15396 * rtlanal.c (nonzero_bits1): Likewise.
15397 (num_sign_bit_copies1): Likewise.
15398 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15399 (reg_num_sign_bit_copies_general): Likewise.
15400 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15401 (reg_nonzero_bits_general): Likewise.
15403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15404 Alan Hayward <alan.hayward@arm.com>
15405 David Sherwood <david.sherwood@arm.com>
15407 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15408 than in subroutines. Return 1 for non-integer modes.
15409 (cached_num_sign_bit_copies): Change the type of the mode parameter
15410 to scalar_int_mode.
15411 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15412 classes. Handle CONST_INT_P first and then check whether X also
15413 has a scalar integer mode. Check the same thing for inner registers
15414 of a SUBREG and for values that are being extended or truncated.
15416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15417 Alan Hayward <alan.hayward@arm.com>
15418 David Sherwood <david.sherwood@arm.com>
15420 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15421 in subroutines. Return the mode mask for non-integer modes.
15422 (cached_nonzero_bits): Change the type of the mode parameter
15423 to scalar_int_mode.
15424 (nonzero_bits1): Likewise. Remove early exit for other mode
15425 classes. Handle CONST_INT_P first and then check whether X
15426 also has a scalar integer mode.
15428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15429 Alan Hayward <alan.hayward@arm.com>
15430 David Sherwood <david.sherwood@arm.com>
15432 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15433 Return a scalar_int_mode and assert that the size is greater than
15435 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15436 (op_by_pieces_d::op_by_pieces_d): Likewise.
15437 (op_by_pieces_d::run): Likewise.
15438 (can_store_by_pieces): Likewise.
15440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15441 Alan Hayward <alan.hayward@arm.com>
15442 David Sherwood <david.sherwood@arm.com>
15444 * combine.c (extract_left_shift): Add a mode argument and update
15446 (make_compound_operation_int): Change the type of the mode parameter
15447 to scalar_int_mode and update the call to extract_left_shift.
15449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15450 Alan Hayward <alan.hayward@arm.com>
15451 David Sherwood <david.sherwood@arm.com>
15453 * combine.c (simplify_and_const_int): Change the type of the mode
15454 parameter to scalar_int_mode.
15455 (simplify_and_const_int_1): Likewise. Update recursive call.
15457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15458 Alan Hayward <alan.hayward@arm.com>
15459 David Sherwood <david.sherwood@arm.com>
15461 * combine.c (simplify_compare_const): Check that the mode is a
15462 scalar_int_mode (rather than VOIDmode) before testing its
15464 (simplify_comparison): Move COMPARISON_P handling out of the
15465 loop and restrict the latter part of the loop to scalar_int_modes.
15466 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15467 and when considering SUBREG_REGs. Use is_int_mode instead of
15468 checking GET_MODE_CLASS against MODE_INT.
15470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15471 Alan Hayward <alan.hayward@arm.com>
15472 David Sherwood <david.sherwood@arm.com>
15474 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15475 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15476 shift_unit_mode and for modes involved in scalar shifts.
15478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15479 Alan Hayward <alan.hayward@arm.com>
15480 David Sherwood <david.sherwood@arm.com>
15482 * combine.c (force_int_to_mode): New function, split out from...
15483 (force_to_mode): ...here. Keep xmode up-to-date and use it
15484 instead of GET_MODE (x).
15486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15487 Alan Hayward <alan.hayward@arm.com>
15488 David Sherwood <david.sherwood@arm.com>
15490 * optabs-query.h (extraction_insn::struct_mode): Change type to
15491 opt_scalar_int_mode and update comment.
15492 (extraction_insn::field_mode): Change type to scalar_int_mode.
15493 (extraction_insn::pos_mode): Likewise.
15494 * combine.c (make_extraction): Update accordingly.
15495 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15496 (get_optab_extraction_insn): Likewise.
15497 * recog.c (simplify_while_replacing): Likewise.
15498 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15499 parameter to opt_scalar_int_mode.
15501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15502 Alan Hayward <alan.hayward@arm.com>
15503 David Sherwood <david.sherwood@arm.com>
15505 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15506 to a scalar_int_mode instead of a machine_mode.
15507 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15508 (get_best_mode): Return a boolean and use a pointer argument to store
15509 the selected mode. Replace the limit mode parameter with a bit limit.
15510 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15511 for the values returned by bit_field_mode_iterator::next_mode.
15512 (store_bit_field): Update call to get_best_mode.
15513 (store_fixed_bit_field): Likewise.
15514 (extract_fixed_bit_field): Likewise.
15515 * expr.c (optimize_bitfield_assignment_op): Likewise.
15516 * fold-const.c (optimize_bit_field_compare): Likewise.
15517 (fold_truth_andor_1): Likewise.
15518 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15519 Update for new type of m_mode.
15520 (get_best_mode): As above.
15522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15523 Alan Hayward <alan.hayward@arm.com>
15524 David Sherwood <david.sherwood@arm.com>
15526 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15527 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15528 (store_bit_field): Check is_a <scalar_int_mode> before calling
15529 strict_volatile_bitfield_p.
15530 (extract_bit_field): Likewise.
15532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15533 Alan Hayward <alan.hayward@arm.com>
15534 David Sherwood <david.sherwood@arm.com>
15536 * target.def (cstore_mode): Return a scalar_int_mode.
15537 * doc/tm.texi: Regenerate.
15538 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15539 * targhooks.h (default_cstore_mode): Likewise.
15540 * targhooks.c (default_cstore_mode): Likewise, using a forced
15542 * expmed.c (emit_cstore): Expect the target of the cstore to be
15545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15546 Alan Hayward <alan.hayward@arm.com>
15547 David Sherwood <david.sherwood@arm.com>
15549 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15551 (niter_desc): Likewise mode.
15552 (iv_analyze): Add a mode parameter.
15554 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15555 and change its type to scalar_int_mode.
15556 * loop-iv.c: Update commentary at head of file.
15557 (iv_constant): Pass the mode paraeter before the rtx it describes
15558 and change its type to scalar_int_mode. Remove VOIDmode handling.
15559 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15560 (iv_extend): Likewise.
15561 (shorten_into_mode): Likewise.
15562 (iv_add): Use scalar_int_mode.
15563 (iv_mult): Likewise.
15564 (iv_shift): Likewise.
15565 (canonicalize_iv_subregs): Likewise.
15566 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15567 it describes and change its mode to scalar_int_mode. Also change
15568 the type of the returned inner_mode to scalar_int_mode.
15569 (get_biv_step): Likewise, turning outer_mode from a pointer
15570 into a direct parameter. Update call to get_biv_step_1.
15571 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15572 iv_constant and get_biv_step.
15573 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15574 and change its type to scalar_int_mode. Don't initialise iv->mode
15575 to VOIDmode and remove later checks for its still being VOIDmode.
15576 Update calls to iv_analyze_op and iv_analyze_expr. Check
15577 is_a <scalar_int_mode> when changing the mode under consideration.
15578 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15579 Update call to iv_analyze_expr.
15580 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15581 inner register is not also a scalar_int_mode. Update call to
15583 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15584 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15585 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15586 separate mode class checks. Update calls to iv_analyze. Remove
15587 fix-up of VOIDmodes after iv_analyze_biv.
15588 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15589 don't have a scalar_int_mode. Update call to biv_p.
15591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15592 Alan Hayward <alan.hayward@arm.com>
15593 David Sherwood <david.sherwood@arm.com>
15595 * cfgexpand.c (convert_debug_memory_address): Use
15596 as_a <scalar_int_mode>.
15597 * combine.c (expand_compound_operation): Likewise.
15598 (make_extraction): Likewise.
15599 (change_zero_ext): Likewise.
15600 (simplify_comparison): Likewise.
15601 * cse.c (cse_insn): Likewise.
15602 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15603 (mem_loc_descriptor): Likewise.
15604 (loc_descriptor): Likewise.
15605 * expmed.c (init_expmed_one_mode): Likewise.
15606 (synth_mult): Likewise.
15607 (emit_store_flag_1): Likewise.
15608 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15609 of a comparison with size.
15610 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15611 (reduce_to_bit_field_precision): Likewise.
15612 * function.c (expand_function_end): Likewise.
15613 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15614 * loop-doloop.c (doloop_modify): Likewise.
15615 * optabs.c (expand_binop): Likewise.
15616 (expand_unop): Likewise.
15617 (expand_copysign_absneg): Likewise.
15618 (prepare_cmp_insn): Likewise.
15619 (maybe_legitimize_operand): Likewise.
15620 * recog.c (const_scalar_int_operand): Likewise.
15621 * rtlanal.c (get_address_mode): Likewise.
15622 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15623 (simplify_cond_clz_ctz): Likewise.
15624 * tree-nested.c (get_nl_goto_field): Likewise.
15625 * tree.c (build_vector_type_for_mode): Likewise.
15626 * var-tracking.c (use_narrower_mode): Likewise.
15628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15629 Alan Hayward <alan.hayward@arm.com>
15630 David Sherwood <david.sherwood@arm.com>
15632 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15633 * builtins.c (expand_builtin_signbit): Use it.
15634 * cfgexpand.c (expand_debug_expr): Likewise.
15635 * dojump.c (do_jump): Likewise.
15636 (do_compare_and_jump): Likewise.
15637 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15638 * expmed.c (make_tree): Likewise.
15639 * expr.c (expand_expr_real_2): Likewise.
15640 (expand_expr_real_1): Likewise.
15641 (try_casesi): Likewise.
15642 * fold-const-call.c (fold_const_call_ss): Likewise.
15643 * fold-const.c (unextend): Likewise.
15644 (extract_muldiv_1): Likewise.
15645 (fold_single_bit_test): Likewise.
15646 (native_encode_int): Likewise.
15647 (native_encode_string): Likewise.
15648 (native_interpret_int): Likewise.
15649 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15650 * internal-fn.c (expand_addsub_overflow): Likewise.
15651 (expand_neg_overflow): Likewise.
15652 (expand_mul_overflow): Likewise.
15653 (expand_arith_overflow): Likewise.
15654 * match.pd: Likewise.
15655 * stor-layout.c (layout_type): Likewise.
15656 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15657 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15658 * tree-ssanames.c (get_range_info): Likewise.
15659 * tree-switch-conversion.c (array_value_type) Likewise.
15660 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15661 (vect_recog_divmod_pattern): Likewise.
15662 (vect_recog_mixed_size_cond_pattern): Likewise.
15663 * tree-vrp.c (extract_range_basic): Likewise.
15664 (simplify_float_conversion_using_ranges): Likewise.
15665 * tree.c (int_fits_type_p): Likewise.
15666 * ubsan.c (instrument_bool_enum_load): Likewise.
15667 * varasm.c (mergeable_string_section): Likewise.
15668 (narrowing_initializer_constant_valid_p): Likewise.
15669 (output_constant): Likewise.
15671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15672 Alan Hayward <alan.hayward@arm.com>
15673 David Sherwood <david.sherwood@arm.com>
15675 * machmode.h (NARROWEST_INT_MODE): New macro.
15676 * expr.c (alignment_for_piecewise_move): Use it instead of
15677 GET_CLASS_NARROWEST_MODE (MODE_INT).
15678 (push_block): Likewise.
15679 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15681 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15684 Alan Hayward <alan.hayward@arm.com>
15685 David Sherwood <david.sherwood@arm.com>
15687 * postreload.c (move2add_valid_value_p): Change the type of the
15688 mode parameter to scalar_int_mode.
15689 (move2add_use_add2_insn): Add a mode parameter and use it instead
15691 (move2add_use_add3_insn): Likewise.
15692 (reload_cse_move2add): Update accordingly.
15694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15695 Alan Hayward <alan.hayward@arm.com>
15696 David Sherwood <david.sherwood@arm.com>
15698 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15700 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15701 * optabs.c (expand_unop): Likewise.
15703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15704 Alan Hayward <alan.hayward@arm.com>
15705 David Sherwood <david.sherwood@arm.com>
15707 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15708 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15709 (popcount_loc_descriptor): Likewise.
15710 (bswap_loc_descriptor): Likewise.
15711 (rotate_loc_descriptor): Likewise.
15712 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15713 calling the functions above.
15715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15716 Alan Hayward <alan.hayward@arm.com>
15717 David Sherwood <david.sherwood@arm.com>
15719 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15721 (try_combine): Likewise.
15722 (simplify_if_then_else): Likewise.
15723 * cse.c (cse_insn): Likewise.
15724 * dwarf2out.c (mem_loc_descriptor): Likewise.
15725 * emit-rtl.c (gen_lowpart_common): Likewise.
15726 * simplify-rtx.c (simplify_truncation): Likewise.
15727 (simplify_binary_operation_1): Likewise.
15728 (simplify_const_relational_operation): Likewise.
15729 (simplify_ternary_operation): Likewise.
15730 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15733 Alan Hayward <alan.hayward@arm.com>
15734 David Sherwood <david.sherwood@arm.com>
15736 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15737 * reload.c (push_reload): Likewise.
15738 (find_reloads): Likewise.
15740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15741 Alan Hayward <alan.hayward@arm.com>
15742 David Sherwood <david.sherwood@arm.com>
15744 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15745 (make_compound_operation_int): Likewise.
15746 (change_zero_ext): Likewise.
15747 * expr.c (convert_move): Likewise.
15748 (convert_modes): Likewise.
15749 * fwprop.c (forward_propagate_subreg): Likewise.
15750 * loop-iv.c (get_biv_step_1): Likewise.
15751 * optabs.c (widen_operand): Likewise.
15752 * postreload.c (move2add_valid_value_p): Likewise.
15753 * recog.c (simplify_while_replacing): Likewise.
15754 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15755 (simplify_binary_operation_1): Likewise. Remove redundant
15756 mode equality check.
15758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15759 Alan Hayward <alan.hayward@arm.com>
15760 David Sherwood <david.sherwood@arm.com>
15762 * combine.c (combine_simplify_rtx): Add checks for
15763 is_a <scalar_int_mode>.
15764 (simplify_if_then_else): Likewise.
15765 (make_field_assignment): Likewise.
15766 (simplify_comparison): Likewise.
15767 * ifcvt.c (noce_try_bitop): Likewise.
15768 * loop-invariant.c (canonicalize_address_mult): Likewise.
15769 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15772 Alan Hayward <alan.hayward@arm.com>
15773 David Sherwood <david.sherwood@arm.com>
15775 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15776 is_a <scalar_int_mode> instead of != BLKmode.
15778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15779 Alan Hayward <alan.hayward@arm.com>
15780 David Sherwood <david.sherwood@arm.com>
15782 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15783 instead of != VOIDmode.
15784 * combine.c (if_then_else_cond): Likewise.
15785 (change_zero_ext): Likewise.
15786 * dwarf2out.c (mem_loc_descriptor): Likewise.
15787 (loc_descriptor): Likewise.
15788 * rtlanal.c (canonicalize_condition): Likewise.
15789 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15792 Alan Hayward <alan.hayward@arm.com>
15793 David Sherwood <david.sherwood@arm.com>
15795 * simplify-rtx.c (simplify_binary_operation_1): Use
15796 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15799 Alan Hayward <alan.hayward@arm.com>
15800 David Sherwood <david.sherwood@arm.com>
15802 * wide-int.h (int_traits<unsigned char>) New class.
15803 (int_traits<unsigned short>) Likewise.
15804 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15805 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15807 * combine.c (set_nonzero_bits_and_sign_copies): Use
15808 is_a <scalar_int_mode>.
15809 (find_split_point): Likewise.
15810 (combine_simplify_rtx): Likewise.
15811 (simplify_logical): Likewise.
15812 (expand_compound_operation): Likewise.
15813 (expand_field_assignment): Likewise.
15814 (make_compound_operation): Likewise.
15815 (extended_count): Likewise.
15816 (change_zero_ext): Likewise.
15817 (simplify_comparison): Likewise.
15818 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15819 (ucompare_loc_descriptor): Likewise.
15820 (minmax_loc_descriptor): Likewise.
15821 (mem_loc_descriptor): Likewise.
15822 (loc_descriptor): Likewise.
15823 * expmed.c (init_expmed_one_mode): Likewise.
15824 * lra-constraints.c (lra_constraint_offset): Likewise.
15825 * optabs.c (prepare_libcall_arg): Likewise.
15826 * postreload.c (move2add_note_store): Likewise.
15827 * reload.c (operands_match_p): Likewise.
15828 * rtl.h (load_extend_op): Likewise.
15829 * rtlhooks.c (gen_lowpart_general): Likewise.
15830 * simplify-rtx.c (simplify_truncation): Likewise.
15831 (simplify_unary_operation_1): Likewise.
15832 (simplify_binary_operation_1): Likewise.
15833 (simplify_const_binary_operation): Likewise.
15834 (simplify_const_relational_operation): Likewise.
15835 (simplify_subreg): Likewise.
15836 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15837 * var-tracking.c (adjust_mems): Likewise.
15838 (prepare_call_arguments): Likewise.
15840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15841 Alan Hayward <alan.hayward@arm.com>
15842 David Sherwood <david.sherwood@arm.com>
15844 * machmode.h (is_int_mode): New fuction.
15845 * combine.c (find_split_point): Use it.
15846 (combine_simplify_rtx): Likewise.
15847 (simplify_if_then_else): Likewise.
15848 (simplify_set): Likewise.
15849 (simplify_shift_const_1): Likewise.
15850 (simplify_comparison): Likewise.
15851 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15852 * cse.c (notreg_cost): Likewise.
15853 (cse_insn): Likewise.
15854 * cselib.c (cselib_lookup_1): Likewise.
15855 * dojump.c (do_jump_1): Likewise.
15856 (do_compare_rtx_and_jump): Likewise.
15857 * dse.c (get_call_args): Likewise.
15858 * dwarf2out.c (rtl_for_decl_init): Likewise.
15859 (native_encode_initializer): Likewise.
15860 * expmed.c (emit_store_flag_1): Likewise.
15861 (emit_store_flag): Likewise.
15862 * expr.c (convert_modes): Likewise.
15863 (store_field): Likewise.
15864 (expand_expr_real_1): Likewise.
15865 * fold-const.c (fold_read_from_constant_string): Likewise.
15866 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15867 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15868 * optabs.c (expand_binop): Likewise.
15869 (expand_unop): Likewise.
15870 (expand_abs_nojump): Likewise.
15871 (expand_one_cmpl_abs_nojump): Likewise.
15872 * simplify-rtx.c (mode_signbit_p): Likewise.
15873 (val_signbit_p): Likewise.
15874 (val_signbit_known_set_p): Likewise.
15875 (val_signbit_known_clear_p): Likewise.
15876 (simplify_relational_operation_1): Likewise.
15877 * tree.c (vector_type_mode): Likewise.
15879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15880 Alan Hayward <alan.hayward@arm.com>
15881 David Sherwood <david.sherwood@arm.com>
15883 * machmode.h (smallest_mode_for_size): Fix formatting.
15884 (smallest_int_mode_for_size): New function.
15885 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15886 instead of smallest_mode_for_size.
15887 * combine.c (make_extraction): Likewise.
15888 * config/arc/arc.c (arc_expand_movmem): Likewise.
15889 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15890 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15891 * config/s390/s390.c (s390_expand_insv): Likewise.
15892 * config/sparc/sparc.c (assign_int_registers): Likewise.
15893 * config/spu/spu.c (spu_function_value): Likewise.
15894 (spu_function_arg): Likewise.
15895 * coverage.c (get_gcov_type): Likewise.
15896 (get_gcov_unsigned_t): Likewise.
15897 * dse.c (find_shift_sequence): Likewise.
15898 * expmed.c (store_bit_field_1): Likewise.
15899 * expr.c (convert_move): Likewise.
15900 (store_field): Likewise.
15901 * internal-fn.c (expand_arith_overflow): Likewise.
15902 * optabs-query.c (get_best_extraction_insn): Likewise.
15903 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15904 * stor-layout.c (layout_type): Likewise.
15905 (initialize_sizetypes): Likewise.
15906 * targhooks.c (default_get_mask_mode): Likewise.
15907 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15910 Alan Hayward <alan.hayward@arm.com>
15911 David Sherwood <david.sherwood@arm.com>
15913 * machmode.h (opt_mode::else_blk): New function.
15914 (int_mode_for_mode): Declare.
15915 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15916 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15918 * cfgexpand.c (expand_debug_expr): Likewise.
15919 * combine.c (gen_lowpart_or_truncate): Likewise.
15920 (gen_lowpart_for_combine): Likewise.
15921 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15922 * config/avr/avr.c (avr_to_int_mode): Likewise.
15923 (avr_out_plus_1): Likewise.
15924 (avr_out_plus): Likewise.
15925 (avr_out_round): Likewise.
15926 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15927 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15928 (s390_expand_vcond): Likewise.
15929 * config/spu/spu.c (spu_split_immediate): Likewise.
15930 (spu_expand_mov): Likewise.
15931 * dse.c (get_stored_val): Likewise.
15932 * expmed.c (store_bit_field_1): Likewise.
15933 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15935 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15936 (extract_low_bits): Likewise.
15937 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15938 handling rather than repeating the check.
15939 (emit_group_store): Likewise.
15940 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15941 * optabs.c (expand_absneg_bit): Likewise.
15942 (expand_copysign_absneg): Likewise.
15943 (expand_copysign_bit): Likewise.
15944 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15945 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15946 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15947 * var-tracking.c (prepare_call_arguments): Likewise.
15948 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15949 int_mode_for_mode instead of mode_for_size.
15950 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15953 Alan Hayward <alan.hayward@arm.com>
15954 David Sherwood <david.sherwood@arm.com>
15956 * machmode.h (int_mode_for_size): New function.
15957 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15958 instead of mode_for_size.
15959 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15961 * combine.c (expand_field_assignment): Use int_mode_for_size
15962 instead of mode_for_size.
15963 (make_extraction): Likewise.
15964 (simplify_shift_const_1): Likewise.
15965 (simplify_comparison): Likewise.
15966 * dojump.c (do_jump): Likewise.
15967 * dwarf2out.c (mem_loc_descriptor): Likewise.
15968 * emit-rtl.c (init_derived_machine_modes): Likewise.
15969 * expmed.c (flip_storage_order): Likewise.
15970 (convert_extracted_bit_field): Likewise.
15971 * expr.c (copy_blkmode_from_reg): Likewise.
15972 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15973 * internal-fn.c (expand_mul_overflow): Likewise.
15974 * lower-subreg.c (simple_move): Likewise.
15975 * optabs-libfuncs.c (init_optabs): Likewise.
15976 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15977 * tree.c (vector_type_mode): Likewise.
15978 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15979 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15980 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15981 * tree-vect-stmts.c (vectorizable_load): Likewise.
15982 (vectorizable_store): Likewise.
15984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15985 Alan Hayward <alan.hayward@arm.com>
15986 David Sherwood <david.sherwood@arm.com>
15988 * coretypes.h (pod_mode): New type.
15989 (scalar_int_mode_pod): New typedef.
15990 * machmode.h (pod_mode): New class.
15991 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15992 * genmodes.c (emit_mode_int_n): Update accordingly.
15993 * lower-subreg.h (target_lower_subreg): Change type to
15994 scalar_int_mode_pod.
15995 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15996 scalar_int_mode_pod.
15998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15999 Alan Hayward <alan.hayward@arm.com>
16000 David Sherwood <david.sherwood@arm.com>
16002 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16003 machine_mode to scalar_int_mode.
16004 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16005 (rs6000_option_override_internal): Remove cast to int.
16006 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16007 machine_mode to scalar_int_mode.
16008 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16009 (rs6000_option_override_internal): Remove cast to int.
16010 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16011 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16013 * config/s390/s390.c (s390_expand_builtin): Likewise.
16014 * coretypes.h (scalar_int_mode): New type.
16015 (opt_scalar_int_mode): New typedef.
16016 * machmode.h (scalar_int_mode): New class.
16017 (scalar_int_mode::includes_p): New function.
16018 (byte_mode): Change type to scalar_int_mode.
16019 (word_mode): Likewise.
16020 (ptr_mode): Likewise.
16021 * emit-rtl.c (byte_mode): Likewise.
16022 (word_mode): Likewise.
16023 (ptr_mode): Likewise.
16024 (init_derived_machine_modes): Update accordingly.
16025 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16026 and MODE_PARTIAL_INT.
16027 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16028 opt_scalar_int_mode.
16030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16031 Alan Hayward <alan.hayward@arm.com>
16032 David Sherwood <david.sherwood@arm.com>
16034 * target.def (libgcc_floating_mode_supported_p): Take a
16036 * doc/tm.texi: Regenerate.
16037 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16039 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16040 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16044 Alan Hayward <alan.hayward@arm.com>
16045 David Sherwood <david.sherwood@arm.com>
16047 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16048 * doc/tm.texi: Regenerate.
16049 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16050 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16051 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16052 * targhooks.h (default_floatn_mode): Likewise.
16053 * targhooks.c (default_floatn_mode): Likewise.
16054 * tree.c (build_common_tree_nodes): Update accordingly.
16056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16057 Alan Hayward <alan.hayward@arm.com>
16058 David Sherwood <david.sherwood@arm.com>
16060 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16061 (mode_iterator::iterate_p): Likewise.
16062 (mode_iterator::get_wider): Likewise.
16063 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16066 Alan Hayward <alan.hayward@arm.com>
16067 David Sherwood <david.sherwood@arm.com>
16069 * coretypes.h (opt_scalar_float_mode): New typedef.
16070 * machmode.h (float_mode_for_size): New function.
16071 * emit-rtl.c (double_mode): Delete.
16072 (init_emit_once): Use float_mode_for_size.
16073 * stor-layout.c (layout_type): Likewise.
16074 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16077 Alan Hayward <alan.hayward@arm.com>
16078 David Sherwood <david.sherwood@arm.com>
16080 * output.h (assemble_real): Take a scalar_float_mode.
16081 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16082 * config/arm/arm.md (consttable_4): Likewise.
16083 (consttable_8): Likewise.
16084 (consttable_16): Likewise.
16085 * config/mips/mips.md (consttable_float): Likewise.
16086 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16087 * varasm.c (assemble_real): Take a scalar_float_mode.
16088 (output_constant_pool_2): Update accordingly.
16089 (output_constant): Likewise.
16091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16092 Alan Hayward <alan.hayward@arm.com>
16093 David Sherwood <david.sherwood@arm.com>
16095 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16096 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16097 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16098 (native_encode_real): Likewise.
16099 (native_interpret_real): Likewise.
16100 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16101 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16104 Alan Hayward <alan.hayward@arm.com>
16105 David Sherwood <david.sherwood@arm.com>
16107 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16108 <scalar_float_mode>. Simplify.
16109 (gen_extend_conv_libfunc): Likewise.
16111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16112 Alan Hayward <alan.hayward@arm.com>
16113 David Sherwood <david.sherwood@arm.com>
16115 * coretypes.h (scalar_float_mode): New type.
16116 * machmode.h (mode_traits::from_int): Use machine_mode if
16117 USE_ENUM_MODES is defined.
16118 (is_a): New function.
16120 (dyn_cast): Likewise.
16121 (scalar_float_mode): New class.
16122 (scalar_float_mode::includes_p): New function.
16123 (is_float_mode): Likewise.
16124 * gdbhooks.py (MachineModePrinter): New class.
16125 (build_pretty_printer): Use it for scalar_float_mode.
16126 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16127 (format_helper::format_helper): Turn into a template.
16128 * genmodes.c (get_mode_class): New function.
16129 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16130 or machine_mode if none.
16131 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16132 as_a <scalar_float_mode>.
16133 * dwarf2out.c (mem_loc_descriptor): Likewise.
16134 (insert_float): Likewise.
16135 (add_const_value_attribute): Likewise.
16136 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16137 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16138 (expand_unop): Update accordingly.
16139 (expand_abs_nojump): Likewise.
16140 (expand_copysign_absneg): Take a scalar_float_mode.
16141 (expand_copysign_bit): Likewise.
16142 (expand_copysign): Update accordingly.
16144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16145 Alan Hayward <alan.hayward@arm.com>
16146 David Sherwood <david.sherwood@arm.com>
16148 * coretypes.h (opt_mode): New class.
16149 * machmode.h (opt_mode): Likewise.
16150 (opt_mode::else_void): New function.
16151 (opt_mode::require): Likewise.
16152 (opt_mode::exists): Likewise.
16153 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16154 (GET_MODE_2XWIDER_MODE): Likewise.
16155 (mode_iterator::get_wider): Update accordingly.
16156 (mode_iterator::get_2xwider): Likewise.
16157 (mode_iterator::get_known_wider): Likewise, turning into a template.
16158 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16159 forcing a wider mode to exist.
16160 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16161 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16162 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16163 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16164 * lower-subreg.c (init_lower_subreg): Likewise.
16165 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16166 on the final iteration.
16167 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16168 a wider mode exists before asking for a move pattern.
16169 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16170 forcing a wider mode to exist.
16171 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16172 returning false if no such mode exists.
16173 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16174 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16175 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16176 Avoid checking for a MODE_INT if we already know the mode is not a
16178 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16179 forcing a wider mode to exist.
16180 (expmed_mult_highpart_optab): Likewise.
16181 (expmed_mult_highpart): Likewise.
16182 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16184 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16185 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16186 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16187 * internal-fn.c (expand_mul_overflow): Update use of
16188 GET_MODE_2XWIDER_MODE.
16189 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16190 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16191 GET_MODE_WIDER_MODE.
16192 (convert_plusminus_to_widen): Likewise.
16193 * tree-switch-conversion.c (array_value_type): Likewise.
16194 * var-tracking.c (emit_note_insn_var_location): Likewise.
16195 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16196 Return false inside rather than outside the loop if no wider mode
16198 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16199 and GET_MODE_2XWIDER_MODE
16200 (can_compare_p): Use else_void.
16201 * gdbhooks.py (OptMachineModePrinter): New class.
16202 (build_pretty_printer): Use it for opt_mode.
16204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16205 Alan Hayward <alan.hayward@arm.com>
16206 David Sherwood <david.sherwood@arm.com>
16208 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16209 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16212 Alan Hayward <alan.hayward@arm.com>
16213 David Sherwood <david.sherwood@arm.com>
16215 * machmode.h (mode_traits): New structure.
16216 (get_narrowest_mode): New function.
16217 (mode_iterator::start): Likewise.
16218 (mode_iterator::iterate_p): Likewise.
16219 (mode_iterator::get_wider): Likewise.
16220 (mode_iterator::get_known_wider): Likewise.
16221 (mode_iterator::get_2xwider): Likewise.
16222 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16223 (FOR_EACH_MODE): Likewise.
16224 (FOR_EACH_MODE_FROM): Likewise.
16225 (FOR_EACH_MODE_UNTIL): Likewise.
16226 (FOR_EACH_WIDER_MODE): Likewise.
16227 (FOR_EACH_2XWIDER_MODE): Likewise.
16228 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16229 * combine.c (simplify_comparison): Likewise
16230 * config/i386/i386.c (type_natural_mode): Likewise.
16231 * cse.c (cse_insn): Likewise.
16232 * dse.c (find_shift_sequence): Likewise.
16233 * emit-rtl.c (init_derived_machine_modes): Likewise.
16234 (init_emit_once): Likewise.
16235 * explow.c (hard_function_value): Likewise.
16236 * expmed.c (extract_fixed_bit_field_1): Likewise.
16237 (extract_bit_field_1): Likewise.
16238 (expand_divmod): Likewise.
16239 (emit_store_flag_1): Likewise.
16240 * expr.c (init_expr_target): Likewise.
16241 (convert_move): Likewise.
16242 (alignment_for_piecewise_move): Likewise.
16243 (widest_int_mode_for_size): Likewise.
16244 (emit_block_move_via_movmem): Likewise.
16245 (copy_blkmode_to_reg): Likewise.
16246 (set_storage_via_setmem): Likewise.
16247 (compress_float_constant): Likewise.
16248 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16249 * optabs-query.c (get_best_extraction_insn): Likewise.
16250 * optabs.c (expand_binop): Likewise.
16251 (expand_twoval_unop): Likewise.
16252 (expand_twoval_binop): Likewise.
16253 (widen_leading): Likewise.
16254 (widen_bswap): Likewise.
16255 (expand_parity): Likewise.
16256 (expand_unop): Likewise.
16257 (prepare_cmp_insn): Likewise.
16258 (prepare_float_lib_cmp): Likewise.
16259 (expand_float): Likewise.
16260 (expand_fix): Likewise.
16261 (expand_sfix_optab): Likewise.
16262 * postreload.c (move2add_use_add2_insn): Likewise.
16263 * reg-stack.c (reg_to_stack): Likewise.
16264 * reginfo.c (choose_hard_reg_mode): Likewise.
16265 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16266 * stor-layout.c (mode_for_size): Likewise.
16267 (smallest_mode_for_size): Likewise.
16268 (mode_for_vector): Likewise.
16269 (finish_bitfield_representative): Likewise.
16270 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16271 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16272 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16273 * var-tracking.c (prepare_call_arguments): Likewise.
16275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16276 Alan Hayward <alan.hayward@arm.com>
16277 David Sherwood <david.sherwood@arm.com>
16279 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16280 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16281 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16282 * machmode.h (mode_size): Move earlier in file.
16283 (mode_precision): Likewise.
16284 (mode_inner): Likewise.
16285 (mode_nunits): Likewise.
16286 (mode_unit_size): Likewise.
16287 (unit_unit_precision): Likewise.
16288 (mode_wider): Likewise.
16289 (mode_2xwider): Likewise.
16290 (machine_mode): New class.
16291 (mode_to_bytes): New function.
16292 (mode_to_bits): Likewise.
16293 (mode_to_precision): Likewise.
16294 (mode_to_inner): Likewise.
16295 (mode_to_unit_size): Likewise.
16296 (mode_to_unit_precision): Likewise.
16297 (mode_to_nunits): Likewise.
16298 (GET_MODE_SIZE): Use mode_to_bytes.
16299 (GET_MODE_BITSIZE): Use mode_to_bits.
16300 (GET_MODE_PRECISION): Use mode_to_precision.
16301 (GET_MODE_INNER): Use mode_to_inner.
16302 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16303 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16304 (GET_MODE_NUNITS): Use mode_to_nunits.
16305 * system.h (ALWAYS_INLINE): New macro.
16306 * config/powerpcspe/powerpcspe-c.c
16307 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16308 int for arg1_mode and arg2_mode.
16310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16311 Alan Hayward <alan.hayward@arm.com>
16312 David Sherwood <david.sherwood@arm.com>
16314 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16315 Prefix mode names with E_ in case statements.
16316 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16317 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16318 (aarch64_split_simd_move): Likewise.
16319 (aarch64_gen_storewb_pair): Likewise.
16320 (aarch64_gen_loadwb_pair): Likewise.
16321 (aarch64_gen_store_pair): Likewise.
16322 (aarch64_gen_load_pair): Likewise.
16323 (aarch64_get_condition_code_1): Likewise.
16324 (aarch64_constant_pool_reload_icode): Likewise.
16325 (get_rsqrte_type): Likewise.
16326 (get_rsqrts_type): Likewise.
16327 (get_recpe_type): Likewise.
16328 (get_recps_type): Likewise.
16329 (aarch64_gimplify_va_arg_expr): Likewise.
16330 (aarch64_simd_container_mode): Likewise.
16331 (aarch64_emit_load_exclusive): Likewise.
16332 (aarch64_emit_store_exclusive): Likewise.
16333 (aarch64_expand_compare_and_swap): Likewise.
16334 (aarch64_gen_atomic_cas): Likewise.
16335 (aarch64_emit_bic): Likewise.
16336 (aarch64_emit_atomic_swap): Likewise.
16337 (aarch64_emit_atomic_load_op): Likewise.
16338 (aarch64_evpc_trn): Likewise.
16339 (aarch64_evpc_uzp): Likewise.
16340 (aarch64_evpc_zip): Likewise.
16341 (aarch64_evpc_ext): Likewise.
16342 (aarch64_evpc_rev): Likewise.
16343 (aarch64_evpc_dup): Likewise.
16344 (aarch64_gen_ccmp_first): Likewise.
16345 (aarch64_gen_ccmp_next): Likewise.
16346 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16347 (alpha_emit_xfloating_libcall): Likewise.
16348 (emit_insxl): Likewise.
16349 (alpha_arg_type): Likewise.
16350 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16351 (arc_preferred_simd_mode): Likewise.
16352 (arc_secondary_reload): Likewise.
16353 (get_arc_condition_code): Likewise.
16354 (arc_print_operand): Likewise.
16355 (arc_legitimate_constant_p): Likewise.
16356 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16357 * config/arc/arc.md (casesi_load): Likewise.
16358 (casesi_compact_jump): Likewise.
16359 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16360 (cc_use_register): Likewise.
16361 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16362 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16363 (arm_init_iwmmxt_builtins): Likewise.
16364 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16365 (neon_expand_vector_init): Likewise.
16366 (arm_attr_length_move_neon): Likewise.
16367 (maybe_get_arm_condition_code): Likewise.
16368 (arm_emit_vector_const): Likewise.
16369 (arm_preferred_simd_mode): Likewise.
16370 (arm_output_iwmmxt_tinsr): Likewise.
16371 (thumb1_output_casesi): Likewise.
16372 (thumb2_output_casesi): Likewise.
16373 (arm_emit_load_exclusive): Likewise.
16374 (arm_emit_store_exclusive): Likewise.
16375 (arm_expand_compare_and_swap): Likewise.
16376 (arm_evpc_neon_vuzp): Likewise.
16377 (arm_evpc_neon_vzip): Likewise.
16378 (arm_evpc_neon_vrev): Likewise.
16379 (arm_evpc_neon_vtrn): Likewise.
16380 (arm_evpc_neon_vext): Likewise.
16381 (arm_validize_comparison): Likewise.
16382 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16383 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16384 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16385 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16386 (c6x_preferred_simd_mode): Likewise.
16387 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16388 (epiphany_rtx_costs): Likewise.
16389 * config/epiphany/predicates.md (proper_comparison_operator):
16391 * config/frv/frv.c (condexec_memory_operand): Likewise.
16392 (frv_emit_move): Likewise.
16393 (output_move_single): Likewise.
16394 (output_condmove_single): Likewise.
16395 (frv_hard_regno_mode_ok): Likewise.
16396 (frv_matching_accg_mode): Likewise.
16397 * config/h8300/h8300.c (split_adds_subs): Likewise.
16398 (h8300_rtx_costs): Likewise.
16399 (h8300_print_operand): Likewise.
16400 (compute_mov_length): Likewise.
16401 (output_logical_op): Likewise.
16402 (compute_logical_op_length): Likewise.
16403 (compute_logical_op_cc): Likewise.
16404 (h8300_shift_needs_scratch_p): Likewise.
16405 (output_a_shift): Likewise.
16406 (compute_a_shift_length): Likewise.
16407 (compute_a_shift_cc): Likewise.
16408 (expand_a_rotate): Likewise.
16409 (output_a_rotate): Likewise.
16410 * config/i386/i386.c (classify_argument): Likewise.
16411 (function_arg_advance_32): Likewise.
16412 (function_arg_32): Likewise.
16413 (function_arg_64): Likewise.
16414 (function_value_64): Likewise.
16415 (ix86_gimplify_va_arg): Likewise.
16416 (ix86_legitimate_constant_p): Likewise.
16417 (put_condition_code): Likewise.
16418 (split_double_mode): Likewise.
16419 (ix86_avx256_split_vector_move_misalign): Likewise.
16420 (ix86_expand_vector_logical_operator): Likewise.
16421 (ix86_split_idivmod): Likewise.
16422 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16423 (ix86_build_const_vector): Likewise.
16424 (ix86_build_signbit_mask): Likewise.
16425 (ix86_match_ccmode): Likewise.
16426 (ix86_cc_modes_compatible): Likewise.
16427 (ix86_expand_branch): Likewise.
16428 (ix86_expand_sse_cmp): Likewise.
16429 (ix86_expand_sse_movcc): Likewise.
16430 (ix86_expand_int_sse_cmp): Likewise.
16431 (ix86_expand_vec_perm_vpermi2): Likewise.
16432 (ix86_expand_vec_perm): Likewise.
16433 (ix86_expand_sse_unpack): Likewise.
16434 (ix86_expand_int_addcc): Likewise.
16435 (ix86_split_to_parts): Likewise.
16436 (ix86_vectorize_builtin_gather): Likewise.
16437 (ix86_vectorize_builtin_scatter): Likewise.
16438 (avx_vpermilp_parallel): Likewise.
16439 (inline_memory_move_cost): Likewise.
16440 (ix86_tieable_integer_mode_p): Likewise.
16441 (x86_maybe_negate_const_int): Likewise.
16442 (ix86_expand_vector_init_duplicate): Likewise.
16443 (ix86_expand_vector_init_one_nonzero): Likewise.
16444 (ix86_expand_vector_init_one_var): Likewise.
16445 (ix86_expand_vector_init_concat): Likewise.
16446 (ix86_expand_vector_init_interleave): Likewise.
16447 (ix86_expand_vector_init_general): Likewise.
16448 (ix86_expand_vector_set): Likewise.
16449 (ix86_expand_vector_extract): Likewise.
16450 (emit_reduc_half): Likewise.
16451 (ix86_emit_i387_round): Likewise.
16452 (ix86_mangle_type): Likewise.
16453 (ix86_expand_round_sse4): Likewise.
16454 (expand_vec_perm_blend): Likewise.
16455 (canonicalize_vector_int_perm): Likewise.
16456 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16457 (expand_vec_perm_1): Likewise.
16458 (expand_vec_perm_interleave3): Likewise.
16459 (expand_vec_perm_even_odd_pack): Likewise.
16460 (expand_vec_perm_even_odd_1): Likewise.
16461 (expand_vec_perm_broadcast_1): Likewise.
16462 (ix86_vectorize_vec_perm_const_ok): Likewise.
16463 (ix86_expand_vecop_qihi): Likewise.
16464 (ix86_expand_mul_widen_hilo): Likewise.
16465 (ix86_expand_sse2_abs): Likewise.
16466 (ix86_expand_pextr): Likewise.
16467 (ix86_expand_pinsr): Likewise.
16468 (ix86_preferred_simd_mode): Likewise.
16469 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16470 * config/i386/sse.md (*andnot<mode>3): Likewise.
16471 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16472 (*<code><mode>3): Likewise.
16473 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16474 (ia64_expand_atomic_op): Likewise.
16475 (ia64_arg_type): Likewise.
16476 (ia64_mode_to_int): Likewise.
16477 (ia64_scalar_mode_supported_p): Likewise.
16478 (ia64_vector_mode_supported_p): Likewise.
16479 (expand_vec_perm_broadcast): Likewise.
16480 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16481 (iq2000_function_arg_advance): Likewise.
16482 (iq2000_function_arg): Likewise.
16483 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16484 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16485 (m68k_libcall_value): Likewise.
16486 (m68k_function_value): Likewise.
16487 (sched_attr_op_type): Likewise.
16488 * config/mcore/mcore.c (mcore_output_move): Likewise.
16489 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16491 (microblaze_function_arg): Likewise.
16492 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16493 (mips_print_operand): Likewise.
16494 (mips_mode_ok_for_mov_fmt_p): Likewise.
16495 (mips_vector_mode_supported_p): Likewise.
16496 (mips_preferred_simd_mode): Likewise.
16497 (mips_expand_vpc_loongson_even_odd): Likewise.
16498 (mips_expand_vec_unpack): Likewise.
16499 (mips_expand_vi_broadcast): Likewise.
16500 (mips_expand_vector_init): Likewise.
16501 (mips_expand_vec_reduc): Likewise.
16502 (mips_expand_msa_cmp): Likewise.
16503 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16504 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16505 (cc_flags_for_mode): Likewise.
16506 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16507 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16508 (nds32_output_casesi_pc_relative): Likewise.
16509 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16510 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16511 (nvptx_gen_unpack): Likewise.
16512 (nvptx_gen_pack): Likewise.
16513 (nvptx_gen_shuffle): Likewise.
16514 (nvptx_gen_wcast): Likewise.
16515 (nvptx_preferred_simd_mode): Likewise.
16516 * config/pa/pa.c (pa_secondary_reload): Likewise.
16517 * config/pa/predicates.md (base14_operand): Likewise.
16518 * config/powerpcspe/powerpcspe-c.c
16519 (altivec_resolve_overloaded_builtin): Likewise.
16520 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16522 (rs6000_preferred_simd_mode): Likewise.
16523 (output_vec_const_move): Likewise.
16524 (rs6000_expand_vector_extract): Likewise.
16525 (rs6000_split_vec_extract_var): Likewise.
16526 (reg_offset_addressing_ok_p): Likewise.
16527 (rs6000_legitimate_offset_address_p): Likewise.
16528 (rs6000_legitimize_address): Likewise.
16529 (rs6000_emit_set_const): Likewise.
16530 (rs6000_const_vec): Likewise.
16531 (rs6000_emit_move): Likewise.
16532 (spe_build_register_parallel): Likewise.
16533 (rs6000_darwin64_record_arg_recurse): Likewise.
16534 (swap_selector_for_mode): Likewise.
16535 (spe_init_builtins): Likewise.
16536 (paired_init_builtins): Likewise.
16537 (altivec_init_builtins): Likewise.
16538 (do_load_for_compare): Likewise.
16539 (rs6000_generate_compare): Likewise.
16540 (rs6000_expand_float128_convert): Likewise.
16541 (emit_load_locked): Likewise.
16542 (emit_store_conditional): Likewise.
16543 (rs6000_output_function_epilogue): Likewise.
16544 (rs6000_handle_altivec_attribute): Likewise.
16545 (rs6000_function_value): Likewise.
16546 (emit_fusion_gpr_load): Likewise.
16547 (emit_fusion_p9_load): Likewise.
16548 (emit_fusion_p9_store): Likewise.
16549 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16550 (fusion_gpr_mem_load): Likewise.
16551 (fusion_addis_mem_combo_load): Likewise.
16552 (fusion_addis_mem_combo_store): Likewise.
16553 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16554 (fusion_gpr_mem_load): Likewise.
16555 (fusion_addis_mem_combo_load): Likewise.
16556 (fusion_addis_mem_combo_store): Likewise.
16557 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16559 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16560 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16561 (rs6000_preferred_simd_mode): Likewise.
16562 (output_vec_const_move): Likewise.
16563 (rs6000_expand_vector_extract): Likewise.
16564 (rs6000_split_vec_extract_var): Likewise.
16565 (reg_offset_addressing_ok_p): Likewise.
16566 (rs6000_legitimate_offset_address_p): Likewise.
16567 (rs6000_legitimize_address): Likewise.
16568 (rs6000_emit_set_const): Likewise.
16569 (rs6000_const_vec): Likewise.
16570 (rs6000_emit_move): Likewise.
16571 (rs6000_darwin64_record_arg_recurse): Likewise.
16572 (swap_selector_for_mode): Likewise.
16573 (paired_init_builtins): Likewise.
16574 (altivec_init_builtins): Likewise.
16575 (rs6000_expand_float128_convert): Likewise.
16576 (emit_load_locked): Likewise.
16577 (emit_store_conditional): Likewise.
16578 (rs6000_output_function_epilogue): Likewise.
16579 (rs6000_handle_altivec_attribute): Likewise.
16580 (rs6000_function_value): Likewise.
16581 (emit_fusion_gpr_load): Likewise.
16582 (emit_fusion_p9_load): Likewise.
16583 (emit_fusion_p9_store): Likewise.
16584 * config/rx/rx.c (rx_gen_move_template): Likewise.
16585 (flags_from_mode): Likewise.
16586 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16587 (s390_slb_comparison): Likewise.
16588 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16589 (s390_vector_mode_supported_p): Likewise.
16590 (s390_cc_modes_compatible): Likewise.
16591 (s390_match_ccmode_set): Likewise.
16592 (s390_canonicalize_comparison): Likewise.
16593 (s390_emit_compare_and_swap): Likewise.
16594 (s390_branch_condition_mask): Likewise.
16595 (s390_rtx_costs): Likewise.
16596 (s390_secondary_reload): Likewise.
16597 (__SECONDARY_RELOAD_CASE): Likewise.
16598 (s390_expand_cs): Likewise.
16599 (s390_preferred_simd_mode): Likewise.
16600 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16601 * config/sh/sh.c (sh_print_operand): Likewise.
16602 (dump_table): Likewise.
16603 (sh_secondary_reload): Likewise.
16604 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16605 * config/sh/sh.md (casesi_worker_1): Likewise.
16606 (casesi_worker_2): Likewise.
16607 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16608 (fcc_comparison_operator): Likewise.
16609 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16610 (emit_soft_tfmode_cvt): Likewise.
16611 (sparc_preferred_simd_mode): Likewise.
16612 (output_cbranch): Likewise.
16613 (sparc_print_operand): Likewise.
16614 (sparc_expand_vec_perm_bmask): Likewise.
16615 (vector_init_bshuffle): Likewise.
16616 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16617 (spu_vector_mode_supported_p): Likewise.
16618 (spu_expand_insv): Likewise.
16619 (spu_emit_branch_or_set): Likewise.
16620 (spu_handle_vector_attribute): Likewise.
16621 (spu_builtin_splats): Likewise.
16622 (spu_builtin_extract): Likewise.
16623 (spu_builtin_promote): Likewise.
16624 (spu_expand_sign_extend): Likewise.
16625 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16626 (tilegx_simd_int): Likewise.
16627 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16628 (tilepro_simd_int): Likewise.
16629 * config/v850/v850.c (const_double_split): Likewise.
16630 (v850_print_operand): Likewise.
16631 (ep_memory_offset): Likewise.
16632 * config/vax/vax.c (vax_rtx_costs): Likewise.
16633 (vax_output_int_move): Likewise.
16634 (vax_output_int_add): Likewise.
16635 (vax_output_int_subtract): Likewise.
16636 * config/visium/predicates.md (visium_branch_operator): Likewise.
16637 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16638 (visium_print_operand_address): Likewise.
16639 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16640 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16641 (xtensa_expand_conditional_branch): Likewise.
16642 (xtensa_copy_incoming_a7): Likewise.
16643 (xtensa_output_literal): Likewise.
16644 * dfp.c (decimal_real_maxval): Likewise.
16645 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16648 Alan Hayward <alan.hayward@arm.com>
16649 David Sherwood <david.sherwood@arm.com>
16651 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16652 (mode_nunits_inline): Likewise.
16653 (mode_inner_inline): Likewise.
16654 (mode_unit_size_inline): Likewise.
16655 (mode_unit_precision_inline): Likewise.
16656 (emit_insn_modes_h): Likewise. Also emit a #define of the
16658 (emit_mode_wider): Add an E_ prefix to mode names.
16659 (emit_mode_complex): Likewise.
16660 (emit_mode_inner): Likewise.
16661 (emit_mode_adjustments): Likewise.
16662 (emit_mode_int_n): Likewise.
16663 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16664 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16665 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16666 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16667 (CRC32_BUILTIN, ENTRY): Likewise.
16668 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16669 (aarch64_pop_regs): Likewise.
16670 (aarch64_process_components): Likewise.
16671 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16672 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16673 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16674 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16675 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16676 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16678 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16679 (pcmpistr): Likewise.
16680 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16681 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16682 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16684 * config/rl78/rl78.c (mduc_regs): Likewise.
16685 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16686 (htm_expand_builtin): Likewise.
16687 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16688 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16689 * config/xtensa/xtensa.c (print_operand): Likewise.
16690 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16691 (NUM_MODE_VECTOR_INT): Likewise.
16692 * genoutput.c (null_operand): Likewise.
16693 (output_operand_data): Likewise.
16694 * genrecog.c (print_parameter_value): Likewise.
16695 * lra.c (debug_operand_data): Likewise.
16697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16699 * dwarf2out.c (scompare_loc_descriptor_wide)
16700 (scompare_loc_descriptor_narrow): New functions, split out from...
16701 (scompare_loc_descriptor): ...here.
16702 * expmed.c (emit_store_flag_int): New function, split out from...
16703 (emit_store_flag): ...here.
16705 2017-08-30 Richard Biener <rguenther@suse.de>
16707 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16708 (dwarf2out_early_finish): Move setting of AT_pubnames from
16709 early debug output to early finish.
16711 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16713 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16714 and -mdata-region to the assembler.
16716 2017-08-30 Richard Biener <rguenther@suse.de>
16718 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16720 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16721 (dwarf2out_early_global_decl): Only generate a DIE for the
16722 abstract origin if it doesn't already exist or is a declaration DIE.
16723 (resolve_addr): Do not add the linkage name twice when
16724 generating a stub DIE for the DW_TAG_GNU_call_site target.
16726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16728 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16729 Use machine_mode rather than int for arg1_mode.
16731 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16734 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16735 that the second argument of the built-in functions to unpack
16736 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16737 switch statement instead a lot of if statements.
16738 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16739 Allow 64-bit values to be in Altivec registers as well as
16740 traditional floating point registers.
16741 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16743 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16745 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16746 MAX_REGS_PER_ADDRESS == 1.
16748 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16750 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16751 * config/i386/i386.c (ix86_option_override_internal): Simplify
16752 setting of opts->x_flag_entry.
16754 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16755 Jakub Jelinek <jakub@redhat.com>
16756 Richard Biener <rguenther@suse.de>
16758 PR tree-optimization/81503
16759 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16760 folded constant fits in the target type; reorder tests for clarity.
16762 2017-08-29 Martin Liska <mliska@suse.cz>
16764 * passes.def: Include pass_lower_switch.
16765 * stmt.c (dump_case_nodes): Remove and move to
16766 tree-switch-conversion.
16767 (case_values_threshold): Likewise.
16768 (expand_switch_as_decision_tree_p): Likewise.
16769 (emit_case_decision_tree): Likewise.
16770 (expand_case): Likewise.
16771 (balance_case_nodes): Likewise.
16772 (node_has_low_bound): Likewise.
16773 (node_has_high_bound): Likewise.
16774 (node_is_bounded): Likewise.
16775 (emit_case_nodes): Likewise.
16776 (struct simple_case_node): New struct.
16777 (add_case_node): Remove.
16778 (emit_case_dispatch_table): Use vector instead of case_list.
16779 (reset_out_edges_aux): Remove.
16780 (compute_cases_per_edge): Likewise.
16781 (expand_case): Build list of simple_case_node.
16782 (expand_sjlj_dispatch_table): Use it.
16783 * tree-switch-conversion.c (struct case_node): Moved from
16784 stmt.c and adjusted.
16785 (emit_case_nodes): Likewise.
16786 (node_has_low_bound): Likewise.
16787 (node_has_high_bound): Likewise.
16788 (node_is_bounded): Likewise.
16789 (case_values_threshold): Likewise.
16790 (reset_out_edges_aux): Likewise.
16791 (compute_cases_per_edge): Likewise.
16792 (add_case_node): Likewise.
16793 (dump_case_nodes): Likewise.
16794 (balance_case_nodes): Likewise.
16795 (expand_switch_as_decision_tree_p): Likewise.
16796 (emit_jump): Likewise.
16797 (emit_case_decision_tree): Likewise.
16798 (try_switch_expansion): Likewise.
16799 (do_jump_if_equal): Likewise.
16800 (emit_cmp_and_jump_insns): Likewise.
16801 (fix_phi_operands_for_edge): New function.
16802 (record_phi_operand_mapping): Likewise.
16803 (class pass_lower_switch): New pass.
16804 (pass_lower_switch::execute): New function.
16805 (make_pass_lower_switch): Likewise.
16806 (conditional_probability):
16807 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16808 * tree-pass.h: Add make_pass_lower_switch.
16810 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16813 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16816 2017-08-29 Richard Biener <rguenther@suse.de>
16818 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16819 we do not add a DW_AT_inline attribute twice.
16820 (gen_subprogram_die): Remove code setting DW_AT_inline on
16821 DECL_ABSTRACT_P nodes.
16823 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16825 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16826 calls to internal functions.
16827 (gimplify_modify_expr): Likewise.
16828 * tree-call-cdce.c (use_internal_fn): Likewise.
16829 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16830 (convert_to_divmod): Set the nothrow flag.
16831 * tree-if-conv.c (predicate_mem_writes): Likewise.
16832 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16833 (vectorizable_call): Likewise.
16834 (vectorizable_store): Likewise.
16835 (vectorizable_load): Likewise.
16836 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16837 (vect_recog_mask_conversion_pattern): Likewise.
16839 2017-08-29 Martin Liska <mliska@suse.cz>
16842 * gcc.c (driver_handle_option): Add new argument.
16843 * opts-common.c (handle_option): Pass
16844 target_option_override_hook.
16845 * opts-global.c (lang_handle_option): Add new option.
16846 (set_default_handlers): Add new argument.
16847 (decode_options): Likewise.
16848 * opts.c (target_handle_option): Likewise.
16849 (common_handle_option): Call target_option_override_hook.
16850 * opts.h (struct cl_option_handler_func): Add hook for
16851 target option override.
16852 (struct cl_option_handlers): Likewise.
16853 (set_default_handlers): Add new argument.
16854 (decode_options): Likewise.
16855 (common_handle_option): Likewise.
16856 (target_handle_option): Likewise.
16857 * toplev.c (toplev::main): Pass targetm.target_option.override
16860 2017-08-29 Richard Biener <rguenther@suse.de>
16861 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16863 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16864 life to the active subtree.
16866 2017-08-28 Jeff Law <law@redhat.com>
16868 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16869 derive_equivalences.
16870 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16871 Code moved into....
16872 (edge_info::derive_equivalences): New private member function
16874 * tree-ssa-dom.c (class edge_info): Changed from a struct
16875 to a class. Add ctor/dtor, methods and data members.
16876 (edge_info::edge_info): Renamed from allocate_edge_info.
16877 Initialize additional members.
16878 (edge_info::~edge_info): New.
16879 (free_dom_edge_info): Delete the edge info.
16880 (record_edge_info): Use new class & associated member functions.
16881 Tighten forms for testing for edge equivalences.
16882 (record_temporary_equivalences): Iterate over the simple
16883 equivalences rather than assuming there's only one per edge.
16884 (cprop_into_successor_phis): Iterate over the simple
16885 equivalences rather than assuming there's only one per edge.
16886 (optimize_stmt): Use operand_equal_p rather than pointer
16887 equality for mini-DSE code.
16889 2017-08-28 Nathan Sidwell <nathan@acm.org>
16891 * gcc.c (execute): Fold SIGPIPE handling into switch
16892 statement. Adjust internal error message.
16894 2017-08-28 Richard Biener <rguenther@suse.de>
16897 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16898 Do nothing for removed DIEs.
16900 2017-08-28 Richard Biener <rguenther@suse.de>
16902 PR tree-optimization/81977
16903 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16906 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16909 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16910 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16911 using targetm.gen_mem_thread_fence.
16913 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16916 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16917 predicate to register_operand. Reorder operands.
16918 (*btr<mode>): Ditto.
16919 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16920 (*btr<mode>_mask): Ditto.
16922 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16924 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16925 * config/rs6000/xmmintrin.h: New file.
16926 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16928 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16931 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16932 parameter and_insn and return it.
16933 (recombine_lvx_pattern): Insert a copy to ensure availability of
16934 the base register of the copied masking operation at the point of
16935 the instruction replacement.
16936 (recombine_stvx_pattern): Likewise.
16938 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16940 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16941 undocumented switches.
16942 (-mpower9-dform-vector): Likewise.
16943 (-mpower9-dform): Likewise.
16944 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16945 comments to delete references to -mpower9-dform* switches.
16946 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16947 Delete reference to -mpower9-dform* switches, test for
16948 -mpower9-vector instead.
16949 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16950 (OTHER_P9_VECTOR_MASKS): Likewise.
16951 (POWERPC_MASKS): Likewise.
16952 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16953 tests against -mpower9-dform* to -mpower9-vector. Delete code
16954 that checked for -mpower9-dform* consistancy with other options.
16955 Add test for -mpower9-misc to enable other power9 switches.
16956 (rs6000_init_hard_regno_mode_ok): Likewise.
16957 (rs6000_option_override_internal): Likewise.
16958 (rs6000_emit_prologue): Likewise.
16959 (rs6000_emit_epilogue): Likewise.
16960 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16961 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16962 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16964 (emit_fusion_p9_store): Likewise.
16965 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16966 resetting these macros if the assembler does not support ISA 3.0
16968 (TARGET_P9_DFORM_VECTOR): Likewise.
16969 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16970 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16972 2017-08-25 Alan Modra <amodra@gmail.com>
16974 PR rtl-optimization/81747
16975 * cse.c (cse_extended_basic_block): Don't attempt to record
16976 equivalences for degenerate conditional branches that branch
16977 to their fall-through.
16979 2017-08-24 Martin Sebor <msebor@redhat.com>
16981 PR middle-end/81908
16982 * gimple-fold.c (size_must_be_zero_p): New function.
16983 (gimple_fold_builtin_memory_op): Call it.
16985 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16987 * config/rs6000/mm_malloc.h: New file.
16989 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16991 PR tree-optimization/81913
16992 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16993 analysis when either IVs in condition can wrap.
16995 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16997 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16998 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17000 2017-08-24 Richard Biener <rguenther@suse.de>
17003 * targhooks.c (default_target_can_inline_p): Properly
17004 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17005 is present and always compare.
17006 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17007 infer -mfpmath=sse from TARGET_SSE_P.
17008 (ix86_can_inline_p): Properly use target_option_default_node when
17009 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17011 2017-08-24 Richard Biener <rguenther@suse.de>
17014 * dwarf2out.c (output_die): Handle flag_generate_offload like
17016 (output_comp_unit): Likewise.
17017 (gen_array_type_die): Likewise.
17018 (dwarf2out_early_finish): Likewise.
17019 (note_variable_value_in_expr): Likewise.
17020 (dwarf2out_finish): Likewise. Adjust assert.
17021 * cgraphunit.c (symbol_table::compile): Move setting of
17022 flag_generate_offload earlier ...
17023 (symbol_table::finalize_compilation_unit): ... here, before
17024 early debug finalization.
17026 2017-08-24 Richard Biener <rguenther@suse.de>
17028 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17029 and ipa-fnsummary.h.
17030 (ix86_can_inline_p): When ix86_fpmath flags do not match
17031 check whether the callee uses FP math at all.
17033 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17035 PR middle-end/81931
17036 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17037 instead of TYPE_PRECISION.
17039 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17040 Alan Hayward <alan.hayward@arm.com>
17041 David Sherwood <david.sherwood@arm.com>
17043 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17045 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17047 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17049 (ix86_frame::stack_realign_allocate): New field.
17050 (struct machine_frame_state): Modify comments.
17051 (machine_frame_state::sp_realigned_fp_end): New field.
17052 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17053 layout calculation.
17054 (sp_valid_at): Add assertion to assure no attempt to access invalid
17055 offset of a realigned stack.
17056 (fp_valid_at): Likewise.
17057 (choose_baseaddr): Modify comments.
17058 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17059 ix86_expand_prologue.
17060 (ix86_expand_prologue): Modify stack realignment and allocation.
17061 (ix86_expand_epilogue): Modify comments.
17062 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17063 avx2_runtime, avx512f, and avx512f_runtime.
17065 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17067 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17068 (mstackrealign): Do not init to -1.
17069 * config/i386/i386.c (ix86_option_override_internal):
17070 Check opts_set, not opts when setting default value of
17071 opts->x_ix86_force_align_arg_pointer.
17073 2017-08-23 Richard Biener <rguenther@suse.de>
17075 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17076 lang_hooks.decl_printable_name.
17077 * print-rtl-function.c (print_rtx_function): Likewise.
17078 * tree-pretty-print.c (dump_function_header): Likewise.
17080 2017-08-23 Richard Biener <rguenther@suse.de>
17083 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17084 -g0 at compile-time.
17086 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17088 PR middle-end/19706
17089 * doc/sourcebuild.texi (Other hardware attributes):
17092 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17094 PR middle-end/19706
17095 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17096 Fix single-use check.
17098 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17100 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17102 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17104 * doc/install.texi: Modify to add more details on running selected
17107 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17109 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17110 is combined with -mabi=ms.
17111 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17114 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17116 PR tree-optimization/81488
17117 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17118 and cached_basis fields.
17119 (MAX_SPREAD): New constant.
17120 (alloc_cand_and_find_basis): Initialize new fields.
17121 (clear_visited): New function.
17122 (create_phi_basis_1): Rename from create_phi_basis, set visited
17123 and cached_basis fields.
17124 (create_phi_basis): New wrapper function.
17125 (phi_add_costs_1): Rename from phi_add_costs, add spread
17126 parameter, set visited field, short-circuit when limits reached.
17127 (phi_add_costs): New wrapper function.
17128 (record_phi_increments_1): Rename from record_phi_increments, set
17130 (record_phi_increments): New wrapper function.
17131 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17132 (phi_incr_cost): New wrapper function.
17133 (all_phi_incrs_profitable_1): Rename from
17134 all_phi_incrs_profitable, set visited field.
17135 (all_phi_incrs_profitable): New wrapper function.
17137 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17138 Alan Hayward <alan.hayward@arm.com>
17139 David Sherwood <david.sherwood@arm.com>
17141 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17142 that takes the outer and inner modes.
17143 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17144 comparison as the canonical test for a paradoxical subreg.
17145 * combine.c (simplify_set): Use paradoxical_subreg_p.
17146 (make_extraction): Likewise.
17147 (force_to_mode): Likewise.
17148 (rtx_equal_for_field_assignment_p): Likewise.
17149 (gen_lowpart_for_combine): Likewise.
17150 (simplify_comparison): Likewise.
17151 * cse.c (equiv_constant): Likewise.
17152 * expmed.c (store_bit_field_1): Likewise.
17153 * final.c (alter_subreg): Likewise.
17154 * fwprop.c (propagate_rtx): Likewise.
17155 (forward_propagate_subreg): Likewise.
17156 * ira-conflicts.c (ira_build_conflicts): Likewise.
17157 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17158 * lra-constraints.c (curr_insn_transform): Likewise.
17159 (split_reg): Likewise.
17160 * lra-eliminations.c (move_plus_up): Likewise.
17161 (lra_eliminate_regs_1): Likewise.
17162 * recog.c (general_operand): Likewise.
17163 * ree.c (combine_reaching_defs): Likewise.
17164 * reload.c (push_reload): Likewise.
17165 (find_reloads): Likewise.
17166 * reload1.c (elimination_effects): Likewise.
17167 (compute_reload_subreg_offset): Likewise.
17168 (choose_reload_regs): Likewise.
17169 * rtlanal.c (subreg_lsb_1): Likewise.
17170 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17171 (simplify_subreg): Likewise.
17172 * var-tracking.c (track_loc_p): Likewise.
17173 * emit-rtl.c (byte_lowpart_offset): Likewise.
17174 (paradoxical_subreg_p): Delete out-of-line definition.
17176 2017-08-22 Jeff Law <law@redhat.com>
17178 PR tree-optimization/81741
17179 PR tree-optimization/71947
17180 * tree-ssa-dom.c: Include tree-inline.h.
17181 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17182 equivalences if one is more expensive to compute than the other.
17183 * tree-ssa-scopedtables.h (class const_or_copies): Make
17184 record_const_or_copy_raw method private.
17185 (class avail_exprs_stack): New method simplify_binary_operation.
17186 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17187 avail_exprs_stack::simplify_binary_operation as needed.
17188 (avail_exprs_stack::simplify_binary_operation): New function.
17190 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17192 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17193 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17194 (DOT_SYMBOLS): Likewise.
17195 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17196 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17197 (RS6000_ABI_NAME): Likewise.
17198 (TARGET_CMODEL): Likewise.
17199 (TOC_SECTION_ASM_OP): Likewise.
17200 (SET_CMODEL): New macro.
17201 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17203 2017-08-22 Richard Biener <rguenther@suse.de>
17205 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17206 to allow for free-lang-data replacements similar to verify_type_variant.
17208 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17209 Alan Hayward <alan.hayward@arm.com>
17210 David Sherwood <david.sherwood@arm.com>
17212 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17213 VOIDmode for the LABEL_REF.
17215 2017-08-22 Richard Biener <rguenther@suse.de>
17217 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17219 2017-08-22 Richard Biener <rguenther@suse.de>
17221 * common.opt (feliminate-dwarf2-dups): Ignore.
17222 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17223 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17224 same_die_p_wrap, compute_section_prefix,
17225 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17226 (comdat_symbol_id, comdat_symbol_number): Likewise.
17227 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17229 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17230 (output_die): Mark unreachable path unreachable.
17231 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17232 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17233 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17234 (dwarf2out_early_finish): Likewise.
17236 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17238 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17240 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17243 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17244 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17245 (avr_attribute_table) <io, io_low, address>: Initialize
17246 .decl_required with true.
17248 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17250 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17251 undocumented debugging options.
17252 (-mvsx-scalar-double): Likewise.
17253 (-mallow-df-permute): Likewise.
17254 (-mvectorize-builtins): Likewise.
17255 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17256 (rs6000_builtin_vectorized_function): Likewise.
17257 (rs6000_builtin_md_vectorized_function): Likewise.
17258 (rs6000_opt_vars): Likewise.
17260 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17263 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17264 (*btrq_imm): Rename from *btrq.
17265 (*btcq_imm): Rename from *btcq.
17266 (btsc): New code attribute.
17267 (*<btsc><mode>): New insn pattern.
17268 (*btr<mode>): Ditto.
17269 (*<btsc><mode>_mask): New insn_and_split pattern.
17270 (*btr<mode>_mask): Ditto.
17272 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17273 Alan Hayward <alan.hayward@arm.com>
17274 David Sherwood <david.sherwood@arm.com>
17276 * function.c (pad_below): Simplify padding calculation.
17278 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17279 Alan Hayward <alan.hayward@arm.com>
17280 David Sherwood <david.sherwood@arm.com>
17282 * target.def (function_prologue): Remove frame size argument.
17283 (function_epilogue): Likewise.
17284 * doc/tm.texi: Regenerate.
17285 * final.c (final_start_function): Update call to function_prologue.
17286 (final_end_function): Update call to function_epilogue.
17287 (default_function_pro_epilogue): Remove frame size argument.
17288 * output.h (default_function_pro_epilogue): Likewise.
17289 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17290 (arm_output_function_prologue): Likewise.
17291 * config/frv/frv.c (frv_function_prologue): Likewise.
17292 (frv_function_epilogue): Likewise.
17293 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17294 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17295 (ia64_output_function_epilogue): Likewise.
17296 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17297 (m32r_output_function_epilogue): Likewise.
17298 * config/microblaze/microblaze.c (microblaze_function_prologue)
17299 (microblaze_function_epilogue): Likewise.
17300 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17301 (mips_output_function_epilogue): Likewise.
17302 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17303 (mmix_target_asm_function_epilogue): Likewise.
17304 * config/msp430/msp430.c (msp430_start_function): Likewise.
17305 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17306 (nds32_asm_function_epilogue): Likewise.
17307 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17308 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17309 (pa_output_function_epilogue): Likewise.
17310 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17311 (rs6000_output_function_epilogue): Likewise.
17312 * config/rl78/rl78.c (rl78_start_function): Likewise.
17313 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17314 (rs6000_output_function_epilogue): Likewise.
17315 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17316 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17317 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17318 (sparc_asm_function_epilogue): Likewise.
17320 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17322 * tree.h (type_has_mode_precision_p): New function.
17323 * convert.c (convert_to_integer_1): Use it.
17324 * expr.c (expand_expr_real_2): Likewise.
17325 (expand_expr_real_1): Likewise.
17326 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17327 * match.pd: Likewise.
17328 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17329 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17330 * tree-tailcall.c (process_assignment): Likewise.
17331 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17332 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17333 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17334 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17335 (vectorizable_assignment): Likewise.
17336 (vectorizable_shift): Likewise.
17337 (vectorizable_operation): Likewise.
17338 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17340 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17342 * match.pd: Add pow (C, x) simplification.
17344 2017-08-21 Richard Biener <rguenther@suse.de>
17346 PR tree-optimization/81900
17347 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17348 for blocks with abnormal predecessors.
17349 (compute_antic): Do not set visited flag prematurely.
17351 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17354 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17356 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17358 * stor-layout.h (vector_type_mode): Move to...
17359 * tree.h (vector_type_mode): ...here.
17360 * stor-layout.c (vector_type_mode): Move to...
17361 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17363 2017-08-21 Richard Biener <rguenther@suse.de>
17365 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17366 register_external_die hooks.
17367 (debug_false_tree_charstarstar_uhwistar): Declare.
17368 (debug_nothing_tree_charstar_uhwi): Likewise.
17369 * debug.c (do_nothing_debug_hooks): Adjust.
17370 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17371 (debug_nothing_tree_charstar_uhwi): Likewise.
17372 * dbxout.c (dbx_debug_hooks): Adjust.
17373 (xcoff_debug_hooks): Likewise.
17374 * sdbout.c (sdb_debug_hooks): Likewise.
17375 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17376 * dwarf2out.c (macinfo_label_base): New global.
17377 (dwarf2out_register_external_die): New function for the
17378 register_external_die hook.
17379 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17380 (dwarf2_debug_hooks): Use them.
17381 (dwarf2_lineno_debug_hooks): Adjust.
17382 (struct die_struct): Add with_offset flag.
17383 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17384 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17385 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17386 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17387 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17388 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17389 defining section names for the early LTO debug variants.
17390 (reset_indirect_string): New helper.
17391 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17392 (print_dw_val): Add support for offsetted symbol references.
17393 (get_ultimate_context): Split out from is_cxx.
17394 (is_cxx): Use get_ultimate_context.
17395 (is_fortran): Add decl overload.
17396 (compute_comp_unit_symbol): Split out worker from
17397 compute_section_prefix.
17398 (compute_section_prefix): Call compute_comp_unit_symbol and
17399 set comdat_type_p here.
17400 (output_die): Skip DIE symbol output for the LTO added one.
17401 Handle DIE symbol references with offset.
17402 (output_comp_unit): Guard section name mangling properly.
17403 For LTO debug sections emit a symbol at the section beginning
17404 which we use to refer to its DIEs.
17405 (add_abstract_origin_attribute): For DIEs registered via
17406 dwarf2out_register_external_die directly refer to the early
17407 DIE rather than indirectly through the shadow one we created.
17408 Remove obsolete call to dwarf2out_abstract_function for
17409 non-function/block origins.
17410 (gen_array_type_die): When generating early LTO debug do
17411 not emit DW_AT_string_length.
17412 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17413 late when in LTO. As suggested place a gcc_unreachable for
17414 the DECL_ABSTRACT_P case.
17415 (gen_subprogram_die): Avoid another specification DIE
17416 for early built declarations/definitions for the late LTO case.
17417 (gen_variable_die): Add type references for late duplicated VLA dies
17419 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17420 we have the abstract instance already.
17421 (process_scope_var): Adjust decl DIE contexts in LTO which
17422 first puts them in limbo.
17423 (gen_decl_die): Do not generate type DIEs late apart from
17424 types for VLAs or for decls we do not yet have a DIE. Do not
17425 call dwarf2out_abstract_function late.
17426 (dwarf2out_early_global_decl): Make sure to create DIEs
17427 for abstract instances of a decl first.
17428 (dwarf2out_late_global_decl): Adjust comment.
17429 (output_macinfo_op): With multiple macro sections use
17430 macinfo_label_base to distinguish labels.
17431 (output_macinfo): Likewise. Update macinfo_label_base.
17432 Pass in the line info label.
17433 (note_variable_value_in_expr): When generating LTO resolve
17434 all variable values here by generating DIEs as needed.
17435 (init_sections_and_labels): Add early LTO debug flag parameter
17436 and generate different sections and names if set. Add generation
17437 counter for the labels so we can have multiple of them.
17438 (reset_dies): Helper to allow DIEs to be output multiple times.
17439 (dwarf2out_finish): When outputting DIEs to the fat part of an
17440 LTO object first reset DIEs.
17441 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17442 (modified_type_die): Check for decl_ultimate_origin being self
17444 (gen_type_die_with_usage): Likewise.
17445 (gen_typedef_die): Allow decl_ultimate_origin being self.
17446 (set_decl_abstract_flags): Remove.
17447 (set_block_abstract_flags): Likewise.
17448 (dwarf2out_abstract_function): Treat the early generated DIEs
17449 as the abstract copy and only add DW_AT_inline and
17450 DW_AT_artificial here and call set_decl_origin_self.
17451 If the DIE has an abstract origin don't do anything.
17452 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17453 if we have none yet (Go fails to build one, PR78628).
17454 (variably_modified_type_p): Prevent endless recursion for Ada
17455 cyclic pointer types.
17456 * lto-streamer-in.c: Include debug.h.
17457 (dref_queue): New global.
17458 (lto_read_tree_1): Stream in DIE references.
17459 (lto_input_tree): Register DIE references.
17460 (input_function): Stream DECL_DEBUG_ARGS.
17461 * lto-streamer-out.c: Include debug.h.
17462 (lto_write_tree_1): Output DIE references.
17463 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17464 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17465 (output_function): Stream DECL_DEBUG_ARGS.
17466 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17467 Stream DECL_ABSTRACT_ORIGIN.
17468 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17469 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17470 DECL_CONTEXT for file-scope decls.
17471 * lto-streamer.h (struct dref_entry): Declare.
17472 (dref_queue): Likewise.
17473 * cfgexpand.c (pass_expand::execute): Do not call the
17474 outlining_inline_function hook here.
17475 * lto-wrapper.c (debug_obj): New global.
17476 (tool_cleanup): Unlink it if required.
17477 (debug_objcopy): New function.
17478 (run_gcc): Handle early debug sections in the IL files by
17479 extracting them to separate files, partially linkin them and
17480 feeding the result back as result to the linker.
17481 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17482 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17483 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17484 sections into a separate segment.
17485 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17487 (darwin_asm_dwarf_section): Likewise.
17488 (darwin_asm_output_dwarf_offset): Likewise.
17489 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17491 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17492 Alan Hayward <alan.hayward@arm.com>
17493 David Sherwood <david.sherwood@arm.com>
17495 * read-md.h (md_reader::record_potential_iterator_use): Replace
17496 pointer argument with an rtx and an index.
17497 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17498 (apply_mode_iterator): Likewise.
17499 (apply_code_iterator): Likewise.
17500 (apply_int_iterator): Likewise.
17501 (apply_subst_iterator): Likewise.
17502 (record_iterator_use): Likewise.
17503 (record_attribute_use): Likewise.
17504 (md_reader::record_potential_iterator_use): Likewise. Update calls
17505 to record_iterator_use and apply_iterator.
17506 (iterator_use): Replace ptr with x and index.
17507 (attribute_use): Likewise.
17508 (apply_attribute_uses): Update calls to apply_iterator.
17509 (apply_iterators): Likewise. Update initialization of iterator_use.
17510 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17511 and record_potential_iterator_use.
17512 (rtx_reader::read_rtx_operand): Likewise.
17514 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17515 Alan Hayward <alan.hayward@arm.com>
17516 David Sherwood <david.sherwood@arm.com>
17518 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17521 2017-08-21 Richard Biener <rguenther@suse.de>
17523 PR middle-end/81884
17524 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17525 at struct end conservatively when comparing common bases.
17527 2017-08-21 Richard Biener <rguenther@suse.de>
17529 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17530 (mem_ref_in_stmt): Remove.
17531 (determine_max_movement): Use ref index to get at the reference.
17532 (invariantness_dom_walker::before_dom_children): Deal with
17533 lim data already initialized.
17534 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17536 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17538 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17539 (TARGET_ISA_ROUND): Ditto.
17540 (TARGET_ROUND): Ditto.
17541 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17542 * config/i386/i386.md: Ditto.
17543 * config/i386/sse.md: Ditto.
17544 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17545 with OPTION_MASK_ISA_SSE4_1.
17547 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17550 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17551 __builtin_ia32_lzcnt_u16.
17553 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17556 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17557 (rs6000_set_current_function): Rewrite function to use it.
17559 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17562 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17563 and TYPE_WARN_IF_NOT_ALIGN.
17564 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17565 (handle_warn_if_not_align): New.
17566 (place_union_field): Call handle_warn_if_not_align.
17567 (place_field): Call handle_warn_if_not_align.
17568 Copy TYPE_WARN_IF_NOT_ALIGN.
17569 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17570 (layout_type): Likewise.
17571 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17573 (tree_decl_common): Add warn_if_not_align.
17574 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17575 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17576 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17577 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17578 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17579 * doc/extend.texi: Document warn_if_not_aligned attribute.
17580 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17582 2017-08-17 Martin Liska <mliska@suse.cz>
17585 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17586 (get_data_dependence): Use it as pointer type.
17587 (distribute_loop): Likewise.
17589 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17591 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17592 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17593 (p8_vmrgow_<mode>_direct): New define_insn.
17594 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17595 handle endianness for vmrgew and vmrgow permute patterns.
17597 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17599 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17600 * config/rs6000/rs6000-cpus.def: Remove comment.
17601 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17602 (POWERPC_MASKS): Likewise.
17603 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17604 use of TARGET_VSX_TIMODE.
17605 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17606 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17607 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17608 (rs6000_option_override_internal): Remove dead code.
17609 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17610 (rs6000_legitimize_reload_address): Likewise.
17611 (rs6000_legitimate_address_p): Likewise.
17612 (rs6000_opt_masks): Delete "vsx-timode".
17613 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17614 from function comment.
17615 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17616 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17617 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17619 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17620 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17621 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17622 (VSX_TI): Likewise.
17624 (define_peephole2): Likewise.
17626 2017-08-17 Martin Sebor <msebor@redhat.com>
17629 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17630 past the end of an array.
17631 (test_pp_format): Add test cases.
17633 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17635 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17636 missing ECF_NOTHROW flags.
17638 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17641 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17642 operands residing in integer registers.
17643 (*vsx_le_perm_load_<mode>): Likewise.
17644 (*vsx_le_perm_store_<mode>): Likewise.
17645 (define_peephole2): Add peepholes to optimize the above.
17647 2017-08-17 Marek Polacek <polacek@redhat.com>
17649 PR middle-end/81814
17650 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17651 to mimic what shorten_compare did. Change the return type to bool.
17652 (fold_cond_expr_with_comparison): Update call to
17653 operand_equal_for_comparison_p.
17654 (fold_ternary_loc): Likewise.
17656 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17658 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17660 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17662 2017-08-17 Richard Biener <rguenther@suse.de>
17664 * tree-ssa-structalias.c (solve_graph): When propagating
17665 to successors update the graphs succ edges and avoid duplicate work.
17667 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17670 * config/i386/i386.c (ix86_option_override_internal): Save target
17671 specific options after ix86_stack_protector_guard_reg was changed.
17673 2017-08-17 Richard Biener <rguenther@suse.de>
17675 PR tree-optimization/81827
17676 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17677 (new_var_info): Initialize it conservatively.
17678 (get_call_vi): Mark register vars.
17679 (new_scalar_tmp_constraint_exp): Likewise.
17680 (handle_rhs_call): Likewise.
17681 (handle_const_call): Likewise.
17682 (create_function_info_for): Likewise.
17683 (solve_constraints): Sort varinfos to separate register from
17684 non-register vars to pack points-to solution bitmaps during
17687 2017-08-17 Marek Polacek <polacek@redhat.com>
17689 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17691 2017-08-17 Richard Biener <rguenther@suse.de>
17693 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17694 to true when overflow is undefined and we saturated the result.
17696 2017-08-17 Alan Modra <amodra@gmail.com>
17699 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17700 Don't use store multiple if only one reg needs saving.
17701 (interesting_frame_related_regno): New function.
17702 (rs6000_frame_related): Don't emit frame info for regs that
17704 (rs6000_emit_epilogue): Likewise.
17706 2017-08-16 Nathan Sidwell <nathan@acm.org>
17708 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17709 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17710 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17711 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17712 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17713 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17714 (verify_type): Adjust for TYPE_BINFO move.
17715 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17716 process TYPE_BINFO directly.
17717 (hash_tree): Likewise.
17718 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17720 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17723 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17725 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17727 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17730 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17731 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17732 Add ix86_binary_operator_ok to insn constraint.
17733 (*iordi_1_bts): Ditto.
17734 (*xordi_1_btc): Ditto.
17735 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17736 Update corresponding peephole2 pattern.
17740 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17742 PR tree-optimization/81832
17743 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17744 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17746 2017-08-16 Marek Polacek <polacek@redhat.com>
17749 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17750 regarding size_zero_node.
17752 2017-08-16 Martin Liska <mliska@suse.cz>
17755 * config.gcc: Respect previously set extra_objs in case
17758 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17760 PR tree-optimization/81835
17761 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17762 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17763 not depend on the phi.
17765 2017-08-16 Alan Modra <amodra@gmail.com>
17767 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17770 2017-08-16 Alan Modra <amodra@gmail.com>
17772 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17773 (save_reg_p): ..into this. Update all callers.
17774 (first_reg_to_save): Simplify.
17776 2017-08-16 Alan Modra <amodra@gmail.com>
17778 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17781 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17785 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17786 constant count if that count is less than 32.
17788 2017-08-15 Nathan Sidwell <nathan@acm.org>
17790 * gcc.c (execute): Emit friendlier message if inferior is killed
17791 by an external cause.
17793 2017-08-15 Richard Biener <rguenther@suse.de>
17795 PR tree-optimization/81790
17796 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17797 CONSTRUCTORs from simplifying and VN.
17799 2017-08-14 Martin Sebor <msebor@redhat.com>
17801 * builtin-attrs.def: Add comments.
17803 2017-08-14 Martin Sebor <msebor@redhat.com>
17806 * doc/extend.texi (attribute nonstring): Document new attribute.
17808 2017-08-14 Martin Sebor <msebor@redhat.com>
17811 * tree-diagnostic.c (default_tree_printer): Handle %G.
17812 * gimple-pretty-print.h (percent_G_format): Declare new function.
17813 * gimple-pretty-print.c (percent_G_format): Define.
17814 * tree-pretty-print.c (percent_K_format): Add argument.
17816 2017-08-14 Martin Sebor <msebor@redhat.com>
17818 PR translation/79998
17819 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17820 Remove a stray space.
17822 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17825 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17826 (*iordi_1_bts): Ditto.
17827 (*xordi_1_btc): Ditto.
17829 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17832 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17833 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17835 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17836 quoted strings, and make more translator-friendly.
17837 (darwin_rs6000_override_options): Likewise.
17838 (rs6000_option_override_internal): Likewise.
17839 (rs6000_return_in_memory): Fix overlong line.
17840 (init_cmulative_args): Use quoted strings, and make more
17841 translator-friendly.
17842 (rs6000_pass_by_reference): Fix overlong line.
17843 (def_builtin): Use quoted strings.
17844 (altivec_expand_predicate_builtin): Use quoted strings, and make
17845 more translator-friendly.
17846 (htm_expand_builtin): Use quoted strings.
17847 (cpu_expand_builtin): Use quoted strings, and make more
17848 translator-friendly.
17849 (altivec_expand_builtin): Likewise.
17850 (paired_expand_predicate_builtin): Likewise.
17851 (rs6000_invalid_builtin): Likewise.
17852 (builtin_function_type): Use quoted strings.
17853 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17854 more translator-friendly.
17855 (rs6000_trampoline_init): Likewise.
17856 (rs6000_handle_altivec_attribute): Likewise.
17857 (rs6000_inner_target_options): Use quoted strings.
17858 (rs6000_disable_incompatible_switches): Likewise.
17859 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17860 strings, and make more translator-friendly.
17861 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17863 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17865 PR tree-optimization/81799
17866 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17867 cond_expr to simple gimple operand.
17869 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17871 PR middle-end/46932
17872 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17874 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17878 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17879 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17880 TARGET_GASISR_PROLOGUES.
17881 * config/avr/avr.c (avr_option_override): Same.
17882 (avr_pass_pre_proep::execute): Same.
17884 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17887 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17888 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17890 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17892 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17893 define_insn_and_split. Split to a memory load from 0 in
17894 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17895 using PTR mode iterator.
17896 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17897 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17898 (*add_tp_<mode>): Redefine as define_insn_and_split.
17899 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17900 address space. Merge with *add_tp_x32 using PTR mode iterator.
17901 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17902 Split to an add with a memory load from 0 in
17903 DEFAULT_TLS_SEG_REG address space.
17905 2017-08-12 Andrew Pinski <apinski@cavium.com>
17907 * config/aarch64/aarch64-option-extensions.def (rdma):
17908 Fix feature string to what Linux prints out in /proc/cpuinfo.
17910 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17913 * dwarf2out.c (modified_type_die): For C typedef types that have
17914 an ultimate origin, process the ultimate origin instead of the
17916 (gen_typedef_die): Assert that input DECLs have no ultimate
17918 (gen_type_die_with_usage): For typedef variants that have an
17919 ultimate origin, just call gen_decl_die on the original DECL.
17920 (process_scope_var): Avoid creating DIEs for local typedefs and
17921 concrete static variables.
17923 2017-08-12 Alan Modra <amodra@gmail.com>
17927 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17928 match gnu-user.h startfile.
17929 (ENDFILE_LINUX_SPEC): Similarly.
17931 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17934 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17936 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17938 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17939 * config/aarch64/aarch64.md (mov<mode>): Change.
17940 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17941 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17942 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17944 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17946 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17947 for storage order barriers.
17949 2017-08-11 Martin Liska <mliska@suse.cz>
17952 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17953 variables of void type.
17955 2017-08-11 Martin Liska <mliska@suse.cz>
17957 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17958 TARGET_SUPPORTS_ALIASES.
17959 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17960 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17961 (optimize_weakref): Likewise.
17962 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17963 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17964 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17966 2017-08-11 Martin Liska <mliska@suse.cz>
17969 * config/i386/i386.c (make_resolver_func): Do complete
17970 refactoring of the function.
17972 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17975 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17976 * config/i386/i386.c (ix86_stack_protect_guard): Use
17977 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17978 * doc/invoke.texi (x86 Options): Document
17979 -mstack-protector-guard-symbol= option.
17981 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17983 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17984 * config/i386/i386.c (ix86_split_stack_guard): New function.
17985 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17986 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17987 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17988 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17989 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17990 (split_stack_space_check): Call ix86_split_stack_guard.
17992 2017-08-10 Martin Sebor <msebor@redhat.com>
17994 * print-tree.c (print_node): Print location using the established
17996 Replace spaces with colons.
17997 (debug_raw, debug): Ditto.
17999 2017-08-10 Martin Sebor <msebor@redhat.com>
18002 * pretty-print.c (pp_format): Correct the handling of %s precision.
18004 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18007 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18009 (ix86_finalize_stack_frame_flags): This. Also clear
18010 frame_pointer_needed if -fno-omit-frame-pointer is used without
18012 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18013 with ix86_finalize_stack_frame_flags.
18014 (ix86_expand_epilogue): Likewise.
18015 (ix86_expand_split_stack_prologue): Likewise.
18016 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18018 2017-08-10 Martin Liska <mliska@suse.cz>
18021 * c-attribs.c (handle_target_attribute):
18022 Report warning for an empty string argument of target attribute.
18024 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18027 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18029 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18030 or DECL_NONLOCAL labels.
18031 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18032 or DECL_NONLOCAL labels here.
18034 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18036 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18037 to indicate when early gimple folding has been disabled.
18038 (rs6000_gimple_fold_builtin): Add debug content.
18039 (rs6000_invalid_builtin): Fix whitespace.
18040 (rs6000_expand_builtin): Fix whitespace.
18041 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18043 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18046 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18047 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18049 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18051 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18052 (qdf24xx): Likewise.
18053 * config/aarch64/aarch64.md: Include falkor.md.
18054 * config/aarch64/falkor.md: New.
18056 2017-08-09 Marek Polacek <polacek@redhat.com>
18059 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18060 * diagnostic.c (emit_diagnostic): Add a comment.
18061 (emit_diagnostic_valist): New function.
18063 2017-08-09 Marek Polacek <polacek@redhat.com>
18066 * input.c (make_location): New overload.
18067 * input.h (make_location): Declare.
18069 2017-08-08 Alan Modra <amodra@gmail.com>
18070 H.J. Lu <hongjiu.lu@intel.com>
18073 * gcc.c (NO_PIE_SPEC): Delete.
18074 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18076 (LINK_PIE_SPEC): ..to here.
18077 (LINK_COMMAND_SPEC): Support -no-pie.
18078 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18079 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18081 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18082 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18083 (ENDFILE_CRTEND_SPEC): Similarly.
18085 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18088 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18089 (mstack-protector-guard-offset=): Ditto.
18090 * config/i386/i386.c (ix86_option_override): Handle
18091 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18093 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18094 ix86_stack_protect_guard_offset variables.
18095 (TARGET_STACK_PROTECT_GUARD): Always define.
18096 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18097 and -mstack-protector-guard-offset= options.
18099 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18101 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18102 boundary case for the last candidate.
18104 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18106 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18107 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18109 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18111 PR middle-end/19706
18112 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18113 * config/aarch64/aarch64-builtins.c
18114 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18115 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18116 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18118 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18119 Andrew Pinski <pinskia@gmail.com>
18121 PR middle-end/19706
18122 * internal-fn.def (XORSIGN): New.
18123 * optabs.def (xorsign_optab): New.
18124 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18125 (convert_expand_mult_copysign): New.
18126 (pass_optimize_widening_mul::execute): Call
18127 convert_expand_mult_copysign.
18129 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18131 PR tree-optimization/81354
18132 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18133 Insert on edges rather than explicitly creating landing pads.
18134 (analyze_candidates_and_replace): Commit edge inserts.
18136 2017-08-08 Richard Biener <rguenther@suse.de>
18138 PR middle-end/81719
18139 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18140 (expand_simple_operations): Also look through ADDR_EXPRs with
18141 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18143 2017-08-08 Richard Biener <rguenther@suse.de>
18145 PR tree-optimization/81723
18146 * tree-vect-slp.c (struct bst_traits): New hash traits.
18147 (bst_fail): New global.
18148 (vect_build_slp_tree_2): New worker, split out from ...
18149 (vect_build_slp_tree): ... this now wrapping it with using
18150 bst_fail set to cache SLP tree build fails. Properly handle
18152 (vect_analyze_slp_instance): Allocate and free bst_fail.
18154 2017-08-08 Martin Liska <mliska@suse.cz>
18157 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18158 LABEL_DECLs that can be from a different function.
18160 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18162 PR tree-optimization/81744
18163 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18164 loop's number of iterations.
18166 2017-08-08 Martin Liska <mliska@suse.cz>
18168 * asan.c: Include header files.
18169 * attribs.c (build_decl_attribute_variant): New function moved
18171 (build_type_attribute_qual_variant): Likewise.
18172 (cmp_attrib_identifiers): Likewise.
18173 (simple_cst_list_equal): Likewise.
18174 (omp_declare_simd_clauses_equal): Likewise.
18175 (attribute_value_equal): Likewise.
18176 (comp_type_attributes): Likewise.
18177 (build_type_attribute_variant): Likewise.
18178 (lookup_ident_attribute): Likewise.
18179 (remove_attribute): Likewise.
18180 (merge_attributes): Likewise.
18181 (merge_type_attributes): Likewise.
18182 (merge_decl_attributes): Likewise.
18183 (merge_dllimport_decl_attributes): Likewise.
18184 (handle_dll_attribute): Likewise.
18185 (attribute_list_equal): Likewise.
18186 (attribute_list_contained): Likewise.
18187 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18188 (lookup_attribute_by_prefix): Likewise.
18189 * bb-reorder.c: Include header files.
18190 * builtins.c: Likewise.
18191 * calls.c: Likewise.
18192 * cfgexpand.c: Likewise.
18193 * cgraph.c: Likewise.
18194 * cgraphunit.c: Likewise.
18195 * convert.c: Likewise.
18196 * dwarf2out.c: Likewise.
18197 * final.c: Likewise.
18198 * fold-const.c: Likewise.
18199 * function.c: Likewise.
18200 * gimple-expr.c: Likewise.
18201 * gimple-fold.c: Likewise.
18202 * gimple-pretty-print.c: Likewise.
18203 * gimple.c: Likewise.
18204 * gimplify.c: Likewise.
18205 * hsa-common.c: Likewise.
18206 * hsa-gen.c: Likewise.
18207 * internal-fn.c: Likewise.
18208 * ipa-chkp.c: Likewise.
18209 * ipa-cp.c: Likewise.
18210 * ipa-devirt.c: Likewise.
18211 * ipa-fnsummary.c: Likewise.
18212 * ipa-inline.c: Likewise.
18213 * ipa-visibility.c: Likewise.
18215 * lto-cgraph.c: Likewise.
18216 * omp-expand.c: Likewise.
18217 * omp-general.c: Likewise.
18218 * omp-low.c: Likewise.
18219 * omp-offload.c: Likewise.
18220 * omp-simd-clone.c: Likewise.
18221 * opts-global.c: Likewise.
18222 * passes.c: Likewise.
18223 * predict.c: Likewise.
18224 * sancov.c: Likewise.
18225 * sanopt.c: Likewise.
18226 * symtab.c: Likewise.
18227 * toplev.c: Likewise.
18228 * trans-mem.c: Likewise.
18229 * tree-chkp.c: Likewise.
18230 * tree-eh.c: Likewise.
18231 * tree-into-ssa.c: Likewise.
18232 * tree-object-size.c: Likewise.
18233 * tree-parloops.c: Likewise.
18234 * tree-profile.c: Likewise.
18235 * tree-ssa-ccp.c: Likewise.
18236 * tree-ssa-live.c: Likewise.
18237 * tree-ssa-loop.c: Likewise.
18238 * tree-ssa-sccvn.c: Likewise.
18239 * tree-ssa-structalias.c: Likewise.
18240 * tree-ssa.c: Likewise.
18241 * tree-streamer-in.c: Likewise.
18242 * tree-vectorizer.c: Likewise.
18243 * tree-vrp.c: Likewise.
18244 * tsan.c: Likewise.
18245 * ubsan.c: Likewise.
18246 * varasm.c: Likewise.
18247 * varpool.c: Likewise.
18248 * tree.c: Remove functions moved to attribs.[ch].
18249 * tree.h: Likewise.
18250 * config/aarch64/aarch64.c: Add attrs.h header file.
18251 * config/alpha/alpha.c: Likewise.
18252 * config/arc/arc.c: Likewise.
18253 * config/arm/arm.c: Likewise.
18254 * config/avr/avr.c: Likewise.
18255 * config/bfin/bfin.c: Likewise.
18256 * config/c6x/c6x.c: Likewise.
18257 * config/cr16/cr16.c: Likewise.
18258 * config/cris/cris.c: Likewise.
18259 * config/darwin.c: Likewise.
18260 * config/epiphany/epiphany.c: Likewise.
18261 * config/fr30/fr30.c: Likewise.
18262 * config/frv/frv.c: Likewise.
18263 * config/ft32/ft32.c: Likewise.
18264 * config/h8300/h8300.c: Likewise.
18265 * config/i386/winnt.c: Likewise.
18266 * config/ia64/ia64.c: Likewise.
18267 * config/iq2000/iq2000.c: Likewise.
18268 * config/lm32/lm32.c: Likewise.
18269 * config/m32c/m32c.c: Likewise.
18270 * config/m32r/m32r.c: Likewise.
18271 * config/m68k/m68k.c: Likewise.
18272 * config/mcore/mcore.c: Likewise.
18273 * config/microblaze/microblaze.c: Likewise.
18274 * config/mips/mips.c: Likewise.
18275 * config/mmix/mmix.c: Likewise.
18276 * config/mn10300/mn10300.c: Likewise.
18277 * config/moxie/moxie.c: Likewise.
18278 * config/msp430/msp430.c: Likewise.
18279 * config/nds32/nds32-isr.c: Likewise.
18280 * config/nds32/nds32.c: Likewise.
18281 * config/nios2/nios2.c: Likewise.
18282 * config/nvptx/nvptx.c: Likewise.
18283 * config/pa/pa.c: Likewise.
18284 * config/pdp11/pdp11.c: Likewise.
18285 * config/powerpcspe/powerpcspe.c: Likewise.
18286 * config/riscv/riscv.c: Likewise.
18287 * config/rl78/rl78.c: Likewise.
18288 * config/rx/rx.c: Likewise.
18289 * config/s390/s390.c: Likewise.
18290 * config/sh/sh.c: Likewise.
18291 * config/sol2.c: Likewise.
18292 * config/sparc/sparc.c: Likewise.
18293 * config/spu/spu.c: Likewise.
18294 * config/stormy16/stormy16.c: Likewise.
18295 * config/tilegx/tilegx.c: Likewise.
18296 * config/tilepro/tilepro.c: Likewise.
18297 * config/v850/v850.c: Likewise.
18298 * config/vax/vax.c: Likewise.
18299 * config/visium/visium.c: Likewise.
18300 * config/xtensa/xtensa.c: Likewise.
18302 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18305 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18306 constraints since the -mupper-regs-* switches have been
18308 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18309 into a vector from a double word element that was extracted from
18310 another vector, and eliminate extra XXPERMDI instructions.
18311 (vsx_concat_<mode>_2): Likewise.
18312 (vsx_concat_<mode>_3): Likewise.
18313 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18314 concat to allow optimizing inserts from previous extracts.
18316 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18318 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18319 memory reference to a SSP offset in TLS address space.
18320 (ix86_print_operand) <case '@'>: Remove.
18321 (ix86_print_operand_punct_valid_p): Remove '@' code.
18322 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18323 UNSPEC_SP_TLS_TEST.
18324 (stack_tls_protect_set_<mode>): Remove.
18325 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18326 (stack_tls_protect_test_<mode>): Remove.
18327 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18329 2017-08-07 Olivier Hainque <hainque@adacore.com>
18332 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18334 2017-08-07 Douglas Rupp <rupp@adacore.com>
18336 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18337 variable was referenced as multidir in command.
18339 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18342 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18345 2017-08-07 Martin Liska <mliska@suse.cz>
18347 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18348 * config/rl78/rl78.c: Add include of attribs.h.
18349 * config/sh/sh.c: Likewise.
18350 * config/v850/v850.c: Likewise.
18352 2017-08-07 Tom de Vries <tom@codesourcery.com>
18354 PR middle-end/78266
18355 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18357 2017-08-07 Martin Liska <mliska@suse.cz>
18359 * config/mips/mips.c: Include attribs.h.
18361 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18364 * doc/invoke.texi: Document change in behvaior for -Ofast for
18367 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18369 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18371 (aarch64_pop_regs): Likewise.
18372 (aarch64_gen_load_pair): Likewise.
18373 (aarch64_save_callee_saves): Likewise.
18374 (aarch64_restore_callee_saves): Likewise.
18376 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18378 * config/i386/i386.c: Revert the last change.
18380 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18383 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18385 (ix86_finalize_stack_frame_flags): This. Also clear
18386 frame_pointer_needed if -fno-omit-frame-pointer is used without
18388 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18389 with ix86_finalize_stack_frame_flags.
18390 (ix86_expand_epilogue): Likewise.
18391 (ix86_expand_split_stack_prologue): Likewise.
18393 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18396 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18397 to P_AES for Westmere.
18399 2017-08-07 Jonathan Yong <10walls@gmail.com>
18401 * config/i386/mingw.opt (fset-stack-executable): Removed.
18402 * config/i386/cygming.opt (fset-stack-executable): Moved
18404 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18406 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18408 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18410 2017-08-07 Marek Polacek <polacek@redhat.com>
18412 PR middle-end/81737
18413 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18415 2017-08-07 Martin Liska <mliska@suse.cz>
18417 * attribs.h (canonicalize_attr_name): New function.
18418 (cmp_attribs): Move from c-format.c and adjusted.
18419 (is_attribute_p): Moved from tree.h.
18420 * tree-inline.c: Add new includes.
18421 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18422 (private_is_attribute_p): Remove.
18423 (private_lookup_attribute): Likewise.
18424 (private_lookup_attribute_by_prefix): Simplify.
18425 (remove_attribute): Use is_attribute_p.
18426 * tree.h: Remove removed declarations.
18428 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18430 PR middle-end/81698
18431 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18432 instead of computing it in the function. Formatting fix.
18433 (expand_case): Don't rely on default_edge being the first edge,
18434 clear it if removing it, pass default_edge to
18435 emit_case_dispatch_table.
18436 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18439 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18441 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18442 insn in the function, emit NOP after the insn.
18444 2017-08-06 Tom de Vries <tom@codesourcery.com>
18446 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18449 2017-08-06 Tom de Vries <tom@codesourcery.com>
18451 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18454 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18456 PR tree-optimization/57371
18457 * match.pd: New pattern.
18459 2017-08-04 Marek Polacek <polacek@redhat.com>
18461 PR middle-end/81695
18462 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18463 perform the computation in offset_int.
18465 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18467 PR tree-optimization/81136
18468 * tree-vectorizer.h: Include tree-hash-traits.h.
18469 (vec_base_alignments): New typedef.
18470 (vec_info): Add a base_alignments field.
18471 (vect_record_base_alignments): Declare.
18472 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18474 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18475 (create_data_ref): Add an is_conditional_in_stmt argument.
18476 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18477 the is_conditional_in_stmt field.
18478 (data_ref_loc): Add an is_conditional_in_stmt field.
18479 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18480 (find_data_references_in_stmt): Update call to create_data_ref.
18481 (graphite_find_data_references_in_stmt): Likewise.
18482 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18483 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18484 (vect_record_base_alignment): New function.
18485 (vect_record_base_alignments): Likewise.
18486 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18487 for nested statements even if we fail to compute a misalignment.
18488 Use pooled base alignments for unconditional references.
18489 (vect_find_same_alignment_drs): Compare base addresses instead
18491 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18492 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18494 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18496 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18497 Add an explicit name for the enum. Use auto_vec for slp_instances
18498 and grouped_stores.
18499 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18501 (_bb_vec_info): Add a constructor and destructor.
18502 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18503 (destroy_loop_vec_info): Delete.
18504 (vect_destroy_datarefs): Likewise.
18505 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18506 (vec_info::vec_info): New function.
18507 (vec_info::~vec_info): Likewise.
18508 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18509 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18510 destroy_loop_vec_info.
18511 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18512 (_loop_vec_info::_loop_vec_info): ...this.
18513 (destroy_loop_vec_info): Replace with...
18514 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18515 the stmt_vec_infos. Leave handling of vec_info information to its
18516 destructor. Remove explicit vector releases.
18517 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18518 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18519 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18520 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18521 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18522 (destroy_bb_vec_info): Replace with...
18523 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18524 information to its destructor.
18525 (vect_slp_analyze_bb_1): Use new and delete instead of
18526 new_bb_vec_info and destroy_bb_vec_info.
18527 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18530 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18532 * tree-data-ref.h (subscript): Add access_fn field.
18533 (data_dependence_relation): Add could_be_independent_p.
18534 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18535 (same_access_functions): Move to tree-data-ref.c.
18536 * tree-data-ref.c (ref_contains_union_access_p): New function.
18537 (access_fn_component_p): Likewise.
18538 (access_fn_components_comparable_p): Likewise.
18539 (dr_analyze_indices): Add a reference to access_fn_component_p.
18540 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18542 (constant_access_functions): Likewise.
18543 (add_other_self_distances): Likewise.
18544 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18545 (initialize_data_dependence_relation): Use XCNEW and remove
18546 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18547 of access functions that have the same type. Allow the
18548 subsequence to end with different bases in some circumstances.
18549 Record the chosen access functions in SUB_ACCESS_FN.
18550 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18551 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18552 (subscript_dependence_tester_1): Likewise dra and drb.
18553 (build_classic_dist_vector): Update calls accordingly.
18554 (subscript_dependence_tester): Likewise.
18555 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18556 DDR_COULD_BE_INDEPENDENT_P.
18557 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18558 comp_alias_ddrs instead of may_alias_ddrs.
18559 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18561 (vect_analyze_data_ref_dependence): Use it if
18562 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18563 distance vectors if that fails.
18564 (dependence_distance_ge_vf): New function.
18565 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18566 LOOP_VINFO_MAY_ALIAS_DDRS.
18568 2017-08-04 Richard Biener <rguenther@suse.de>
18570 PR middle-end/81705
18571 * fold-const.c (fold_binary_loc): Properly restrict
18572 minus_var0 && minus_var1 case when associating undefined overflow
18575 2017-08-04 Tom de Vries <tom@codesourcery.com>
18577 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18579 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18581 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18582 Don't start diagnostic messages with a capital letter.
18583 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18585 (rs6000_invalid_builtin): Likewise.
18586 (rs6000_trampoline_init): Likewise.
18588 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18591 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18592 after setting changeable df flags.
18594 2017-08-03 Richard Biener <rguenther@suse.de>
18596 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18597 up if the use is in USE - X.
18599 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18601 * toplev.c (dumpfile.h): New include.
18602 (internal_error_reentered): New static function. Use it...
18603 (internal_error_function): ...here to handle reentered internal_error.
18605 2017-08-03 Richard Biener <rguenther@suse.de>
18607 PR middle-end/81148
18608 * fold-const.c (split_tree): Add minus_var and minus_con
18609 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18610 here but always use minus_*.
18611 (associate_trees): Assert we never associate with MINUS_EXPR
18612 and NULL first operand. Do not recurse for PLUS_EXPR operands
18613 when associating as MINUS_EXPR either.
18614 (fold_binary_loc): Track minus_var and minus_con.
18616 2017-08-03 Tom de Vries <tom@codesourcery.com>
18619 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18620 ACCEL_COMPILER, apply finish_options on
18621 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18623 2017-08-03 Tom de Vries <tom@codesourcery.com>
18626 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18627 function_entry_patch_area_size > 0.
18629 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18632 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18633 instead of 10??LU, perform unit multiplication in wide_int,
18634 don't change alloc_object_size_limit if the limit is larger
18637 PR tree-optimization/81655
18638 PR tree-optimization/81588
18639 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18640 the case when ranges[i].low and high are 1 for unsigned type with
18643 PR middle-end/81052
18644 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18645 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18647 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18649 * tree-vrp.h: Add include guard.
18651 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18654 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18655 (ud2): New insn pattern.
18656 * config/i386/i386.c (ix86_expand_epilogue):
18657 For naked functions, generate ud2 instead of trap insn.
18659 2017-08-02 Marek Polacek <polacek@redhat.com>
18662 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18664 2017-08-02 Tom de Vries <tom@codesourcery.com>
18665 Cesar Philippidis <cesar@codesourcery.com>
18667 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18668 Add missing edge probabilities.
18670 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18672 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18673 Correct endianness.
18675 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18677 PR middle-end/79499
18678 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18679 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18680 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18681 prologue_seq sequences - if any.
18683 2017-08-02 Richard Biener <rguenther@suse.de>
18685 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18686 via vectors if supported, integer extracts via punning if supported
18687 or otherwise vector extracts.
18689 2017-08-02 Richard Biener <rguenther@suse.de>
18691 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18693 (bitmap_insert_into_set): ... this.
18695 2017-08-02 Richard Biener <rguenther@suse.de>
18697 PR tree-optimization/81633
18699 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18701 PR tree-optimization/71752
18702 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18704 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18706 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18707 (machine_function::call_ms2sysv_pad_out): Remove field.
18708 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18709 (ix86_compute_frame_layout): Likewise.
18711 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18714 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18715 attribute with interrupt attribute.
18717 2017-08-01 Andrew Pinski <apinski@cavium.com>
18719 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18720 BIT_INSERT_EXPR's operand 1
18721 to see if the types precision matches.
18723 2017-08-01 Martin Liska <mliska@suse.cz>
18725 PR middle-end/70140
18726 * builtins.c (expand_builtin_memcpy_args): Remove.
18727 (expand_builtin_memcpy): Call newly added function
18728 expand_builtin_memory_copy_args.
18729 (expand_builtin_memcpy_with_bounds): Likewise.
18730 (expand_builtin_mempcpy): Remove last argument.
18731 (expand_builtin_mempcpy_with_bounds): Likewise.
18732 (expand_builtin_memory_copy_args): New function created from
18733 expand_builtin_mempcpy_args with small modifications.
18734 (expand_builtin_mempcpy_args): Remove.
18735 (expand_builtin_stpcpy): Remove unused argument.
18736 (expand_builtin): Likewise.
18737 (expand_builtin_with_bounds): Likewise.
18739 2017-08-01 Martin Liska <mliska@suse.cz>
18742 Make mempcpy more optimal (PR middle-end/70140).
18744 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18748 __builtin_vec_cmpne verify both arguments are compatible vectors
18749 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18750 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18751 move computation of aligned to after checking the argument types.
18755 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18756 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18759 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18760 Doug Rupp <rupp@adacore.com>
18761 Olivier Hainque <hainque@adacore.com>
18763 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18764 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18766 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18767 for TARGET_OS_CPP_BUILTIN.
18768 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18769 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18771 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18772 passing required abi options to the assembler for EABI configurations.
18773 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18774 of .text.hot and .text.unlikely sections for kernel modules when
18775 using ARM style exceptions.
18776 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18777 options. Add EXTRA_CC1_SPEC.
18778 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18780 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18782 (ARM_TARGET2_DWARF_FORMAT): Define.
18783 * config/arm/t-vxworks: Adjust multilib control to removal of the
18784 Diab command line options.
18786 2017-08-01 Martin Liska <mliska@suse.cz>
18788 PR gcov-profile/81561
18789 * gcov.c (unblock): Make unblocking safe as we need to preserve
18790 index correspondence of blocks and block_lists.
18792 2017-08-01 Richard Biener <rguenther@suse.de>
18794 PR tree-optimization/81181
18795 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18796 (compute_antic): ... end of iteration here.
18798 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18800 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18801 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18802 (ftree-slp-vectorize): Likewise.
18803 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18804 can no longer be set independent of flag_tree_loop_vectorize.
18805 * omp-general.c (emp_max_vf): Likewise.
18806 * opts.c (enable_fdo_optimizations): Remove references to
18807 flag_tree_vectorize, these are now implicit.
18808 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18809 and leave it for the options machinery.
18811 2017-08-01 Martin Liska <mliska@suse.cz>
18813 PR middle-end/70140
18814 * builtins.c (expand_builtin_memcpy_args): Remove.
18815 (expand_builtin_memcpy): Call newly added function
18816 expand_builtin_memory_copy_args.
18817 (expand_builtin_memcpy_with_bounds): Likewise.
18818 (expand_builtin_mempcpy): Remove last argument.
18819 (expand_builtin_mempcpy_with_bounds): Likewise.
18820 (expand_builtin_memory_copy_args): New function created from
18821 expand_builtin_mempcpy_args with small modifications.
18822 (expand_builtin_mempcpy_args): Remove.
18823 (expand_builtin_stpcpy): Remove unused argument.
18824 (expand_builtin): Likewise.
18825 (expand_builtin_with_bounds): Likewise.
18827 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18830 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18831 print "ds:" only for immediates in generic address space.
18833 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18836 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18837 (ix86_function_ok_for_sibcall): Return false for naked functions.
18839 2017-08-01 Richard Biener <rguenther@suse.de>
18841 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18842 (compute_antic): Seed worklist with exit block predecessors.
18843 * cfganal.c (dfs_find_deadend): For a cycle return the source
18844 of the edge closing it.
18846 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18848 * config/aarch64/aarch64.c
18849 (aarch64_can_const_movi_rtx_p): Move 0 check.
18851 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18853 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18854 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18856 * match.pd: Ditto in address comparison pattern.
18858 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18860 PR tree-optimization/81627
18861 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18862 closed ssa form for store-store chain.
18864 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18866 PR tree-optimization/81620
18867 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18868 for store-store chain.
18870 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18872 PR tree-optimization/81588
18873 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18874 ranges[i].in_p, invert comparison code ccode. For >/>=,
18875 swap rhs1 and rhs2 and comparison code unconditionally,
18876 for </<= don't do that. Don't swap rhs1/rhs2 again if
18877 ranges[i].in_p, instead invert comparison code ccode if
18878 opcode or oe->rank is BIT_IOR_EXPR.
18881 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18882 a direct optab to conversion optab.
18883 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18884 with GET_MODE_INNER as last argument instead of optab_handler.
18885 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18886 vector extraction if possible and optab is available.
18887 * expr.c (store_constructor): Use convert_optab_handler instead
18888 of optab_handler. Use vector initialization from smaller
18889 vectors if possible and optab is available.
18890 * tree-vect-stmts.c (vectorizable_load): Likewise.
18891 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18892 now have two modes.
18893 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18894 of vec_init from half-sized vectors with the same element mode.
18895 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18896 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18897 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18898 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18899 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18900 after mode in gen_vec_extract* calls.
18901 (vec_extract<mode>): Renamed to ...
18902 (vec_extract<mode><ssescalarmodelower>): ... this.
18903 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18904 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18905 element mode after mode in gen_vec_init* calls.
18906 (VEC_INIT_HALF_MODE): New mode iterator.
18907 (vec_init<mode>): Renamed to ...
18908 (vec_init<mode><ssescalarmodelower>): ... this.
18909 (vec_init<mode><ssehalfvecmodelower>): New expander.
18910 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18911 (vec_extractv2sfsf): ... this.
18912 (vec_initv2sf): Renamed to ...
18913 (vec_initv2sfsf): ... this.
18914 (vec_extractv2si): Renamed to ...
18915 (vec_extractv2sisi): ... this.
18916 (vec_initv2si): Renamed to ...
18917 (vec_initv2sisi): ... this.
18918 (vec_extractv4hi): Renamed to ...
18919 (vec_extractv4hihi): ... this.
18920 (vec_initv4hi): Renamed to ...
18921 (vec_initv4hihi): ... this.
18922 (vec_extractv8qi): Renamed to ...
18923 (vec_extractv8qiqi): ... this.
18924 (vec_initv8qi): Renamed to ...
18925 (vec_initv8qiqi): ... this.
18926 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18927 (vec_init<mode>): Renamed to ...
18928 (vec_init<mode><VEC_base_l>): ... this.
18929 (vec_extract<mode>): Renamed to ...
18930 (vec_extract<mode><VEC_base_l>): ... this.
18931 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18932 (vec_initv2sfsf): ... this.
18933 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18934 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18935 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18936 element mode after mode in gen_vec_init* calls.
18937 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18938 (vec_init<mode><Vel>): ... this.
18939 (vec_extract<mode>): Renamed to ...
18940 (vec_extract<mode><Vel>): ... this.
18941 * config/aarch64/iterators.md (Vel): New mode attribute.
18942 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18943 Add element mode after mode in gen_vec_extract* calls.
18944 * config/s390/vector.md (non_vec_l): New mode attribute.
18945 (vec_extract<mode>): Renamed to ...
18946 (vec_extract<mode><non_vec_l>): ... this.
18947 (vec_init<mode>): Renamed to ...
18948 (vec_init<mode><non_vec_l>): ... this.
18949 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18950 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18952 * config/arm/iterators.md (V_elem_l): New mode attribute.
18953 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18954 (vec_extract<mode><V_elem_l>): ... this.
18955 (vec_extractv2di): Renamed to ...
18956 (vec_extractv2didi): ... this.
18957 (vec_init<mode>): Renamed to ...
18958 (vec_init<mode><V_elem_l>): ... this.
18959 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18960 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18961 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18962 Add element mode after gen_vec_extract* calls.
18963 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18964 (vec_init<mode><unitmode>): ... this.
18965 (vec_extract<mode>): Renamed to ...
18966 (vec_extract<mode><unitmode>): ... this.
18967 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18968 (vec_init<mode><unitmode>): ... this.
18969 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18970 (vec_initv2sfsf): ... this.
18971 (vec_extractv2sf): Renamed to ...
18972 (vec_extractv2sfsf): ... this.
18973 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18974 Add element mode after gen_vec_extract* calls.
18975 * config/mips/mips.md (unitmode): New mode iterator.
18976 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18977 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18978 * config/spu/spu.md (inner_l): New mode attribute.
18979 (vec_init<mode>): Renamed to ...
18980 (vec_init<mode><inner_l>): ... this.
18981 (vec_extract<mode>): Renamed to ...
18982 (vec_extract<mode><inner_l>): ... this.
18983 * config/sparc/sparc.md (veltmode): New mode iterator.
18984 (vec_init<VMALL:mode>): Renamed to ...
18985 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18986 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18987 (vec_initv2sisi): ... this.
18988 (vec_initv2sf): Renamed to ...
18989 (vec_initv2sfsf): ... this.
18990 (vec_extractv2sf): Renamed to ...
18991 (vec_extractv2sfsf): ... this.
18992 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18993 (vec_init<mode>): Renamed to ...
18994 (vec_init<mode><VEC_base_l>): ... this.
18995 (vec_extract<mode>): Renamed to ...
18996 (vec_extract<mode><VEC_base_l>): ... this.
18997 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18998 (vec_initv2sfsf): ... this.
18999 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19000 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19001 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19002 gen_vec_init* calls.
19004 2017-08-01 Richard Biener <rguenther@suse.de>
19006 PR tree-optimization/81297
19007 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19008 TREE_OVERFLOW from INTEGER_CSTs.
19010 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19012 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19014 2017-07-31 Carl Love <cel@us.ibm.com>
19016 * config/rs6000/rs6000-c: Add support for built-in functions
19017 vector signed char vec_xl_be (signed long long, signed char *);
19018 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19019 vector signed int vec_xl_be (signed long long, signed int *);
19020 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19021 vector signed long long vec_xl_be (signed long long, signed long long *);
19022 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19023 vector signed short vec_xl_be (signed long long, signed short *);
19024 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19025 vector double vec_xl_be (signed long long, double *);
19026 vector float vec_xl_be (signed long long, float *);
19027 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19028 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19029 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19031 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19032 (altivec_expand_builtin): Add switch statement to call
19033 altivec_expand_xl_be for each builtin.
19034 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19035 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19036 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19037 __builtin_vsx_le_be_v16qi.
19038 * doc/extend.texi: Update the built-in documentation file for the
19039 new built-in functions.
19041 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19044 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19046 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19048 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19050 * config.gcc: Add z14.
19051 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19052 CPU model numbers for z13s and z14.
19053 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19055 * config/s390/s390-opts.h (enum processor_type): Rename
19056 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19057 * config/s390/s390.c (processor_table): Add field for CPU name to
19058 be passed to Binutils.
19059 (s390_asm_output_machine_for_arch): Use the new field in
19060 processor_table for Binutils.
19061 (s390_expand_builtin): Replace arch12 with z14.
19062 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19063 (s390_get_sched_attrmask): Likewise.
19064 (s390_get_unit_mask): Likewise.
19065 * config/s390/s390.opt: Add z14 to processor_type enum.
19067 2017-07-31 Martin Jambor <mjambor@suse.cz>
19070 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19071 regardless of optimization level.
19073 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19074 Martin Liska <mliska@suse.cz>
19076 * predict.def: Remove old comment and adjust probability.
19077 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19078 PREDICT statements.
19080 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19083 * config/i386/i386.c (ix86_function_naked): New function.
19084 (ix86_can_use_return_insn_p): Return false for naked functions.
19085 (ix86_expand_prologue): Skip prologue for naked functions.
19086 (ix86_expand_epilogue): Skip epilogue for naked functions
19087 and emit trap instruction.
19088 (ix86_warn_func_return): New function.
19089 (ix86_attribute_table): Add "naked" attribute specification.
19090 (TARGET_WARN_FUNC_RETURN): Define.
19091 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19093 2017-07-31 Martin Liska <mliska@suse.cz>
19095 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19096 (dump_gimple_bb_header): Always dump BB info.
19097 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19099 2017-07-31 Martin Liska <mliska@suse.cz>
19102 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19103 also with current_function_decl non-null equality.
19105 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19108 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19109 change type to the element type, instead add eltype variable and
19110 use it where we are interested in the element type.
19112 PR tree-optimization/81603
19113 * ipa-polymorphic-call.c
19114 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19115 offset arithmetic in offset_int, bail out if the resulting bit offset
19116 doesn't fit into shwi.
19118 2017-07-31 Martin Liska <mliska@suse.cz>
19120 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19121 (gimplify_save_expr): Fix comment.
19123 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19126 * config/i386/i386.c (ix86_function_arg): Update arguments for
19128 (ix86_compute_frame_layout): Set the initial stack offset to
19129 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19130 INCOMING_FRAME_SP_OFFSET.
19131 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19132 stack before exception handler returns.
19133 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19134 the 'ERROR_CODE' for exception handler.
19136 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19138 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19139 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19140 (ASM_OUTPUT_REG_POP): Ditto.
19141 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19142 instead of asm_fprintf to output pure string.
19144 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19146 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19147 to imported_module_or_decl hook.
19148 (debug_nothing_tree_tree_tree_bool): Remove.
19149 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19150 * debug.c (do_nothing_debug_hooks): Use
19151 debug_nothing_tree_tree_tree_bool_bool instead of
19152 debug_nothing_tree_tree_tree_bool.
19153 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19154 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19155 * sdbout.c (sdb_debug_hooks): Likewise.
19156 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19157 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19159 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19160 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19161 attribute, don't add anything.
19163 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19165 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19166 (fold_build2_stat_loc): Likewise.
19167 (fold_build3_stat_loc): Likewise.
19168 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19169 (fold_build1_loc): Remove macro.
19170 (fold_build2_loc): Likewise.
19171 (fold_build3_loc): Likewise.
19173 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19175 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19176 (gimple_build_debug_bind_source_stat): Likewise.
19177 * gimple.h (gimple_build_debug_bind): Remove macro.
19178 (gimple_build_debug_bind_source): Likewise.
19180 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19182 * bitmap.c (bitmap_alloc): Adjust.
19183 (bitmap_gc_alloc): Likewise.
19184 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19186 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19188 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19189 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19190 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19191 (bitmap_gc_alloc_stat): Likewise.
19192 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19196 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19197 * rtl.h (shallow_copy_rtx): Remove macro.
19199 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19201 * emit-rtl.c (gen_raw_REG): Adjust.
19202 * gengenrtl.c (gendef): Likewise.
19203 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19204 * rtl.h (rtx_alloc): Remove macro.
19206 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19208 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19209 (build_tree_list_stat): Likewise.
19210 * tree.h (build_tree_list): Remove macro.
19211 (build_tree_list_vec): Likewise.
19213 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19215 * tree.c (make_vector_stat): Remove _stat from name.
19216 (build_vector_stat): Likewise.
19217 * tree.h (make_vector_stat): Remove macro.
19218 (build_vector_stat): Likewise.
19220 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19222 * tree.h (build_var_debug_value): Remove prototype.
19224 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19226 * tree.c (tree_cons_stat): Remove _stat from name.
19227 * tree.h (tree_cons): Remove macro.
19229 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19231 * tree.c (build_vl_exp_stat): Remove _stat from name.
19232 * tree.h (build_vl_exp): Remove macro.
19234 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19236 * tree.c (build_decl_stat): Remove _stat from name.
19237 * tree.h (build_decl): Remove macro.
19239 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19241 * gimple.c (gimple_build_with_ops_stat): Adjust.
19242 (gimple_alloc_stat): Remove _stat from name.
19243 * gimple.h (gimple_alloc): Remove macro.
19245 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19247 * tree.c (make_tree_vec_stat): Remove _stat from name.
19248 (grow_tree_vec_stat): Likewise.
19249 * tree.h (make_tree_vec_stat): Adjust prototype.
19250 (grow_tree_vec_stat): Likewise.
19251 (make_tree_vec): Remove macro.
19252 (grow_tree_vec): Likewise.
19254 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19256 * fold-const.c (fold_build1_stat_loc): Adjust.
19257 (fold_build2_stat_loc): Likewise.
19258 (fold_build3_stat_loc): Likewise.
19259 * tree.c (build0_stat): Remove _stat from name.
19260 (build1_stat): Likewise.
19261 (build2_stat): Likewise.
19262 (build3_stat): Likewise.
19263 (build4_stat): Likewise.
19264 (build5_stat): Likewise.
19265 * tree.h (build1_loc): Remove macro, and rename _stat function
19267 (build2_loc): Likewise.
19268 (build3_loc): Likewise.
19269 (build4_loc): Likewise.
19270 (build5_loc): Likewise.
19272 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19274 * tree.c (make_int_cst_stat): Remove _stat from name.
19275 * tree.h (make_int_cst_stat): Adjust prototype.
19276 (make_int_cst): Remove macro.
19278 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19280 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19281 * tree.h (make_tree_binfo_stat): Adjust prototype.
19282 (make_tree_binfo): Remove.
19284 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19286 * tree.c (copy_node_stat): Rename to copy_node.
19287 (build_distinct_type_copy): Adjust.
19288 * tree.h (copy_node_stat): Adjust prototype.
19289 (copy_node): Remove macro.
19291 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19293 * tree.c (make_node_stat): rename to make_node.
19294 (build_tree_list_stat): Adjust.
19295 (build0_stat): Likewise.
19296 (build2_stat): Likewise.
19297 (build3_stat): Likewise.
19298 (build4_stat): Likewise.
19299 (build5_stat): Likewise.
19300 (build_decl_stat): Likewise.
19301 * tree.h (make_node_stat): Adjust prototype.
19302 (make_node): remove macro.
19304 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19306 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19307 (PPC_FEATURE2_SCV): Likewise.
19308 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19310 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19312 * config/aarch64/aarch64.c
19313 (aarch64_internal_mov_immediate): Add new special pattern.
19314 * config/aarch64/aarch64.md (*movdi_aarch64):
19315 Add reg/32bit const mov case.
19317 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19318 Richard Sandiford <richard.sandiford@linaro.org>
19320 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19321 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19322 Add integer and movi cases.
19323 (movi-split-hf-df-sf split, fp16): New.
19324 (enabled): Added TARGET_FP_F16INST.
19325 * config/aarch64/iterators.md (GPF_HF): New.
19326 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19328 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19330 * config/aarch64/aarch64.c
19331 (aarch64_simd_container_mode): Add prototype.
19332 (aarch64_expand_mov_immediate): Add HI support.
19333 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19334 (aarch64_can_const_movi_rtx_p): New.
19335 (aarch64_preferred_reload_class):
19336 Remove restrictions of using FP registers for certain SIMD operations.
19337 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19338 (aarch64_valid_floating_const): Add integer move validation.
19339 (aarch64_simd_imm_scalar_p): Remove.
19340 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19341 (aarch64_legitimate_constant_p): Expand list of supported cases.
19342 * config/aarch64/aarch64-protos.h
19343 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19344 (aarch64_reinterpret_float_as_int): New.
19345 (aarch64_simd_imm_scalar_p): Remove.
19346 * config/aarch64/constraints.md (Uvi): New.
19347 (Dd): Split into Ds and new Dd.
19348 * config/aarch64/aarch64.md (*movsi_aarch64):
19350 (*movdi_aarch64): Add SIMD mov case.
19352 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19354 * tree-predcom.c: (struct chain): Handle store-store chain in which
19355 stores for elimination only store loop invariant values.
19356 (execute_pred_commoning_chain): Ditto.
19357 (prepare_initializers_chain_store_elim): Ditto.
19358 (prepare_finalizers): Ditto.
19359 (is_inv_store_elimination_chain): New function.
19360 (initialize_root_vars_store_elim_1): New function.
19362 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19364 * tree-predcom.c: Revise general description of the pass.
19365 (enum chain_type): New enum type for store elimination.
19366 (struct chain): New field supporting store elimination.
19367 (struct component): Ditto.
19368 (dump_chain): Dump store-stores chain.
19369 (release_chain): Release resources.
19370 (split_data_refs_to_components): Compute and create component
19371 contains only stores for elimination.
19372 (get_chain_last_ref_at): New function.
19373 (make_invariant_chain): Initialization.
19374 (make_rooted_chain): Specify chain type in parameter and record it.
19375 (add_looparound_copies): Skip for store-stores chain.
19376 (determine_roots_comp): Compute type of chain and pass it to
19378 (initialize_root_vars_store_elim_2): New function.
19379 (finalize_eliminated_stores): New function.
19380 (remove_stmt): Handle store for elimination.
19381 (execute_pred_commoning_chain): Execute predictive commoning on
19382 store-store chains.
19383 (determine_unroll_factor): Skip unroll for store-stores chain.
19384 (prepare_initializers_chain_store_elim): New function.
19385 (prepare_initializers_chain): Hanlde store-store chain.
19386 (prepare_finalizers_chain, prepare_finalizers): New function.
19387 (tree_predictive_commoning_loop): Return integer value indicating
19388 if loop is unrolled or lcssa form is corrupted.
19389 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19391 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19393 * tree-predcom.c (initialize_root): Delete.
19394 (execute_pred_commoning_chain): Initialize root vars and replace
19395 reference of non-combined chain directly, rather than call above
19398 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19400 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19401 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19403 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19405 * tree-predcom.c (struct chain): New field init_seq.
19406 (release_chain): Release init_seq.
19407 (prepare_initializers_chain): Record intialization stmts in above
19409 (insert_init_seqs): New function.
19410 (tree_predictive_commoning_loop): Call insert_init_seqs.
19412 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19414 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19416 2017-07-28 Richard Biener <rguenther@suse.de>
19418 * match.pd: Remove superfluous :c.
19419 * genmatch.c (simplify::id): Add member.
19420 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19422 (current_id): New global.
19423 (dt_node::parent): Move from ...
19424 (dt_operand::parent): ... here. Add for_id member.
19425 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19426 (decision_tree::find_node): Relax order requirement when
19427 merging DT_TRUE nodes to ones inbetween the current simplify
19428 and the one we try to merge with. Add diagnostic whenever
19429 we need to enforce pattern order by not merging.
19430 (decision_tree::insert): Set current_id.
19431 (decision_tree::print_node): Dump parent node and for_id.
19432 (parser::last_id): Add member.
19433 (parser::push_simplify): Assign unique id.
19434 (parser::parser): Initialize last_id.
19436 2017-07-28 Martin Liska <mliska@suse.cz>
19439 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19440 gimple_build_debug_bind.
19442 2017-07-28 Richard Biener <rguenther@suse.de>
19444 PR tree-optimization/81502
19445 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19447 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19449 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19450 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19451 for BIT_FIELD_REF args.
19452 * fold-const.c (make_bit_field_ref): Likewise.
19453 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19455 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19458 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19459 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19460 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19461 Or it into SANITIZER_UNDEFINED.
19462 * ubsan.c: Include gimple-fold.h and varasm.h.
19463 (ubsan_expand_ptr_ifn): New function.
19464 (instrument_pointer_overflow): New function.
19465 (maybe_instrument_pointer_overflow): New function.
19466 (instrument_object_size): Formatting fix.
19467 (pass_ubsan::execute): Call instrument_pointer_overflow
19468 and maybe_instrument_pointer_overflow.
19469 * internal-fn.c (expand_UBSAN_PTR): New function.
19470 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19471 * sanitizer.def (__ubsan_handle_pointer_overflow,
19472 __ubsan_handle_pointer_overflow_abort): New builtins.
19473 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19474 * internal-fn.def (UBSAN_PTR): New internal function.
19475 * opts.c (sanitizer_opts): Add pointer-overflow.
19476 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19477 * fold-const.c (build_range_check): Compute pointer range check in
19478 integral type if pointer arithmetics would be needed. Formatting
19481 2017-07-28 Martin Liska <mliska@suse.cz>
19484 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19485 parameters that are of a variable-length.
19487 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19489 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19491 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19492 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19493 (CRT_CALL_STATIC_FUNCTION): Likewise.
19494 (ASM_DEFAULT_SPEC): New define.
19495 (ASM_SPEC32): Likewise.
19496 (ASM_SPEC64): Likewise.
19497 (ASM_SPEC_COMMON): Likewise.
19498 (ASM_SPEC): Likewise.
19499 (INVALID_64BIT): Likewise.
19500 (LINK_OS_DEFAULT_SPEC): Likewise.
19501 (LINK_OS_SPEC32): Likewise.
19502 (LINK_OS_SPEC64): Likewise.
19503 (POWERPC_LINUX): Likewise.
19504 (PTRDIFF_TYPE): Likewise.
19505 (RESTORE_FP_PREFIX): Likewise.
19506 (RESTORE_FP_SUFFIX): Likewise.
19507 (SAVE_FP_PREFIX): Likewise.
19508 (SAVE_FP_SUFFIX): Likewise.
19509 (SIZE_TYPE): Likewise.
19510 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19511 (TARGET_64BIT): Likewise.
19512 (TARGET_64BIT): Likewise.
19513 (TARGET_AIX): Likewise.
19514 (WCHAR_TYPE_SIZE): Likewise.
19515 (WCHAR_TYPE): Undefine.
19516 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19517 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19518 (CPP_OS_RTEMS_SPEC): Delete.
19519 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19520 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19522 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19524 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19526 PR tree-optimization/81578
19527 * tree-parloops.c (build_new_reduction): Bail out if
19528 reduction_code isn't one of the standard OpenMP reductions.
19529 Move the details printing after that decision.
19531 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19533 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19534 related to reload_in_progress.
19535 (splat_input_operand): Likewise.
19536 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19538 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19540 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19541 (TARGET_INSTANTIATE_DECLS): Likewise.
19542 (legitimate_indexed_address_p): Delete reload_in_progress code.
19543 (rs6000_debug_legitimate_address_p): Likewise.
19544 (rs6000_eliminate_indexed_memrefs): Likewise.
19545 (rs6000_emit_le_vsx_store): Likewise.
19546 (rs6000_emit_move_si_sf_subreg): Likewise.
19547 (rs6000_emit_move): Likewise.
19548 (register_to_reg_type): Likewise.
19549 (rs6000_pre_atomic_barrier): Likewise.
19550 (rs6000_machopic_legitimize_pic_address): Likewise.
19551 (rs6000_allocate_stack_temp): Likewise.
19552 (rs6000_address_for_fpconvert): Likewise.
19553 (rs6000_address_for_altivec): Likewise.
19554 (rs6000_secondary_memory_needed_rtx): Delete function.
19555 (rs6000_check_sdmode): Likewise.
19556 (rs6000_alloc_sdmode_stack_slot): Likewise.
19557 (rs6000_instantiate_decls): Likewise.
19558 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19559 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19560 Delete reload_in_progress.
19561 (*vec_reload_and_plus_<mptrsize>): Likewise.
19562 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19563 (vsx_div_v2di): Likewise.
19564 (vsx_udiv_v2di): Likewise.
19566 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19568 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19569 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19570 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19571 (rs6000_debug_reg_global): Delete print of LRA status.
19572 (rs6000_option_override_internal): Delete dead LRA related code.
19573 (rs6000_lra_p): Delete function.
19574 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19576 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19578 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19579 * config/riscv/rtems.h: New file.
19581 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19582 Sudakshina Das <sudi.das@arm.com>
19584 * config/aarch64/aarch64.md
19585 (define_split for and<mode>3nr_compare): Move
19586 non aarch64_logical_operand to a register.
19587 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19588 register immediate operand to a register.
19589 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19591 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19593 PR middle-end/81564
19594 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19596 2017-07-27 Richard Biener <rguenther@suse.de>
19598 PR tree-optimization/81573
19599 PR tree-optimization/81494
19600 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19601 multi defuse cycle case.
19603 2017-07-27 Richard Biener <rguenther@suse.de>
19605 PR tree-optimization/81571
19606 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19609 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19611 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19612 earlier and only if MASK_FPU is set. Adjust formatting.
19614 2017-07-27 Martin Liska <mliska@suse.cz>
19616 * opt-functions.awk: Add validation of value of Init.
19617 * optc-gen.awk: Pass new argument.
19619 2017-07-27 Martin Liska <mliska@suse.cz>
19621 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19622 Fix wrong condition.
19624 2017-07-27 Martin Liska <mliska@suse.cz>
19626 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19627 BBs and edges seen by autoFDO.
19629 2017-07-27 Richard Biener <rguenther@suse.de>
19631 PR tree-optimization/81502
19632 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19633 with incompatible but same sized type.
19634 (execute_update_addresses_taken): Likewise.
19636 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19638 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19639 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19641 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19644 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19645 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19646 Change s_operand to memory_operand.
19648 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19650 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19651 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19652 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19653 Emit instructions rather than returning an expression. Handle TFmode
19654 and KFmode by casting to TImode.
19655 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19656 (rs6000_emit_le_vsx_store): Likewise.
19657 * config/rs6000/vsx.md (VSX_TI): New iterator.
19658 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19659 (*vsx_le_undo_permute_<mode>): Likewise.
19660 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19661 emit the split sequence.
19662 (*vsx_le_perm_store_<mode>): Likewise.
19664 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19666 PR tree-optimization/81555
19667 PR tree-optimization/81556
19668 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19669 if true, force CHANGED for the recursive invocation.
19670 (reassociate_bb): Remember original length of ops array, pass
19671 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19673 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19674 attributes for noipa attribute. For naked attribute use
19675 lookup_attribute first before lookup_attribute_spec.
19676 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19678 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19679 for functions with noipa attribute.
19680 (cgraph_externally_visible_p): Return true for functions with noipa
19682 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19683 for functions with noipa attribute.
19684 * doc/extend.texi: Document noipa function attribute.
19685 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19686 also for functions with noipa attribute.
19687 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19689 2017-07-26 Andrew Pinski <apinski@cavium.com>
19691 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19692 vec_unalign_load_cost and vec_unalign_store_cost.
19694 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19696 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19697 -mvsx-small-integer option.
19698 (ISA_3_0_MASKS_IEEE): Likewise.
19699 (OTHER_VSX_VECTOR_MASKS): Likewise.
19700 (POWERPC_MASKS): Likewise.
19701 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19702 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19703 code, only testing for DImode being allowed in non-VSX floating
19705 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19706 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19708 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19709 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19710 TARGET_P8_VECTOR test.
19711 (rs6000_secondary_reload_simple_move): Likewise.
19712 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19713 since TARGET_P9_VECTOR was already tested.
19714 (rs6000_opt_masks): Remove -mvsx-small-integer.
19715 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19716 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19718 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19719 test for TARGET_VEXTRACTUB was used, and that uses
19721 (p9 extract splitter): Likewise.
19722 (vsx_extract_<mode>_di_p9): Likewise.
19723 (vsx_extract_<mode>_store_p9): Likewise.
19724 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19725 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19726 the elimination of TARGET_VSX_SMALL_INTEGER.
19727 (vsx_extract_<mode>_p8): Likewise.
19728 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19729 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19730 (vsx_set_<mode>_p9): Likewise.
19731 (vsx_set_v4sf_p9): Likewise.
19732 (vsx_set_v4sf_p9_zero): Likewise.
19733 (vsx_insert_extract_v4sf_p9): Likewise.
19734 (vsx_insert_extract_v4sf_p9_2): Likewise.
19735 * config/rs6000/rs6000.md (sign extend splitter): Change
19736 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19737 (floatsi<mode>2_lfiwax_mem): Likewise.
19738 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19739 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19740 since a test for TARGET_P9_VECTOR was used.
19741 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19742 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19743 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19744 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19745 TARGET_P8_VECTOR test.
19746 (fix_trunc<mode>si2_stfiwx): Likewise.
19747 (fix_trunc<mode>si2_internal): Likewise.
19748 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19749 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19751 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19752 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19753 TARGET_P8_VECTOR test.
19754 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19755 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19756 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19758 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19759 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19760 since a test for TARGET_P9_VECTOR was used.
19761 (splitter for loading small constants): Likewise.
19763 2017-07-26 Andrew Pinski <apinski@cavium.com>
19765 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19768 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19771 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19772 (fp_valid_at): Likewise.
19774 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19776 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19777 (qdf24xx_addrcost_table): Likewise.
19778 (cortexa57_tunings): Update to use generic_branch_cost.
19779 (cortexa72_tunings): Likewise.
19780 (cortexa73_tunings): Likewise.
19781 (qdf24xx_tunings): Likewise.
19783 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19785 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19786 (thunderx2t99_branch_cost): Likewise.
19787 (cortexa35_tunings): Update to use generic_branch_cost.
19788 (cortexa53_tunings): Likewise.
19789 (cortexa57_tunings): Likewise.
19790 (cortexa72_tunings): Likewise.
19791 (cortexa73_tunings): Likewise.
19792 (thunderx2t99_tunings): Likewise.
19794 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19796 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19797 (sparc_option_override): Honour MASK_FSMULD.
19798 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19799 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19800 * config/sparc/sparc.opt (mfsmuld): New option.
19801 * doc/invoke.texi (mfsmuld): Document option.
19803 2017-07-26 Marek Polacek <polacek@redhat.com>
19805 PR middle-end/70992
19806 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19808 2017-07-26 Richard Biener <rguenther@suse.de>
19810 * gimple-match-head.c (do_valueize): Return OP if valueize
19812 (get_def): New helper to get at the def stmt of a SSA name
19813 if valueize allows.
19814 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19815 do_valueize to get at the def stmt.
19816 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19818 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19820 PR middle-end/46932
19821 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19823 2017-07-26 Martin Liska <mliska@suse.cz>
19826 * function.c (expand_function_start): Make expansion of
19827 nonlocal_goto_save_area after parm_birth_insn.
19829 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19831 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19832 from all CPU target flags enable members.
19834 2017-07-26 Richard Biener <rguenther@suse.de>
19836 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19837 (decision_tree::gen): Make 'type' const.
19838 (write_predicate): Likewise.
19840 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19842 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19843 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19844 (rs6000_option_override_internal): Likewise.
19845 (rs6000_expand_vector_set): Likewise.
19846 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19847 (TARGET_UPPER_REGS_SF): Likewise.
19848 (TARGET_UPPER_REGS_DI): Likewise.
19849 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19850 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19851 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19852 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19853 (Splitters for DI constants in Altivec registers): Likewise.
19854 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19855 (vsx_set_v4sf_p9): Likewise.
19856 (vsx_set_v4sf_p9_zero): Likewise.
19857 (vsx_insert_extract_v4sf_p9): Likewise.
19858 (vsx_insert_extract_v4sf_p9_2): Likewise.
19860 2017-07-25 Carl Love <cel@us.ibm.com>
19862 * doc/extend.texi: Update the built-in documentation file for the
19863 existing built-in functions
19864 vector signed char vec_cnttz (vector signed char);
19865 vector unsigned char vec_cnttz (vector unsigned char);
19866 vector signed short vec_cnttz (vector signed short);
19867 vector unsigned short vec_cnttz (vector unsigned short);
19868 vector signed int vec_cnttz (vector signed int);
19869 vector unsigned int vec_cnttz (vector unsigned int);
19870 vector signed long long vec_cnttz (vector signed long long);
19871 vector unsigned long long vec_cnttz (vector unsigned long long);
19873 2017-07-25 Andrew Pinski <apinski@cavium.com>
19875 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19876 accesses where the use is for the first operand of a BIT_INSERT.
19878 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19881 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19882 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19884 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19886 * config/i386/gstabs.h: Delete.
19887 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19889 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19891 * config/i386/i386.c (ix86_decompose_address): Do not check for
19892 register RTX when looking at index_reg or base_reg.
19893 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19895 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19897 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19898 to update EH info here.
19900 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19902 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19904 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19906 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19908 2017-07-25 Torsten Duwe <duwe@suse.de>
19910 * common.opt: Introduce -fpatchable-function-entry
19911 command line option, and its variables function_entry_patch_area_size
19912 and function_entry_patch_area_start.
19913 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19914 including a two-value parser.
19915 * target.def (print_patchable_function_entry): New target hook.
19916 * targhooks.h (default_print_patchable_function_entry): New function.
19917 * targhooks.c (default_print_patchable_function_entry): Likewise.
19918 * toplev.c (process_options): Switch off IPA-RA if
19919 patchable function entries are being generated.
19920 * varasm.c (assemble_start_function): Look at the
19921 patchable-function-entry command line switch and current
19922 function attributes and maybe generate NOP instructions by
19923 calling the print_patchable_function_entry hook.
19924 * doc/extend.texi: Document patchable_function_entry attribute.
19925 * doc/invoke.texi: Document -fpatchable_function_entry
19926 command line option.
19927 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19929 * doc/tm.texi: Re-generate.
19931 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19934 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19935 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19937 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19939 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19941 2017-07-25 Richard Biener <rguenther@suse.de>
19943 PR tree-optimization/81455
19944 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19945 not walk in cycles when looking for guards.
19947 2017-07-25 Richard Biener <rguenther@suse.de>
19949 PR tree-optimization/81529
19950 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19951 when optimizing backedge uses.
19953 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19955 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19957 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19958 to dl_section_ref. On AIX, append an expression to subtract
19959 the size of the section length to dl_section_ref.
19961 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19963 * configure.ac: If any of the config.* scripts fail, exit 1.
19964 * configure: Regenerate.
19966 2017-07-25 Richard Biener <rguenther@suse.de>
19968 PR middle-end/81546
19969 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19970 of immediate uses, be more verbose on errors.
19972 2017-07-25 Richard Biener <rguenther@suse.de>
19974 PR tree-optimization/81510
19975 * tree-vect-loop.c (vect_is_simple_reduction): When the
19976 reduction stmt is not inside the loop bail out.
19978 2017-07-25 Richard Biener <rguenther@suse.de>
19980 PR tree-optimization/81303
19981 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19982 profitability check against LOOP_VINFO_NITERSM1.
19984 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19986 * domwalk.c (cmp_bb_postorder): Simplify.
19987 (sort_bbs_postorder): New function. Use it...
19988 (dom_walker::walk): ...here to optimize common cases.
19990 2017-07-25 Martin Liska <mliska@suse.cz>
19993 * ipa-visibility.c (function_and_variable_visibility): Make the
19994 redirection just on target that supports aliasing.
19995 Fix GNU coding style.
19997 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20000 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20002 * config/arm/rtems.h: Likewise.
20003 * config/bfin/rtems.h: Likewise.
20004 * config/i386/rtemself.h: Likewise.
20005 * config/lm32/rtems.h: Likewise.
20006 * config/m32c/rtems.h: Likewise.
20007 * config/m68k/rtemself.h: Likewise.
20008 * config/microblaze/rtems.h: Likewise.
20009 * config/mips/rtems.h: Likewise.
20010 * config/moxie/rtems.h: Likewise.
20011 * config/nios2/rtems.h: Likewise.
20012 * config/powerpcspe/rtems.h: Likewise.
20013 * config/rs6000/rtems.h: Likewise.
20014 * config/rtems.h: Likewise.
20015 * config/sh/rtems.h: Likewise.
20016 * config/sh/rtemself.h: Likewise.
20017 * config/sparc/rtemself.h: Likewise.
20019 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20022 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20023 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20024 * tree-ssa-structalias.c (alias_get_name): Same.
20026 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20029 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20030 instructions if no du chain is found.
20032 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20034 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20036 2017-07-25 Richard Biener <rguenther@suse.de>
20038 PR middle-end/81505
20039 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20042 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20044 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20045 upper-regs options.
20046 (ISA_2_7_MASKS_SERVER): Likewise.
20047 (ISA_3_0_MASKS_IEEE): Likewise.
20048 (OTHER_P8_VECTOR_MASKS): Likewise.
20049 (OTHER_VSX_VECTOR_MASKS): Likewise.
20050 (POWERPC_MASKS): Likewise.
20051 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20052 duplicate list of options.
20053 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20054 explicit -mupper-regs options.
20055 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20056 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20057 alias for -mupper-regs-df.
20058 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20059 (rs6000_init_hard_regno_mode_ok): Likewise.
20060 (rs6000_option_override_internal): Likewise.
20061 (rs6000_opt_masks): Likewise.
20062 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20063 options in terms of whether -mvsx or -mpower8-vector was used.
20064 (TARGET_UPPER_REGS_DI): Likewise.
20065 (TARGET_UPPER_REGS_SF): Likewise.
20066 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20067 -mupper-regs-* options.
20069 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20071 * passes.c (emergency_dump_function): Print some empty lines and a
20072 header before the RTL dump.
20074 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20076 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20078 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20081 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20082 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20084 2017-07-24 Carl Love <cel@us.ibm.com>
20086 * config/rs6000/rs6000-c.c: Add support for built-in functions
20087 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20088 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20089 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20090 vec_extract_fp_from_shortl): Add defines for the two builtins.
20091 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20092 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20094 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20095 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20096 * doc/extend.texi: Update the built-in documentation file for the
20097 new built-in function.
20099 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20102 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20104 * doc/generic.texi: Likewise.
20105 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20106 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20108 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20110 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20111 (aarch64_mls_elt_merge<mode>): Likewise.
20113 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20115 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20116 having __cxa_atexit.
20118 2017-07-23 Michael Collison <michael.collison@arm.com>
20120 * config/arm/arm.c (arm_option_override): Deprecate
20121 use of -mstructure-size-boundary.
20122 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20123 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20125 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20128 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20129 Reduce cost estimate for direct moves.
20131 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20134 * config/i386/i386.c (ix86_option_override_internal): Disable
20135 BMI, BMI2 and TBM instructions for -m16.
20137 2017-07-21 Carl Love <cel@us.ibm.com>
20139 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20140 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20141 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20142 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20143 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20144 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20145 VMULOSW): New enum "unspec" values.
20146 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20147 altivec_vmulosw): New patterns.
20148 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20149 VMULOSW): Add definitions.
20151 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20153 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20154 (qdf24xx): Likewise.
20155 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20156 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20157 (AARCH64_FL_V8_1): Renumber.
20158 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20159 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20160 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20161 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20162 rdma to feature modifiers list.
20164 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20166 PR middle-end/56727
20167 * ipa-visibility (function_and_variable_visibility): Convert
20168 recursive PLT call to direct call if appropriate.
20170 2017-07-21 Andrew Pinski <apinski@cavium.com>
20172 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20173 operand 1 to see if the types precision matches.
20174 * fold-const.c (operand_equal_p): Likewise.
20176 2017-07-21 Richard Biener <rguenther@suse.de>
20178 PR tree-optimization/81303
20179 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20180 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20181 (vect_peeling_hash_get_lowest_cost): Adjust.
20182 (vect_enhance_data_refs_alignment): Likewise. Use
20183 vect_get_peeling_costs_all_drs to compute the penalty for no
20184 peeling to match up costs.
20186 2017-07-21 Richard Biener <rguenther@suse.de>
20188 PR tree-optimization/81500
20189 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20190 we didn't identify a reduction path.
20192 2017-07-21 Tom de Vries <tom@codesourcery.com>
20193 Cesar Philippidis <cesar@codesourcery.com>
20195 PR gcov-profile/81442
20196 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20199 2017-07-21 Tom de Vries <tom@codesourcery.com>
20202 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20204 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20205 nvptx_override_options_after_change.
20207 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20209 * dwarf2out.c (output_file_names): Avoid double testing for
20210 dwarf_version >= 5.
20212 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20214 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20216 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20218 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20220 (try_crossjump_to_edge): Do not punt on partitioned functions.
20222 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20224 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20225 Put all BBs reachable only via paths crossing cold region to cold
20227 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20229 2016-07-21 Richard Biener <rguenther@suse.de>
20231 PR tree-optimization/81303
20232 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20233 into account prologue and epilogue iterations when raising
20234 min_profitable_iters to sth at least covering one vector iteration.
20236 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20238 * config/arm/arm.c (arm_test_cpu_arch_dat):
20241 2017-07-20 Nathan Sidwell <nathan@acm.org>
20243 Remove TYPE_METHODS.
20244 * tree.h (TYPE_METHODS): Delete.
20245 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20246 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20247 (dbxout_type_methods): Scan TYPE_FIELDS.
20248 (dbxout_type): Don't check TYPE_METHODS here.
20249 * function.c (use_register_for_decl): Always ignore register for
20250 class types when not optimizing.
20251 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20252 * tree.c (free_lang_data_in_type): Stitch out member functions and
20253 templates from TYPE_FIELDS.
20254 (build_distinct_type_copy, verify_type_variant,
20255 verify_type): Member fns are on TYPE_FIELDS.
20256 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20257 * tree-pretty-print.c (dump_generic_node): Likewise.
20259 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20262 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20263 V2TImode and V4TImode.
20264 (ix86_expand_vector_extract): Likewise.
20265 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20266 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20267 (ssescalarmode): Handle V4TImode and V2TImode.
20268 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20269 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20270 (VEXTRACTI128_MODE): New mode iterator.
20271 (splitter for *vec_extractv?ti first element): New.
20272 (VEC_INIT_MODE): New mode iterator.
20273 (vec_init<mode>): Consolidate 3 expanders into one using
20274 VEC_INIT_MODE mode iterator.
20276 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20278 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20279 non_spilled_static_chain_regno_p.
20281 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20283 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20285 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20287 * bb-reorder.c (connect_traces): Allow copying of blocks within
20290 2017-07-20 Richard Biener <rguenther@suse.de>
20292 * gimple.h (gimple_phi_result): Add gphi * overload.
20293 (gimple_phi_result_ptr): Likewise.
20294 (gimple_phi_arg): Likewise. Adjust index assert to only
20295 allow actual argument accesses rather than all slots available
20297 (gimple_phi_arg_def): Add gphi * overload.
20298 * tree-phinodes.c (make_phi_node): Initialize only actual
20300 (resize_phi_node): Clear memory not covered by old node,
20301 do not initialize excess argument slots.
20302 (reserve_phi_args_for_new_edge): Initialize new argument slot
20305 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20307 PR tree-optimization/81388
20309 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20311 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20312 by removing computation of may_be_zero.
20314 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20315 Tom de Vries <tom@codesourcery.com>
20317 PR middle-end/81030
20318 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20319 when gimple level profile disagrees with what RTL expander did.
20321 2017-07-20 Richard Biener <rguenther@suse.de>
20323 PR tree-optimization/61171
20324 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20325 (vect_analyze_stmt): Add slp instance parameter.
20326 (vectorizable_reduction): Likewise.
20327 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20328 (vect_is_simple_reduction): Deal with chains not detected
20329 as SLP reduction chain, specifically not properly associated
20330 chains containing a mix of plus/minus.
20331 (get_reduction_op): Remove.
20332 (get_initial_defs_for_reduction): Simplify, pass in whether
20333 this is a reduction chain, pass in the SLP node for the PHIs.
20334 (vect_create_epilog_for_reduction): Get the SLP instance as
20336 (vectorizable_reduction): Get the SLP instance as arg.
20337 During analysis remember the SLP node with the PHIs in the
20338 instance. Simplify getting at the vectorized reduction PHIs.
20339 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20340 through SLP instance.
20341 (vect_slp_analyze_operations): Likewise.
20342 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20343 (vect_transform_stmt): Likewise.
20345 2017-07-20 Tom de Vries <tom@codesourcery.com>
20347 PR tree-optimization/81489
20348 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20349 read of phi arg location to before loop that modifies phi.
20351 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20353 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20356 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20358 PR middle-end/81331
20359 * except.c (execute): Fix ordering issue.
20361 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20363 PR rtl-optimization/81423
20364 * combine.c (make_compound_operation_int): Don't try to optimize
20365 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20367 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20369 PR rtl-optimization/81423
20370 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20371 with a constant that is -1 in the truncated to mode.
20373 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20375 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20376 (determine_unlikely_bbs): ... here.
20377 * predict.h (propagate_unlikely_bbs_forward): Declare.
20378 * cfgexpand.c (pass_expand::execute): Use it.
20379 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20381 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20382 propagate_unlikely_bbs_forward.
20384 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20386 PR middle-end/81331
20387 * except.c (maybe_add_nop_after_section_switch): New function.
20390 2017-07-19 Tom de Vries <tom@codesourcery.com>
20392 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20394 2017-07-19 Tom de Vries <tom@codesourcery.com>
20396 * gimple.h (gimple_phi_arg): Make assert more strict.
20398 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20400 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20401 * config/rs6000/mmintrin.h: New file.
20402 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20404 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20406 PR tree-optimization/81346
20407 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20409 2017-07-19 Tom de Vries <tom@codesourcery.com>
20411 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20413 2017-07-19 Tom de Vries <tom@codesourcery.com>
20415 * config/nvptx/nvptx-modes.def: Add V2DImode.
20416 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20417 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20418 (nvptx_output_mov_insn): Handle lack of mov.b128.
20419 (nvptx_print_operand): Handle 'H' and 'L' codes.
20420 (nvptx_vector_mode_supported): Allow V2DImode.
20421 (nvptx_preferred_simd_mode): New function.
20422 (nvptx_data_alignment): New function.
20423 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20424 nvptx_preferred_simd_mode.
20425 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20427 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20429 2017-07-19 Tom de Vries <tom@codesourcery.com>
20431 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20432 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20433 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20434 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20435 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20436 (mov<VECIM>_insn): New define_insn.
20437 (define_expand "mov<VECIM>): New define_expand.
20439 2017-07-19 Tom de Vries <tom@codesourcery.com>
20441 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20443 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20445 PR tree-optimization/81346
20446 * fold-const.h (fold_div_compare, range_check_type): Declare.
20447 * fold-const.c (range_check_type): New function.
20448 (build_range_check): Use range_check_type.
20449 (fold_div_compare): No longer static, rewritten into
20450 a match.pd helper function.
20451 (fold_comparison): Don't call fold_div_compare here.
20452 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20453 as helper function.
20455 2017-07-19 Nathan Sidwell <nathan@acm.org>
20457 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20458 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20459 * tree.c (find_decls_types_r, verify_type): Use
20460 TYPE_{MIN,MAX}_VALUE_RAW.
20461 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20462 (hash_tree): Likewise.
20463 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20465 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20468 2017-07-18 Tom de Vries <tom@codesourcery.com>
20470 PR middle-end/81464
20471 * omp-expand.c (expand_omp_for_static_chunk): Handle
20472 equal-argument loop exit phi.
20474 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20477 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20478 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20479 operand 2 predicate.
20480 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20481 operand 2 predicate.
20482 (ror,rol -> rorx splitters): Use const_int_operand as
20483 operand 2 predicate.
20485 2017-06-18 Richard Biener <rguenther@suse.de>
20487 PR tree-optimization/81410
20488 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20489 the gap in the ! slp_perm SLP case after each group.
20491 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20493 PR middle-end/81463
20494 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20497 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20499 PR middle-end/81462
20500 * predict.c (set_even_probabilities): Cleanup; do not affect
20501 probabilities that are already known.
20502 (combine_predictions_for_bb): Call even when count is set.
20504 2017-07-18 Nathan Sidwell <nathan@acm.org>
20506 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20509 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20512 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20513 optimization for loop niter analysis.
20515 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20518 * config/avr/avr.c (avr_optimize_casesi): Don't use
20519 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20521 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20523 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20524 body_cost_vec from _vect_peel_extended_info.
20525 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20526 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20529 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20531 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20533 2017-07-18 Richard Biener <rguenther@suse.de>
20535 PR tree-optimization/80620
20536 PR tree-optimization/81403
20537 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20538 info when re-using a VN table entry.
20540 2017-07-18 Richard Biener <rguenther@suse.de>
20542 PR tree-optimization/81418
20543 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20544 vectype_in. Verify that with lane-reducing reduction operations
20545 we have a single def-use cycle.
20547 2017-07-17 Carl Love <cel@us.ibm.com>
20549 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20551 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20552 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20553 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20554 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20555 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20556 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20557 VMULOSW): New enum "unspec" values.
20558 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20559 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20560 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20561 altivec_vmulosw): New patterns.
20562 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20563 VMULOSW): Add definitions.
20565 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20567 * config/alpha/alpha.c: Include predict.h.
20569 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20571 * tree-vrp.c (compare_assert_loc): Fix comparison function
20572 to return predictable results.
20574 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20576 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20578 (subdi3): Likewise.
20579 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20580 * doc/invoke.texi (mexpand-adddi): Update text.
20582 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20584 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20585 that also clobbers the CC register. The old expand code is moved
20587 (*arc_clzsi2): ... here.
20588 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20589 the CC register. The old expand code is moved to ...
20590 (arc_ctzsi2): ... here.
20592 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20594 * config/arc/arc.opt (mindexed-loads): Use initial value
20595 TARGET_INDEXED_LOADS_DEFAULT.
20596 (mauto-modify-reg): Use initial value
20597 TARGET_AUTO_MODIFY_REG_DEFAULT.
20598 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20599 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20600 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20601 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20603 2017-07-17 Martin Liska <mliska@suse.cz>
20606 * opts.c (finish_options): Do not allow -fgnu-tm
20607 w/ -fsanitize={kernel-,}address. Say sorry.
20609 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20612 * tree-loop-distribution.c (classify_partition): Only assert on
20613 numer of iterations.
20614 (merge_dep_scc_partitions): Delete prameter. Update function call.
20615 (distribute_loop): Remove code handling loop with unknown niters.
20616 (pass_loop_distribution::execute): Skip loop with unknown niters.
20618 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20621 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20622 function sort_partitions_by_post_order.
20624 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20626 PR tree-optimization/81374
20627 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20628 the max index of basic blocks, rather than number of basic blocks.
20630 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20632 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20634 (arc_legitimate_pic_operand_p): Likewise.
20635 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20637 (arc_needs_pcl_p): Likewise.
20638 (arc_legitimate_pc_offset_p): Likewise.
20639 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20640 function is also used in constrains.md.
20641 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20642 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20643 PLUS. Only return true/false in known cases, otherwise assert.
20644 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20645 is already called in arc_legitimate_constant_p.
20646 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20648 (LEGITIMATE_PIC_OPERAND_P): Use
20649 arc_raw_symbolic_reference_mentioned_p function.
20650 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20655 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20656 Andrew Burgess <andrew.burgess@embecosm.com>
20658 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20659 (arc_return_address_register): New function.
20660 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20661 (arc_handle_fndecl_attribute): Add naked attribute.
20662 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20663 (TARGET_WARN_FUNC_RETURN): Likewise.
20664 (arc_allocate_stack_slots_for_args): New function.
20665 (arc_warn_func_return): Likewise.
20666 (machine_function): Change type fn_type.
20667 (arc_compute_function_type): Consider new naked function type,
20668 change function return type.
20669 (arc_must_save_register): Adapt to handle new
20670 arc_compute_function_type's return type.
20671 (arc_expand_prologue): Likewise.
20672 (arc_expand_epilogue): Likewise.
20673 (arc_return_address_regs): Delete.
20674 (arc_return_address_register): New function.
20675 (arc_epilogue_uses): Use above function.
20676 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20677 (arc_function_type): Change encoding, add naked type.
20678 (ARC_INTERRUPT_P): Change to handle the new encoding.
20679 (ARC_FAST_INTERRUPT_P): Likewise.
20680 (ARC_NORMAL_P): Define.
20681 (ARC_NAKED_P): Likewise.
20682 (arc_compute_function_type): Delete prototype.
20683 * config/arc/arc.md (in_ret_delay_slot): Use
20684 arc_return_address_register function.
20685 (simple_return): Likewise.
20686 (p_return_i): Likewise.
20688 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20690 PR tree-optimization/81428
20691 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20692 can't be built for those types.
20694 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20696 Remove stuff dead since r239246.
20698 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20699 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20700 (avr_inform_devices): Remove dead stuff.
20702 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20704 * config/arm/arm_neon.h: Fix softp typo.
20706 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20708 PR tree-optimization/81365
20709 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20710 aggregate moves onto bb predecessor edges, make sure there are no
20711 loads that could alias the lhs in between the start of bb and the
20714 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20717 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20718 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20719 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20721 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20723 PR tree-optimization/81396
20724 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20725 (init_symbolic_number): Initialize it to 1.
20726 (perform_symbolic_merge): Add n_ops from both operands into the new
20728 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20729 without base_addr as useless if they need more than one operation.
20730 (bswap_replace): Handle !bswap case for NULL base_addr.
20732 2017-07-17 Tom de Vries <tom@codesourcery.com>
20735 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20738 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20740 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20741 conditional builtin define __FIX_LEON3FT_B2BST.
20743 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20745 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20746 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20749 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20751 PR rtl-optimization/81424
20752 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20753 to remove potential trapping from operands if -fnon-call-exceptions.
20755 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20757 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20758 profile_proability for scalling.
20759 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20761 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20763 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20765 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20767 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20768 fixpoint arithmetics.
20770 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20772 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20773 fixpoint arithmetics.
20775 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20777 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20778 fixpoint arithmetics.
20780 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20782 * profile-count.h (profile_probability::from_reg_br_prob_note,
20783 profile_probability::to_reg_br_prob_note): New functions.
20784 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20785 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20786 * predict.c (probability_reliable_p): Update.
20787 (edge_probability_reliable_p): Update.
20788 (br_prob_note_reliable_p): Update.
20789 (invert_br_probabilities): Update.
20790 (add_reg_br_prob_note): New function.
20791 (combine_predictions_for_insn): Update.
20792 * asan.c (asan_clear_shadow): Update.
20793 * cfgbuild.c (compute_outgoing_frequencies): Update.
20794 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20795 (update_br_prob_note): Update.
20796 (rtl_verify_edges): Update.
20797 (purge_dead_edges): Update.
20798 (fixup_reorder_chain): Update.
20799 * emit-rtl.c (try_split): Update.
20800 * ifcvt.c (cond_exec_process_insns): Update.
20801 (cond_exec_process_if_block): Update.
20802 (dead_or_predicable): Update.
20803 * internal-fn.c (expand_addsub_overflow): Update.
20804 (expand_neg_overflow): Update.
20805 (expand_mul_overflow): Update.
20806 * loop-doloop.c (doloop_modify): Update.
20807 * loop-unroll.c (compare_and_jump_seq): Update.
20808 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20809 * predict.h: Update.
20810 * reorg.c (mostly_true_jump): Update.
20812 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20813 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20814 * config/arc/arc.c: (emit_unlikely_jump): Update.
20815 * config/arm/arm.c: (emit_unlikely_jump): Update.
20816 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20817 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20818 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20819 (ix86_print_operand): Update.
20820 (ix86_split_fp_branch): Update.
20821 (predict_jump): Update.
20822 * config/ia64/ia64.c (ia64_print_operand): Update.
20823 * config/mmix/mmix.c (mmix_print_operand): Update.
20824 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20825 (rs6000_expand_split_stack_prologue): Update.
20826 * config/rs6000/rs6000.c: Update.
20827 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20828 (s390_expand_vec_movstr): Update.
20829 (s390_expand_cs_tdsi): Update.
20830 (s390_expand_split_stack_prologue): Update.
20831 * config/sh/sh.c (sh_print_operand): Update.
20832 (expand_cbranchsi4): Update.
20833 (expand_cbranchdi4): Update.
20834 * config/sparc/sparc.c (output_v9branch): Update.
20835 * config/spu/spu.c (get_branch_target): Update.
20836 (ea_load_store_inline): Update.
20837 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20838 * config/tilepro/tilepro.c: Update.
20840 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20842 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20843 (gimplify_save_expr): Likewise.
20845 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20847 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20849 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20851 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20853 * timevar.def (TV_IPA_FNSUMMARY): Define.
20855 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20857 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20858 to back store errata sensitive sequence from being generated.
20859 (sqrtdf2_fix): Likewise.
20861 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20863 * tree-ssa-threadupdate.c (compute_path_counts,
20864 update_joiner_offpath_counts): Use profile_probability.
20866 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20869 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20871 * config/arm/arm-c.c (arm_cpu_builtins): Define
20872 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20874 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20876 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20877 array entries to represent __ieee128 versions of the
20878 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20879 scalar_extract_sig, and scalar_insert_exp built-in functions.
20880 (altivec_resolve_overloaded_builtin): Add special case handling
20881 for the __builtin_scalar_insert_exp function, as represented by
20882 the P9V_BUILTIN_VEC_VSIEDP constant.
20883 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20884 exponent support for __ieee128 argument.
20885 (VSESQP): Add scalar extract signature support for __ieee128
20887 (VSTDCNQP): Add scalar test negative support for __ieee128
20889 (VSIEQP): Add scalar insert exponent support for __int128 argument
20890 with __ieee128 result.
20891 (VSIEQPF): Add scalar insert exponent support for __ieee128
20892 argument with __ieee128 result.
20893 (VSTDCQP): Add scalar test data class support for __ieee128
20895 (VSTDCNQP): Add overload support for scalar test negative with
20896 __ieee128 argument.
20897 (VSTDCQP): Add overload support for scalar test data class
20898 __ieee128 argument.
20899 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20900 UNSPEC_VSX_SXSIGDP.
20901 (UNSPEC_VSX_SIEXPQP): New constant.
20902 (xsxexpqp): New insn for VSX scalar extract exponent quad
20904 (xsxsigqp): New insn for VSX scalar extract significand quad
20906 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20907 precision with floating point argument.
20908 (xststdcqp): New expand for VSX scalar test data class quad
20910 (xststdcnegqp): New expand for VSX scalar test negative quad
20912 (xststdcqp): New insn to match expansions for VSX scalar test data
20913 class quad precision and VSX scalar test negative quad precision.
20914 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20915 special case operand checking to enforce that second operand of
20916 VSX scalar test data class with quad precision argument is a 7-bit
20918 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20919 prototypes and descriptions of __ieee128 versions of
20920 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20921 scalar_test_data_class, and scalar_test_neg built-in functions.
20923 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20925 PR tree-optimization/81162
20926 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20927 replace a negate with an add.
20929 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20931 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20933 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20935 * config/arm/arm-c.c (arm_cpu_builtins): Define
20936 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20938 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20940 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20941 (armv8-r): Set ARM Cortex-R52 as default CPU.
20942 * config/arm/arm-tables.opt: Regenerate.
20943 * config/arm/arm-tune.md: Regenerate.
20944 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20946 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20947 extension for -mcpu=cortex-r52.
20949 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20951 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20952 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20953 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20954 (fp-armv8): Define it as FP_ARMv8 only.
20955 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20956 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20958 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20959 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20960 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20961 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20962 than TARGET_FPU_ARMV8.
20963 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20964 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20965 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20967 * config/arm/neon.md (neon_vrint): Likewise.
20968 (neon_vcvt): Likewise.
20969 (neon_<fmaxmin_op><mode>): Likewise.
20970 (<fmaxmin><mode>3): Likewise.
20971 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20972 * config/arm/predicates.md (arm_cond_move_operator): Check against
20973 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20975 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20977 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20978 to top of function.
20980 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20982 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20983 loop in comment with memset.
20985 2017-07-14 Martin Liska <mliska@suse.cz>
20987 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20988 * dwarf2out.c (is_java): Remove the function.
20989 (output_pubname): Remove usage of the function.
20990 (lower_bound_default): Remove usage of DW_LANG_Java.
20991 (gen_compile_unit_die): Likewise.
20992 * gcc.c: Remove compiler defaults for .java and .zip files.
20993 * gimple-expr.c (remove_suffix): Change as there's no longer
20994 extension than 4-letter one.
20995 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20996 (gimplify_save_expr): Likewise.
20997 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20998 as it's possible even for other languages than Java.
20999 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21000 * lto-opts.c (lto_write_options): Remove reference to Java.
21001 * opts.c (strip_off_ending): Update file extension handling.
21002 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21003 * tree-eh.c (lower_resx): Likewise.
21004 * tree.c (free_lang_data_in_type): Remove dead code.
21005 (find_decls_types_r): Likewise.
21006 (build_common_builtin_nodes): Remove Java from a comment.
21007 (verify_type): Remove dead code.
21008 * varasm.c (assemble_external): Remove Java from a comment.
21010 2017-07-14 Martin Liska <mliska@suse.cz>
21012 * opts.c (finish_options): Add quotes.
21013 (common_handle_option): Likewise.
21015 2017-07-14 Martin Liska <mliska@suse.cz>
21017 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21018 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21019 Remove N_SO_PASCAL.
21020 * dwarf2out.c (lower_bound_default): Do not handle
21022 (gen_compile_unit_die): Likewise.
21023 * gcc.c: Remove default extension binding for GNU Pascal.
21024 * stmt.c: Remove Pascal language from a comment.
21025 * xcoffout.c: Likewise.
21027 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21030 * diagnostic-show-locus.c (fixit_cmp): New function.
21031 (layout::layout): Sort m_fixit_hints.
21032 (column_range::column_range): Assert that the values are valid.
21033 (struct char_span): New struct.
21034 (correction::overwrite): New method.
21035 (struct source_line): New struct.
21036 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21037 calls in terms of classes source_line and char_span, and
21038 correction::overwrite.
21039 (selftest::test_overlapped_fixit_printing_2): New function.
21040 (selftest::diagnostic_show_locus_c_tests): Call it.
21042 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21044 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21045 early if there is no lhs.
21047 2017-07-13 Martin Liska <mliska@suse.cz>
21049 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21050 (gen_reference_type_die): Likewise.
21051 * stor-layout.c: Remove Pascal-related comment.
21053 2017-07-13 Martin Liska <mliska@suse.cz>
21055 * opts.c (finish_options): Add quotes to error messages.
21056 (parse_sanitizer_options): Likewise.
21058 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21060 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21062 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21064 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21066 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21068 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21070 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21072 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21075 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21076 provides the hardware capability bits, define the macro
21077 __BUILTIN_CPU_SUPPORTS__.
21078 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21079 if GLIBC does not provide the hardware capability bits. Add a
21080 gcc_unreachable call if the built-in cpu function is neither
21081 __builtin_cpu_is nor __builtin_cpu_supports.
21082 (rs6000_get_function_versions_dispatcher): Change the warning
21083 that an old GLIBC is used which does not export the capability
21084 bits to be an error.
21085 * doc/extend.texi (target_clones attribute): Document the
21086 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21087 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21088 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21089 the macros defined by GCC if the newer GLIBC is available.
21091 2017-07-12 Jeff Law <law@redhat.com>
21093 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21094 remaining includes slightly.
21095 * config/riscv/riscv-builtins.c: Include profile-count.h.
21097 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21100 * config/avr/avr.c (avr_set_current_function): In diagnostic
21101 messages: Quote keywords and (parts of) identifiers.
21102 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21105 2017-07-12 Carl Love <cel@us.ibm.com>
21107 * config/rs6000/rs6000-c.c: Add support for built-in functions
21108 vector bool char vec_revb (vector bool char);
21109 vector bool short vec_revb (vector short char);
21110 vector bool int vec_revb (vector bool int);
21111 vector bool long long vec_revb (vector bool long long);
21112 * doc/extend.texi: Update the built-in documentation file for the
21113 new built-in functions.
21115 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21117 * config/s390/s390.md: Remove movcc splitter.
21119 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21121 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21122 load/store on condition.
21124 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21127 * config/avr/avr.c (avr_encode_section_info)
21128 [progmem && !TREE_READONLY]: Error if progmem object needs
21131 2017-07-11 Michael Collison <michael.collison@arm.com>
21133 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21136 2017-07-11 Carl Love <cel@us.ibm.com>
21138 * config/rs6000/rs6000-c.c: Add support for builtins
21139 vector unsigned int vec_parity_lsbb (vector signed int);
21140 vector unsigned int vec_parity_lsbb (vector unsigned int);
21141 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21142 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21143 vector unsigned long long vec_parity_lsbb (vector signed long long);
21144 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21145 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21146 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21147 * doc/extend.texi: Update the built-in documentation file for the
21148 new built-in functions.
21150 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21152 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21153 (layout::m_primary_loc): New field.
21154 (layout::layout): Initialize new field. Move location filtering
21155 logic from here to...
21156 (layout::maybe_add_location_range): ...this new method. Add
21157 support for filtering to just the lines already specified by other
21159 (layout::will_show_line_p): New method.
21160 (gcc_rich_location::add_location_if_nearby): New method.
21161 (selftest::test_add_location_if_nearby): New test function.
21162 (selftest::diagnostic_show_locus_c_tests): Call it.
21163 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21166 2017-07-11 Tom de Vries <tom@codesourcery.com>
21168 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21169 (bb_first_real_insn): New function.
21170 (nvptx_single): Add extra initialization of broadcasted condition
21173 2017-07-11 Nathan Sidwell <nathan@acm.org>
21175 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21177 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21179 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21180 Binutils doc as TEXI will mess them up.
21181 * doc/invoke.texi (AVR Options): Same here.
21183 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21185 * config/sparc/sparc.opt (mfix-ut700): New option.
21186 (mfix-gr712rc): Likewise.
21187 (sparc_fix_b2bst): New variable.
21188 * doc/invoke.texi (SPARC options): Document them.
21189 (ARM options): Fix warnings.
21190 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21191 instructions to prevent sequences that can trigger the store-store
21192 errata for certain LEON3FT processors.
21193 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21194 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21195 * config/sparc/sparc.md (fix_b2bst): New attribute.
21196 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21198 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21201 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21203 (*rsqrtsf2_sse): Ditto.
21205 (div<mode>3): Macroize insn from divdf3 and divsf3
21206 using MODEF mode iterator.
21208 2017-07-10 Martin Sebor <msebor@redhat.com>
21210 PR tree-optimization/80397
21211 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21212 instead of testing for equality to INTEGER_TYPE.
21214 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21216 * config.gcc: Remove uclibc from arc target spec.
21218 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21220 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21222 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21225 * lto-wrapper.c (remove_option): New function.
21226 (merge_and_complain): Merge PIC/PIE options more realistically.
21228 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21230 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21234 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21235 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21236 * config.in: Regenerate.
21237 * configure: Regenerate.
21238 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21239 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21240 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21241 (TARGET_GASISR_PROLOGUES): ...target mask.
21242 * common/config/avr/avr-common.c
21243 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21244 Set -mgas-isr-prologues.
21245 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21246 INSERT_PASS_BEFORE for it.
21247 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21248 * config/avr/avr.c (avr_option_override)
21249 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21250 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21251 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21252 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21253 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21255 (make_avr_pass_pre_proep): New function.
21256 (emit_push_sfr) <treg>: Add argument to function and use it
21257 instead of TMP_REG.
21258 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21259 and set machine->gasisr.yes.
21260 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21261 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21262 __gcc_isr.n_pushed to .L__stack_usage.
21263 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21264 (avr_asm_final_postscan_insn): ...this new static function.
21265 * config/avr/avr.h (machine_function)
21266 <is_no_gccisr, use_L__stack_usage>: New fields.
21267 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21268 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21269 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21270 (gasisr, *gasisr): New expander and insn.
21271 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21272 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21273 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21275 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21277 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21280 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21282 Move jump-tables out of .text again.
21285 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21286 (ASM_OUTPUT_ADDR_VEC): New function.
21287 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21288 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21289 INSN_ADDRESSes as asm comment.
21290 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21291 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21292 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21293 * config/avr/avr.md (*tablejump): Adjust comment.
21294 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21295 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21297 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21298 (avr_output_addr_vec): New proto.
21299 (avr_log_t) <insn_addresses>: New field.
21301 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21304 * config/i386/i386.c (ix86_function_arg_advance): Set
21305 outgoing_args_on_stack to true if there are outgoing arguments
21307 (ix86_function_arg): Likewise.
21308 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21309 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21310 * config/i386/i386.h (machine_function): Add
21311 outgoing_args_on_stack.
21313 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21315 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21316 supporting pthreds.
21317 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21319 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21321 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21322 (REAL_H): Remove $(MACHMODE_H).
21323 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21325 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21326 $(MACHMODE_H) and double-int.h.
21327 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21329 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21332 2017-07-07 Andrew Pinski <apinski@cavium.com>
21334 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21335 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21337 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21339 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21340 Add warning if GCC was not configured to link against a GLIBC that
21341 exports the hardware capability bits.
21342 (make_resolver_func): Make resolver function private and not a
21343 COMDAT function. Create the name with clone_function_name instead
21344 of make_unique_name.
21347 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21348 correct operand in doing the split.
21350 2017-07-07 Carl Love <cel@us.ibm.com>
21352 * config/rs6000/rs6000-c: Add support for built-in function
21353 vector unsigned short vec_pack_to_short_fp32 (vector float,
21355 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21356 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21357 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21358 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21359 (convert_4f32_8i16): Add define_expand.
21360 * doc/extend.texi: Update the built-in documentation file for the
21361 new built-in function.
21363 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21365 * config/sparc/m8.md: New file.
21366 * config/sparc/sparc.md: Include m8.md.
21368 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21370 * config/sparc/sparc.opt: New option -mvis4b.
21371 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21372 (sparc_option_override): Handle VIS4B.
21373 (enum sparc_builtins): Define
21374 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21375 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21376 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21377 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21378 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21379 (check_constant_argument): New function.
21380 (sparc_vis_init_builtins): Define builtins
21381 __builtin_vis_dictunpack{8,16,32},
21382 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21383 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21384 __builtin_vis_fpcmpde{8,16,32}shl and
21385 __builtin_vis_fpcmpur{8,16,32}shl.
21386 (sparc_expand_builtin): Check that the constant operands to
21387 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21388 constant and in range.
21389 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21391 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21392 (SPARC_IMM5_P): Likewise.
21393 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21394 (enabled): Handle vis4b.
21395 (UNSPEC_DICTUNPACK): New unspec.
21396 (UNSPEC_FPCMPSHL): Likewise.
21397 (UNSPEC_FPUCMPSHL): Likewise.
21398 (UNSPEC_FPCMPDESHL): Likewise.
21399 (UNSPEC_FPCMPURSHL): Likewise.
21400 (cpu_feature): New CPU feature `vis4b'.
21401 (dictunpack{8,16,32}): New insns.
21402 (FPCSMODE): New mode iterator.
21403 (fpcscond): New code iterator.
21404 (fpcsucond): Likewise.
21405 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21406 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21407 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21408 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21409 * config/sparc/constraints.md: Define constraints `q' for unsigned
21410 2-bit integer constants and `t' for unsigned 5-bit integer
21412 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21414 (imm5_operand_dictunpack16): Likewise.
21415 (imm5_operand_dictunpack32): Likewise.
21416 (imm2_operand): Likewise.
21417 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21418 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21419 ditunpack* and fpcmp*shl builtins.
21421 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21423 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21424 * config.in: Add HAVE_AS_SPARC6 define.
21425 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21427 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21429 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21430 (CPP_CPU_SPEC): Handle m8.
21431 (ASM_CPU_SPEC): Likewise.
21432 * config/sparc/sparc-opts.h (enum processor_type): Add
21434 * config/sparc/sparc.c (m8_costs): New struct.
21435 (sparc_option_override): Handle TARGET_CPU_m8.
21436 (sparc32_initialize_trampoline): Likewise.
21437 (sparc64_initialize_trampoline): Likewise.
21438 (sparc_issue_rate): Likewise.
21439 (sparc_register_move_cost): Likewise.
21440 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21441 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21442 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21443 (CPP_CPU_SPEC): Handle M8.
21444 (ASM_CPU_SPEC): Likewise.
21445 (AS_M8_FLAG): Define.
21446 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21447 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21448 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21450 * configure: Regenerate.
21451 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21454 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21456 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21458 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21459 ("*movdi_insn_sp32"): Do not set v3pipe.
21460 ("*movsi_insn"): Likewise.
21461 ("*movdi_insn_sp64"): Likewise.
21462 ("*movsf_insn"): Likewise.
21463 ("*movdf_insn_sp32"): Likewise.
21464 ("*movdf_insn_sp64"): Likewise.
21465 ("*zero_extendsidi2_insn_sp64"): Likewise.
21466 ("*sign_extendsidi2_insn"): Likewise.
21467 ("*mov<VM32:mode>_insn"): Likewise.
21468 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21469 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21470 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21471 ("<vlop:code><VL:mode>3"): Likewise.
21472 ("*not_<vlop:code><VL:mode>3"): Likewise.
21473 ("*nand<VL:mode>_vis"): Likewise.
21474 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21475 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21476 ("one_cmpl<VL:mode>2"): Likewise.
21477 ("faligndata<VM64:mode>_vis"): Likewise.
21478 ("alignaddrsi_vis"): Likewise.
21479 ("alignaddrdi_vis"): Likweise.
21480 ("alignaddrlsi_vis"): Likewise.
21481 ("alignaddrldi_vis"): Likewise.
21482 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21483 ("bmaskdi_vis"): Likewise.
21484 ("bmasksi_vis"): Likewise.
21485 ("bshuffle<VM64:mode>_vis"): Likewise.
21486 ("cmask8<P:mode>_vis"): Likewise.
21487 ("cmask16<P:mode>_vis"): Likewise.
21488 ("cmask32<P:mode>_vis"): Likewise.
21489 ("pdistn<P:mode>_vis"): Likewise.
21490 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21492 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21494 * config/sparc/sparc.md ("subtype"): New insn attribute.
21495 ("*wrgsr_sp64"): Set insn subtype.
21496 ("*rdgsr_sp64"): Likewise.
21497 ("alignaddrsi_vis"): Likewise.
21498 ("alignaddrdi_vis"): Likewise.
21499 ("alignaddrlsi_vis"): Likewise.
21500 ("alignaddrldi_vis"): Likewise.
21501 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21502 ("fexpand_vis"): Likewise.
21503 ("fpmerge_vis"): Likewise.
21504 ("faligndata<VM64:mode>_vis"): Likewise.
21505 ("bshuffle<VM64:mode>_vis"): Likewise.
21506 ("cmask8<P:mode>_vis"): Likewise.
21507 ("cmask16<P:mode>_vis"): Likewise.
21508 ("cmask32<P:mode>_vis"): Likewise.
21509 ("fchksm16_vis"): Likewise.
21510 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21511 ("fmean16_vis"): Likewise.
21512 ("fp<plusminus_insn>64_vis"): Likewise.
21513 ("<plusminus_insn>v8qi3"): Likewise.
21514 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21515 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21516 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21517 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21518 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21519 ("*movqi_insn"): Likewise.
21520 ("*movhi_insn"): Likewise.
21521 ("*movsi_insn"): Likewise.
21522 ("movsi_pic_gotdata_op"): Likewise.
21523 ("*movdi_insn_sp32"): Likewise.
21524 ("*movdi_insn_sp64"): Likewise.
21525 ("movdi_pic_gotdata_op"): Likewise.
21526 ("*movsf_insn"): Likewise.
21527 ("*movdf_insn_sp32"): Likewise.
21528 ("*movdf_insn_sp64"): Likewise.
21529 ("*zero_extendhisi2_insn"): Likewise.
21530 ("*zero_extendqihi2_insn"): Likewise.
21531 ("*zero_extendqisi2_insn"): Likewise.
21532 ("*zero_extendqidi2_insn"): Likewise.
21533 ("*zero_extendhidi2_insn"): Likewise.
21534 ("*zero_extendsidi2_insn_sp64"): Likewise.
21535 ("ldfsr"): Likewise.
21536 ("prefetch_64"): Likewise.
21537 ("prefetch_32"): Likewise.
21538 ("tie_ld32"): Likewise.
21539 ("tie_ld64"): Likewise.
21540 ("*tldo_ldub_sp32"): Likewise.
21541 ("*tldo_ldub1_sp32"): Likewise.
21542 ("*tldo_ldub2_sp32"): Likewise.
21543 ("*tldo_ldub_sp64"): Likewise.
21544 ("*tldo_ldub1_sp64"): Likewise.
21545 ("*tldo_ldub2_sp64"): Likewise.
21546 ("*tldo_ldub3_sp64"): Likewise.
21547 ("*tldo_lduh_sp32"): Likewise.
21548 ("*tldo_lduh1_sp32"): Likewise.
21549 ("*tldo_lduh_sp64"): Likewise.
21550 ("*tldo_lduh1_sp64"): Likewise.
21551 ("*tldo_lduh2_sp64"): Likewise.
21552 ("*tldo_lduw_sp32"): Likewise.
21553 ("*tldo_lduw_sp64"): Likewise.
21554 ("*tldo_lduw1_sp64"): Likewise.
21555 ("*tldo_ldx_sp64"): Likewise.
21556 ("*mov<VM32:mode>_insn"): Likewise.
21557 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21558 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21560 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21562 * config/sparc/sparc.md ("type"): New insn type viscmp.
21563 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21565 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21566 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21567 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21568 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21570 ("n7_vis_logical_11cycle"): Likewise.
21571 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21572 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21573 * config/sparc/niagara.md ("niag_vis"): Likewise.
21574 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21575 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21577 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21579 * config/sparc/sparc.md: New instruction type `bmask'.
21580 (bmaskdi_vis): Use the `bmask' type.
21581 (bmasksi_vis): Likewise.
21582 * config/sparc/ultra3.md (us3_array): Likewise.
21583 * config/sparc/niagara7.md (n7_array): Likewise.
21584 * config/sparc/niagara4.md (n4_array): Likewise.
21585 * config/sparc/niagara2.md (niag2_vis): Likewise.
21586 (niag3_vis): Likewise.
21587 * config/sparc/niagara.md (niag_vis): Likewise.
21589 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21591 * ipa-comdats.c: Remove optimize check from gate.
21592 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21593 for functions not optimized.
21594 (ipa_fn_summary_read): Skip optimize check.
21595 (ipa_fn_summary_write): Likewise.
21596 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21598 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21600 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21601 (check_callers): Give up on caller which is not optimized.
21602 (inline_small_functions): Likewise.
21603 (ipa_inline): Do not give up when not optimizing.
21604 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21605 away unoptimizes cdtors.
21606 (whole_program_function_and_variable_visibility): Do
21607 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21608 * ipa.c (process_references): Do not check optimize.
21609 (symbol_table::remove_unreachable_nodes): Update optimize check.
21610 (set_writeonly_bit): Update optimize check.
21611 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21612 (pass_ipa_single_use::gate): Remove.
21614 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21616 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21617 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21618 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21619 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21620 permute_load, permute_store, adjust_extract, adjust_splat,
21621 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21622 replace_swap_with_copy, dump_swap_insn_table,
21623 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21624 recombine_lvx_pattern, recombine_stvx_pattern,
21625 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21626 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21627 to file rs6000-p8swap.c.
21628 * config/rs6000/rs6000-p8swap.c: New file.
21629 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21630 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21631 and rs6000*-*-* targets.
21633 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21635 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21637 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21639 * lto-wrapper.c (merge_and_complain): Do not merge
21640 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21641 fsigned_zeros, ftrapping_math, fwrapv.
21642 (append_compiler_options): Do not track these options.
21643 (append_linker_options): Likewie
21645 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21647 * cgraphunit.c (cgraph_node::finalize_function): When
21648 !flag_toplevel_reorde set no_reorder flag.
21649 (varpool_node::finalize_decl): Likewise.
21650 (symbol_table::compile): Drop no toplevel reorder path.
21652 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21654 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21655 edges; zero probability is not better than uninitialized.
21657 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21659 * asan.h (asan_sanitize_allocas_p): Declare.
21660 * asan.c (asan_sanitize_allocas_p): New function.
21661 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21662 (handle_builtin_alloca): Likewise.
21663 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21664 if !asan_sanitize_allocas_p.
21665 * params.def (asan-instrument-allocas): Add new option.
21666 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21667 * opts.c (common_handle_option): Disable allocas sanitization for
21670 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21672 * asan.c: Include gimple-fold.h.
21673 (get_last_alloca_addr): New function.
21674 (handle_builtin_stackrestore): Likewise.
21675 (handle_builtin_alloca): Likewise.
21676 (asan_emit_allocas_unpoison): Likewise.
21677 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21678 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21679 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21680 (instrument_builtin_call): Pass gimple iterator to
21681 get_mem_refs_of_builtin_call.
21682 (last_alloca_addr): New global.
21683 * asan.h (asan_emit_allocas_unpoison): Declare.
21684 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21685 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21686 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21687 if function calls alloca.
21688 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21689 * gimple-fold.h (replace_call_with_value): Declare.
21690 * internal-fn.c: Include asan.h.
21691 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21692 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21694 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21696 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21697 (C_SELFTEST_FLAGS): New.
21698 (CPP_SELFTEST_FLAGS): New.
21699 (SELFTEST_DEPS): New, from deps of s-selftest.
21700 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21701 (CPP_SELFTEST_DEPS): New.
21702 (selftest): Add dependency on s-selftest-c++.
21703 (s-selftest): Rename to...
21704 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21705 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21706 than SELFTEST_FLAGS.
21707 (selftest-gdb): Rename to...
21708 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21710 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21711 (selftest-valgrind): Rename to...
21712 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21714 (selftest-valgrind): Reintroduce as an alias for
21715 selftest-c-valgrind.
21716 (s-selftest-c++): New.
21717 (selftest-c++-gdb): New.
21718 (selftest-c++-valgrind): New.
21720 2017-07-06 Olivier Hainque <hainque@adacore.com>
21722 * gcc.c (process_command): When deciding if undefined variables
21723 should be ignored when processing specs, accept "gcc -v" as well.
21725 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21727 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21728 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21730 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21732 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21733 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21734 * config/arm/arm-tables.opt: Regenerate.
21735 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21737 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21739 2017-07-06 Carl Love <cel@us.ibm.com>
21741 * ChangeLog: Clean up from mid air collision
21743 2017-07-06 Carl Love <cel@us.ibm.com>
21745 * config/rs6000/rs6000-c.c: Add support for built-in functions
21746 vector signed int vec_subc (vector signed int, vector signed int);
21747 vector signed __int128 vec_subc (vector signed __int128,
21748 vector signed __int128);
21749 vector unsigned __int128 vec_subc (vector unsigned __int128,
21750 vector unsigned __int128);
21751 vector signed int vec_sube (vector signed int, vector signed int,
21752 vector signed int);
21753 vector unsigned int vec_sube (vector unsigned int,
21754 vector unsigned int,
21755 vector unsigned int);
21756 vector signed __int128 vec_sube (vector signed __int128,
21757 vector signed __int128,
21758 vector signed__int128);
21759 vector unsigned __int128 vec_sube (vector unsigned __int128,
21760 vector unsigned __int128,
21761 vector unsigned __int128);
21762 vector signed int vec_subec (vector signed int, vector signed int,
21763 vector signed int);
21764 vector unsigned int vec_subec (vector unsigned int,
21765 vector unsigned int,
21766 vector unsigned int);
21767 vector signed __int128 vec_subec (vector signed __int128,
21768 vector signed __int128,
21769 vector signed__int128);
21770 vector unsigned __int128 vec_subec (vector unsigned __int128,
21771 vector unsigned __int128,
21772 vector unsigned __int128);
21773 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21774 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21775 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21776 BU_ALTIVEC_OVERLOAD_X definitions.
21777 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21778 * doc/extend.texi: Update the built-in documentation file for the new
21779 built-in functions.
21781 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21784 * diagnostic-show-locus.c (layout::layout): Use start and finish
21785 spelling location for the start and finish of each range.
21786 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21787 Add unused aspect param.
21788 * input.c (expand_location_1): Add "aspect" param, and use it
21789 to access the correct part of the location.
21790 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21792 (expand_location_to_spelling_point): Likewise.
21793 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21794 param, and pass it to expand_location_1.
21796 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21798 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21799 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21800 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21801 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21802 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21803 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21804 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21805 _mm_maskz_getmant_ss): New intrinsics.
21806 (__builtin_ia32_getexpss128_mask): Changed to ...
21807 __builtin_ia32_getexpss128_round ... this.
21808 (__builtin_ia32_getexpsd128_mask): Changed to ...
21809 __builtin_ia32_getexpsd128_round ... this.
21810 * config/i386/i386-builtin-types.def
21811 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21812 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21813 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21814 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21815 __builtin_ia32_getmantss_mask_round): New builtins.
21816 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21817 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21818 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21819 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21820 * config/i386/sse.md
21821 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21822 avx512f_sgetexp<mode><mask_scalar_name>
21823 <round_saeonly_scalar_name> ... this.
21824 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21825 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21826 vgetexp<ssescalarmodesuffix>
21827 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21828 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21829 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21830 avx512f_vgetmant<mode><mask_scalar_name>
21831 <round_saeonly_scalar_name> ... this.
21832 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21833 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21834 vgetmant<ssescalarmodesuffix>
21835 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21836 %0<mask_scalar_operand4>, %1, %2
21837 <round_saeonly_scalar_mask_op4>, %3} ... this.
21838 * config/i386/subst.md (mask_scalar_operand4,
21839 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21840 round_saeonly_scalar_nimm_predicate): New subst attributes.
21842 2017-07-06 Julia Koval <julia.koval@intel.com>
21844 * config/i386/i386.c (ix86_erase_embedded_rounding):
21845 Remove code for old rounding pattern.
21847 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21849 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21851 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21853 * doc/sourcebuild.texi (Test Directives, Variants of
21854 dg-require-support): Add documentation for dg-require-stack-check.
21856 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21858 * config/i386/subst.md (mask_scalar, round_scalar,
21859 round_saeonly_scalar): New meta-templates.
21860 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21861 round_scalar_mask_operand3, round_scalar_mask_op3,
21862 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21863 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21864 round_saeonly_scalar_constraint,
21865 round_saeonly_scalar_prefix): New subst attribute.
21866 * config/i386/sse.md
21867 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21868 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21869 <round_scalar_name> ... this.
21870 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21871 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21872 <round_scalar_name> ... this.
21873 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21874 <sse>_vm<code><mode>3<mask_scalar_name>
21875 <round_saeonly_scalar_name> ... this.
21876 (v<plusminus_mnemonic><ssescalarmodesuffix>
21877 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21878 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21879 v<plusminus_mnemonic><ssescalarmodesuffix>
21880 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21881 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21882 (v<multdiv_mnemonic><ssescalarmodesuffix>
21883 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21884 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21885 v<multdiv_mnemonic><ssescalarmodesuffix>
21886 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21887 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21888 (v<maxmin_float><ssescalarmodesuffix>
21889 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21890 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21891 v<maxmin_float><ssescalarmodesuffix>
21892 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21893 %0<mask_scalar_operand3>, %1, %<iptr>2
21894 <round_saeonly_scalar_mask_op3>} ... this.
21896 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21898 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21899 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21901 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21902 Alan Hayward <alan.hayward@arm.com>
21903 David Sherwood <david.sherwood@arm.com>
21905 * combine.c (simplify_if_then_else): Remove "enum" before
21907 * compare-elim.c (can_eliminate_compare): Likewise.
21908 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21910 (aarch64_lookup_simd_builtin_type): Likewise.
21911 (aarch64_simd_builtin_type): Likewise.
21912 (aarch64_init_simd_builtin_types): Likewise.
21913 (aarch64_simd_expand_args): Likewise.
21914 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21916 (aarch64_reverse_mask): Likewise.
21917 (aarch64_simd_emit_reg_reg_move): Likewise.
21918 (aarch64_gen_adjusted_ldpstp): Likewise.
21919 (aarch64_ccmp_mode_to_code): Likewise.
21920 (aarch64_operands_ok_for_ldpstp): Likewise.
21921 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21922 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21924 (aarch64_min_divisions_for_recip_mul): Likewise.
21925 (aarch64_reassociation_width): Likewise.
21926 (aarch64_get_condition_code_1): Likewise.
21927 (aarch64_simd_emit_reg_reg_move): Likewise.
21928 (aarch64_simd_attr_length_rglist): Likewise.
21929 (aarch64_reverse_mask): Likewise.
21930 (aarch64_operands_ok_for_ldpstp): Likewise.
21931 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21932 (aarch64_gen_adjusted_ldpstp): Likewise.
21933 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21935 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21936 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21937 (arm_lookup_simd_builtin_type): Likewise.
21938 (arm_simd_builtin_type): Likewise.
21939 (arm_init_simd_builtin_types): Likewise.
21940 (arm_expand_builtin_args): Likewise.
21941 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21942 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21943 (ft32_setup_incoming_varargs): Likewise.
21944 (ft32_function_arg): Likewise.
21945 (ft32_function_arg_advance): Likewise.
21946 (ft32_pass_by_reference): Likewise.
21947 (ft32_arg_partial_bytes): Likewise.
21948 (ft32_valid_pointer_mode): Likewise.
21949 (ft32_addr_space_pointer_mode): Likewise.
21950 (ft32_addr_space_legitimate_address_p): Likewise.
21951 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21953 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21954 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21955 (iamcu_alignment): Likewise.
21956 (canonicalize_vector_int_perm): Likewise.
21957 (ix86_noce_conversion_profitable_p): Likewise.
21958 (ix86_mpx_bound_mode): Likewise.
21959 (ix86_operands_ok_for_move_multiple): Likewise.
21960 * config/microblaze/microblaze-protos.h
21961 (microblaze_expand_conditional_branch_reg): Likewise.
21962 * config/microblaze/microblaze.c
21963 (microblaze_expand_conditional_branch_reg): Likewise.
21964 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21966 (rs6000_reassociation_width): Likewise.
21967 (rs6000_invalid_binary_op): Likewise.
21968 (fusion_p9_p): Likewise.
21969 (emit_fusion_p9_load): Likewise.
21970 (emit_fusion_p9_store): Likewise.
21971 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21973 (riscv_hard_regno_mode_ok_p): Likewise.
21974 (riscv_address_insns): Likewise.
21975 (riscv_split_symbol): Likewise.
21976 (riscv_legitimize_move): Likewise.
21977 (riscv_function_value): Likewise.
21978 (riscv_hard_regno_nregs): Likewise.
21979 (riscv_expand_builtin): Likewise.
21980 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21981 (riscv_build_integer): Likewise.
21982 (riscv_split_integer): Likewise.
21983 (riscv_legitimate_constant_p): Likewise.
21984 (riscv_cannot_force_const_mem): Likewise.
21985 (riscv_regno_mode_ok_for_base_p): Likewise.
21986 (riscv_valid_base_register_p): Likewise.
21987 (riscv_valid_offset_p): Likewise.
21988 (riscv_valid_lo_sum_p): Likewise.
21989 (riscv_classify_address): Likewise.
21990 (riscv_legitimate_address_p): Likewise.
21991 (riscv_address_insns): Likewise.
21992 (riscv_load_store_insns): Likewise.
21993 (riscv_force_binary): Likewise.
21994 (riscv_split_symbol): Likewise.
21995 (riscv_force_address): Likewise.
21996 (riscv_legitimize_address): Likewise.
21997 (riscv_move_integer): Likewise.
21998 (riscv_legitimize_const_move): Likewise.
21999 (riscv_legitimize_move): Likewise.
22000 (riscv_address_cost): Likewise.
22001 (riscv_subword): Likewise.
22002 (riscv_output_move): Likewise.
22003 (riscv_canonicalize_int_order_test): Likewise.
22004 (riscv_emit_int_order_test): Likewise.
22005 (riscv_function_arg_boundary): Likewise.
22006 (riscv_pass_mode_in_fpr_p): Likewise.
22007 (riscv_pass_fpr_single): Likewise.
22008 (riscv_pass_fpr_pair): Likewise.
22009 (riscv_get_arg_info): Likewise.
22010 (riscv_function_arg): Likewise.
22011 (riscv_function_arg_advance): Likewise.
22012 (riscv_arg_partial_bytes): Likewise.
22013 (riscv_function_value): Likewise.
22014 (riscv_pass_by_reference): Likewise.
22015 (riscv_setup_incoming_varargs): Likewise.
22016 (riscv_print_operand): Likewise.
22017 (riscv_elf_select_rtx_section): Likewise.
22018 (riscv_save_restore_reg): Likewise.
22019 (riscv_for_each_saved_reg): Likewise.
22020 (riscv_register_move_cost): Likewise.
22021 (riscv_hard_regno_mode_ok_p): Likewise.
22022 (riscv_hard_regno_nregs): Likewise.
22023 (riscv_class_max_nregs): Likewise.
22024 (riscv_memory_move_cost): Likewise.
22025 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22026 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22027 (rl78_addr_space_address_mode): Likewise.
22028 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22030 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22031 (rs6000_reassociation_width): Likewise.
22032 (rs6000_invalid_binary_op): Likewise.
22033 (fusion_p9_p): Likewise.
22034 (emit_fusion_p9_load): Likewise.
22035 (emit_fusion_p9_store): Likewise.
22036 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22037 (ok_for_simple_move_operands): Likewise.
22038 (ok_for_simple_move_strict_operands): Likewise.
22039 (ok_for_simple_arith_logic_operands): Likewise.
22040 (visium_legitimize_reload_address): Likewise.
22041 (visium_select_cc_mode): Likewise.
22042 (output_cbranch): Likewise.
22043 (visium_split_double_move): Likewise.
22044 (visium_expand_copysign): Likewise.
22045 (visium_expand_int_cstore): Likewise.
22046 (visium_expand_fp_cstore): Likewise.
22047 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22048 (visium_function_arg): Likewise.
22049 (visium_function_arg_advance): Likewise.
22050 (visium_libcall_value): Likewise.
22051 (visium_setup_incoming_varargs): Likewise.
22052 (visium_legitimate_constant_p): Likewise.
22053 (visium_legitimate_address_p): Likewise.
22054 (visium_legitimize_address): Likewise.
22055 (visium_secondary_reload): Likewise.
22056 (visium_register_move_cost): Likewise.
22057 (visium_memory_move_cost): Likewise.
22058 (prepare_move_operands): Likewise.
22059 (ok_for_simple_move_operands): Likewise.
22060 (ok_for_simple_move_strict_operands): Likewise.
22061 (ok_for_simple_arith_logic_operands): Likewise.
22062 (visium_function_value_1): Likewise.
22063 (rtx_ok_for_offset_p): Likewise.
22064 (visium_legitimize_reload_address): Likewise.
22065 (visium_split_double_move): Likewise.
22066 (visium_expand_copysign): Likewise.
22067 (visium_expand_int_cstore): Likewise.
22068 (visium_expand_fp_cstore): Likewise.
22069 (visium_split_cstore): Likewise.
22070 (visium_select_cc_mode): Likewise.
22071 (visium_split_cbranch): Likewise.
22072 (output_cbranch): Likewise.
22073 (visium_print_operand_address): Likewise.
22074 * expmed.c (flip_storage_order): Likewise.
22075 * expmed.h (emit_cstore): Likewise.
22076 (flip_storage_order): Likewise.
22077 * genrecog.c (validate_pattern): Likewise.
22078 * hsa-gen.c (gen_hsa_addr): Likewise.
22079 * internal-fn.c (expand_arith_overflow): Likewise.
22080 * ira-color.c (allocno_copy_cost_saving): Likewise.
22081 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22082 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22083 (process_invariant_for_inheritance): Likewise.
22084 * lra-eliminations.c (move_plus_up): Likewise.
22085 * omp-low.c (lower_oacc_reductions): Likewise.
22086 * simplify-rtx.c (simplify_subreg): Likewise.
22087 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22088 (TARGET_CHKP_BOUND_MODE): Likewise..
22089 * targhooks.c (default_chkp_bound_mode): Likewise.
22090 (default_setup_incoming_vararg_bounds): Likewise.
22091 * targhooks.h (default_chkp_bound_mode): Likewise.
22092 (default_setup_incoming_vararg_bounds): Likewise.
22093 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22094 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22095 (have_whole_vector_shift): Likewise.
22096 * tree-vect-stmts.c (vectorizable_load): Likewise.
22097 * doc/tm.texi: Regenerate.
22099 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22101 Graceful degrade if Binutils PR21472 is not available.
22104 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22105 .rodata in flash test fails.
22106 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22107 * confgure: Regenerate.
22108 * config.in: Regenerate.
22109 * config/avr/avr.c (avr_asm_named_section)
22110 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22111 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22112 (avr_asm_init_sections): Same.
22114 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22116 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22117 (fma<VH:mode>4_intrinsic): Likewise.
22118 (*fmsub<VCVTF:mode>4): Likewise.
22119 (*fmsub<VH:mode>4_intrinsic): Likewise.
22121 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22124 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22125 Don't depend on "optimize > 0".
22126 (out_movhi_r_mr, out_movqi_mr_r): Same.
22127 (out_movhi_mr_r, out_movqi_r_mr): Same.
22128 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22129 io_address_operand on "optimize > 0".
22131 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22133 * tree-loop-distribution.c: Add general explanantion on the pass.
22134 (generate_loops_for_partition): Mark distributed loop.
22135 (pg_add_dependence_edges): New parameter. Handle alias data
22136 dependence specially and record it in the parameter if asked.
22137 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22138 (init_partition_graph_vertices, add_partition_graph_edge): New.
22139 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22140 (free_partition_graph_vdata, build_partition_graph): New.
22141 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22142 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22143 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22144 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22145 (version_for_distribution_p, finalize_partitions): New.
22146 (distribute_loop): Handle alias data dependence specially. Factor
22147 out loop fusion code as functions and call these functions.
22149 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22151 * tree-loop-distribution.c (classify_partition): New parameter and
22152 better handle reduction statement.
22153 (rdg_build_partitions): Revise comment.
22154 (distribute_loop): Compute statements in all partitions and pass it
22155 to classify_partition.
22157 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22159 * tree-loop-distribution.c (enum partition_type): New.
22160 (struct partition): New field type.
22161 (partition_merge_into): Add parameter. Update partition type.
22162 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22163 (build_rdg_partition_for_vertex): Compute partition type.
22164 (rdg_build_partitions): Dump partition type.
22165 (distribute_loop): Update calls to partition_merge_into.
22167 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22169 * tree-loop-distribution.c (struct ddr_hasher): New.
22170 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22172 (classify_partition): Call get_data_dependence.
22173 (pg_add_dependence_edges): Ditto.
22174 (distribute_loop): Release data dependence hash table.
22176 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22178 * tree-loop-distribution.c (ref_base_address): Delete.
22179 (similar_memory_accesses): Rename ...
22180 (share_memory_accesses): ... to this. Check if partitions access
22181 the same memory reference.
22182 (distribute_loop): Call share_memory_accesses.
22184 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22186 * tree-loop-distribution.c (struct partition): New field recording
22187 its data reference.
22188 (partition_alloc, partition_free): Init and release data refs.
22189 (partition_merge_into): Merge data refs.
22190 (build_rdg_partition_for_vertex): Collect data refs for partition.
22191 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22193 (distribute_loop): Remve data refs from vertice data of partition
22196 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22198 * tree-loop-distribution.c (params.h): Include header file.
22199 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22200 (datarefs_vec): New global var.
22201 (create_rdg_vertices): Use datarefs_vec directly.
22202 (free_rdg): Don't free data references.
22203 (build_rdg): Update use. Don't free data references.
22204 (distribute_loop): Compute global variable for data references.
22205 Bail out if there are too many data references.
22207 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22209 * tree-loop-distribution.c (loop_nest): New global var.
22210 (build_rdg): Use loop directly, rather than loop nest.
22211 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22213 (distribute_loop): Compute global variable loop nest. Update use.
22215 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22217 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22218 (partition_merge_into): New parameter. Dump reason for fusion.
22219 (distribute_loop): Update use of partition_merge_into.
22221 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22223 * tree-loop-distribution.c (bb_top_order_index): New.
22224 (bb_top_order_index_size, bb_top_order_cmp): New.
22225 (stmts_from_loop): Use topological order.
22226 (pass_loop_distribution::execute): Compute and release topological
22227 order for basic blocks.
22229 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22231 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22234 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22236 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22237 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22238 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22239 * internal-fn.def (LOOP_DIST_ALIAS): New.
22240 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22241 (fold_loop_internal_call): ... this.
22242 (vect_loop_dist_alias_call): New function.
22243 (set_uid_loop_bbs): Call fold_loop_internal_call.
22244 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22247 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22250 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22251 Require dead FLAGS_REG at the beginning of a peephole.
22253 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22256 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22257 arguments in the call to __builtin_ia32_sbb_u32.
22258 (_subborrow_u64): Swap _X and _Y arguments in the call to
22259 __builtin_ia32_sbb_u64.
22261 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22264 * tree-vrp.c (compare_assert_loc): Turn into a function template
22265 with stable template parameter. Only test if a->e is NULL,
22266 !a->e == !b->e has been verified already. Use e == NULL or
22267 e != NULL instead of e or ! e tests. If stable is true, don't use
22268 iterative_hash_expr, on the other side allow a or b or both NULL
22269 and sort the NULLs last.
22270 (process_assert_insertions): Sort using compare_assert_loc<false>
22271 instead of compare_assert_loc, later sort using
22272 compare_assert_loc<true> before calling process_assert_insertions_for
22273 in a loop. Use break instead of continue once seen NULL pointer.
22275 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22277 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22278 Cortex-R7 and Cortex-R8 processors.
22280 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22282 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22283 uninitialized while src is not.
22285 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22287 * common/config/arm/arm-common.c: Adjust include path for
22289 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22290 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22291 (arm-cpu-data.h): Likewise.
22292 (arm-cpu-cdata.h): Likewise.
22293 * config/arm/arm-cpu.h: Delete.
22294 * config/arm/arm-cpu-cdata.h: Delete.
22295 * config/arm/arm-cpu-data.h: Delete.
22297 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22299 * config/arm/arm-cpus.in (cortex-a55): New.
22300 (cortex-a75): Likewise.
22301 (cortex-a75.cortex-a55): Likewise.
22302 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22304 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22305 * config/arm/arm-cpu-cdata.h: Regenerate.
22306 * config/arm/arm-cpu-data.h: Regenerate.
22307 * config/arm/arm-cpu.h: Regenerate.
22308 * config/arm/arm-tables.opt: Regenerate.
22309 * config/arm/arm-tune.md: Regenerate.
22311 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22313 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22315 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22317 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22320 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22322 PR tree-optimization/81292
22323 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22324 full_string_p, also call adjust_related_strinfos if the adjustment
22325 is simple, otherwise invalidate related strinfos.
22327 2017-07-04 Martin Liska <mliska@suse.cz>
22330 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22331 newly created variable as DECL_IGNORED_P.
22333 2017-07-04 Martin Liska <mliska@suse.cz>
22336 * ipa-inline.c (inline_small_functions):
22337 Use xstrdup_for_dump.
22339 2017-07-04 Tom de Vries <tom@codesourcery.com>
22341 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22343 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22346 * config/darwin.c (darwin_function_switched_text_sections):
22349 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22351 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22353 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22355 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22357 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22359 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22360 min_profitable_iters, and th as inclusive lower bounds.
22361 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22362 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22363 for min_profitable_iters and min_profitable_estimate.
22364 (vect_transform_loop): Treat th as an inclusive lower bound.
22365 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22367 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22370 * config/darwin.c (darwin_function_switched_text_sections):
22371 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22372 in two pieces, and suppress the use of buf.
22374 2017-07-03 Nathan Sidwell <nathan@acm.org>
22376 * hash-table.h (hash_table_mod1): Fix indentation.
22378 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22380 PR middle-end/81290
22381 * predict.c (force_edge_cold): Be more careful about propagation
22383 * profile-count.h (profile_probability::guessed,
22384 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22386 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22388 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22390 * doc/invoke.texi (rcpc architecture extension): Document it.
22392 2017-07-03 Richard Biener <rguenther@suse.de>
22394 PR tree-optimization/60510
22395 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22396 the scalar reduction PHI and use it.
22397 (vectorizable_reduction): Properly guard the single_defuse_cycle
22398 path for non-SLP reduction chains where we cannot use it.
22399 Rework reduc_def/index and vector type deduction. Rework
22400 vector operand gathering during reduction op code-gen.
22401 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22402 chains dissolve the chain and leave it to non-SLP reduction
22405 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22407 * tree-data-ref.h (dr_alignment): Declare.
22408 * tree-data-ref.c (dr_alignment): New function.
22409 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22410 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22412 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22414 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22416 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22417 and base_misalignment fields.
22418 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22419 * tree-data-ref.c: Include builtins.h.
22420 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22421 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22422 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22423 * tree-vect-data-refs.c: Include tree-cfg.h.
22424 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22425 fields instead of calculating an alignment here.
22426 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22427 innermost_loop_behavior fields.
22429 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22431 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22433 (DR_STEP_ALIGNMENT): New macro.
22434 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22435 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22436 (create_data_ref): Print it.
22437 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22438 to tell whether the step preserves vector (mis)alignment.
22439 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22440 Move the check for an integer step and generalise to all INTEGER_CST.
22441 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22442 Print the outer step alignment.
22444 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22446 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22447 with offset_alignment.
22448 (DR_ALIGNED_TO): Delete.
22449 (DR_OFFSET_ALIGNMENT): New macro.
22450 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22451 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22452 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22453 (create_data_ref): Likewise.
22454 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22455 (vect_analyze_data_refs): Likewise.
22456 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22457 creating dummy innermost behavior.
22459 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22461 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22462 with a "innermost_loop_behavior *" and refeence tree.
22463 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22464 (create_data_ref): Update call accordingly.
22465 * tree-predcom.c (find_looparound_phi): Likewise.
22467 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22469 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22470 fields with dr_wrt_vec_loop.
22471 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22472 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22473 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22474 (vect_dr_behavior): New function.
22475 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22476 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22477 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22478 track whether the step preserves the misalignment.
22479 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22480 Use vect_dr_behavior.
22481 (vect_setup_realignment): Update call accordingly.
22482 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22483 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22484 call to vect_create_addr_base_for_vector_ref.
22485 (vect_create_cond_for_align_checks): Likewise.
22486 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22487 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22488 (vect_recog_mask_conversion_pattern): Likewise.
22489 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22490 (new_stmt_vec_info): Remove redundant zeroing.
22492 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22494 * common/config/arm/arm-common.c (arm_be8_option): New function.
22495 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22496 (ISA_ARMv6): Add isa_bit_be8.
22497 * config/arm/arm.h (arm_be8_option): Add prototype.
22498 (BE8_SPEC_FUNCTION): New define.
22499 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22500 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22501 (mlittle-endian): Similarly.
22502 (mbe8, mbe32): New options.
22503 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22504 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22506 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22508 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22510 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22512 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22513 (cleanup_tree_cfg_bb): Use it.
22514 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22516 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22518 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22521 * loop-doloop.c (add_test): Update profile.
22523 2017-07-03 Martin Liska <mliska@suse.cz>
22526 * sanopt.c (rewrite_usage_of_param): New function.
22527 (sanitize_rewrite_addressable_params): Likewise.
22528 (pass_sanopt::execute): Call rewrite_usage_of_param.
22530 2017-07-03 Richard Biener <rguenther@suse.de>
22532 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22533 back to using VIEW_CONVERT_EXPR.
22535 2017-07-03 Martin Liska <mliska@suse.cz>
22538 * doc/extend.texi: Document when a resolver function is
22539 generated for target_clones.
22541 2017-07-03 Martin Liska <mliska@suse.cz>
22543 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22544 and shadow memory of auto variables which are subject of
22545 use-after-scope sanitization.
22546 (asan_expand_mark_ifn): Add do set only when is_poison.
22548 2016-07-03 Richard Biener <rguenther@suse.de>
22550 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22552 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22553 (vectorizable_reduction): Perform reduction PHI creation when
22554 visiting a reduction PHI and adjust and simplify code generation
22555 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22556 (vect_transform_loop): Visit reduction PHIs.
22557 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22558 defs into the SLP tree.
22559 (vect_build_slp_tree): Reduction defs terminate the recursion.
22560 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22562 (vect_get_vec_defs_for_stmt_copy): Export.
22563 (vect_get_vec_defs): Likewise.
22564 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22566 (vect_get_vec_defs_for_stmt_copy): Declare.
22567 (vect_get_vec_defs): Likewise.
22569 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22571 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22572 parameter with a "loop" parameter and use it instead of the
22573 loop containing DR_STMT. Don't check simple_iv when doing
22574 BB analysis. Describe the two analysis modes in the comment.
22576 2017-07-03 Tom de Vries <tom@codesourcery.com>
22578 PR tree-optimization/69468
22579 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22580 (find_same_succ_bb): Handle ignore_edge_flags.
22582 2017-07-03 Tom de Vries <tom@codesourcery.com>
22584 PR tree-optimization/81192
22585 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22587 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22589 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22591 2017-07-03 Tom de Vries <tom@codesourcery.com>
22593 PR tree-optimization/81192
22594 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22595 BB_SAME_SUCC (bb) == NULL.
22597 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22599 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22602 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22604 * dumpfile.c: Include profile-count.h
22605 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22607 (insert_cond_bb): Update profile.
22608 * tree-cfg.h (insert_cond_bb): Update prototype.
22609 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22610 * tree-dump.c: Do not include tree-cfg.
22612 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22614 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22616 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22618 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22621 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22623 * tree-complex.c (expand_complex_div_wide): update profile.
22625 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22626 Alan Hayward <alan.hayward@arm.com>
22627 David Sherwood <david.sherwood@arm.com>
22629 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22630 (CORETYPES_H): New define.
22631 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22632 (insn-modes-inline.h, s-modes-inline-h): New rules.
22633 (generated_files): Add insn-modes-inline.h.
22634 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22635 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22636 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22637 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22638 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22639 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22640 (build/gencodes.o, build/genconditions.o): Likewise.
22641 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22642 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22643 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22644 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22645 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22646 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22647 * coretypes.h: Include everything up to real.h for generators.
22648 Include insn-modes.h first. Include wide-int-print.h after
22649 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22650 * machmode.h: Don't include insn-modes.h here.
22651 * function-tests.c: Remove includes of signop.h, machmode.h,
22652 double-int.h and wide-int.h.
22654 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22656 * optc-save-gen.awk: Likewise.
22657 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22658 * godump.c: Remove include of wide-int-print.h.
22659 * pretty-print.h: Likewise.
22660 * wide-int-print.cc: Likewise.
22661 * wide-int.cc: Likewise.
22662 * hash-map-tests.c: Remove include of signop.h.
22663 * hash-set-tests.c: Likewise.
22664 * rtl-tests.c: Likewise.
22665 * mkconfig.sh: Remove include of machmode.h.
22666 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22668 (emit_insn_modes_inline_h): ...this new function. Emit the code
22669 into an insn-modes-inline.h header file, adding appropriate
22670 include guards and end comments.
22671 (emit_insn_modes_c_header): Remove include of machmode.h.
22672 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22674 (main): Handle -i flag and call emit_insn_modes_inline_h when
22677 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22679 * tree-ssa-strlen.c (strinfo): Rename the length field to
22680 nonzero_chars. Add a full_string_p field.
22681 (compare_nonzero_chars, zero_length_string_p): New functions.
22682 (get_addr_stridx): Add an offset_out parameter.
22683 Use compare_nonzero_chars.
22684 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22685 (new_strinfo): Update after above changes to strinfo.
22686 (set_endptr_and_length): Set full_string_p.
22687 (get_string_length): Update after above changes to strinfo.
22688 (unshare_strinfo): Update call to new_strinfo.
22689 (maybe_invalidate): Likewise.
22690 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22691 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22692 as a uhwi instead of an shwi. Update after above changes to
22693 strinfo and new_strinfo.
22694 (zero_length_string): Assert that chainsi contains full strings.
22695 Use zero_length_string_p. Update call to new_strinfo.
22696 (adjust_related_strinfos): Update after above changes to strinfo.
22697 Copy full_string_p from origsi.
22698 (adjust_last_stmt): Use zero_length_string_p.
22699 (handle_builtin_strlen): Update after above changes to strinfo and
22700 new_strinfo. Install the lhs as the string length if the previous
22701 entry didn't describe a full string.
22702 (handle_builtin_strchr): Update after above changes to strinfo
22704 (handle_builtin_strcpy): Likewise.
22705 (handle_builtin_strcat): Likewise.
22706 (handle_builtin_malloc): Likewise.
22707 (handle_pointer_plus): Likewise.
22708 (handle_builtin_memcpy): Likewise. Track nonzero characters
22709 that aren't necessarily followed by a nul terminator.
22710 (handle_char_store): Likewise.
22712 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22714 PR tree-optimization/80769
22715 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22716 for malloc and calloc. Document the new invariant that all related
22717 strinfos have delayed lengths or none do.
22718 (verify_related_strinfos): Move earlier in file.
22719 (set_endptr_and_length): New function, split out from...
22720 (get_string_length): ...here. Also set the lengths of related
22722 (zero_length_string): Assert that chainsi has known (rather than
22724 (adjust_related_strinfos): Likewise.
22726 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22728 PR tree-optimization/81136
22729 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22730 assert that two references with the same misalignment have the same
22731 compile-time misalignment if those compile-time misalignments
22734 2017-07-01 Andi Kleen <ak@linux.intel.com>
22736 * print-tree.c (print_node): Print all attributes.
22738 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22740 * cfg.c (scale_bbs_frequencies): New function.
22741 * cfg.h (scale_bbs_frequencies): Declare it.
22742 * cfgloopanal.c (single_likely_exit): Cleanup.
22743 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22745 (scale_loop_profile): Likewise.
22746 (loop_version): Likewise.
22747 (create_empty_loop_on_edge): Update.
22748 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22749 scale_loop_frequencies, scale_loop_profile, loopify,
22750 loop_version): Update prototypes.
22751 * modulo-sched.c (sms_schedule): Update.
22752 * predict.c (unlikely_executed_edge_p): Also check probability.
22753 (probably_never_executed_edge_p): Fix typo.
22754 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22755 * tree-parloops.c (gen_parallel_loop): Update.
22756 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22757 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22758 * tree-ssa-loop-split.c (split_loop): Update.
22759 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22760 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22761 (vect_loop_versioning): Update.
22762 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22764 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22766 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22768 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22770 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22771 to keep profile consistent.
22773 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22775 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22776 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22777 * profile-count.h (max_safe_multiplier): Make unsigned.
22778 (profile_count::guessed_zero): New.
22780 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22782 * bb-reorder.c (fix_up_crossing_landing_pad,
22783 fix_crossing_conditional_branches): Use make_single_succ_edge
22784 to keep profile consistent.
22786 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22788 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22791 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22794 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22795 the right scopes, make sure cond_jump isn't preserved between multiple
22796 iterations. Search for fallthru edge whenever there are 3+ edges and
22797 use find_fallthru_edge for it.
22799 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22801 Patch by Alexander Monakov <amonakov@ispras.ru>
22802 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22803 probabilities consistently.
22805 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22807 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22808 emit_cmp_and_jump_insns.
22810 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22813 * tree-inline.c (expand_call_inline): Combine profile statuses.
22815 2017-06-30 Andrew Pinski <apinski@cavium.com>
22817 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22818 fold_stmt returned true.
22820 2017-06-30 Nathan Sidwell <nathan@acm.org>
22822 * ggc.h (empty_string): Delete.
22823 * cfgexpand.c (expand_asm_stmt): Use plain "".
22824 * optabs.c (expand_asm_memory_barrier): Likewise.
22825 * stringpool.c (empty_string): Delete.
22826 (digit_vector, digit_string): Delete.
22827 (ggc_alloc_string): Use plain "", don't optimize single digit
22828 strings. Use ggc_alloc_atomic.
22830 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22832 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22833 comparison set and one other set, use the cost of the non-comparison
22836 2017-06-30 Nathan Sidwell <nathan@acm.org>
22838 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22841 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22843 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22844 loops. Remove now unneeded calls to gimple_switch_set_label() that
22845 just set removed labels to NULL_TREE.
22847 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22849 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22850 (set_range_info): ...here. Only call set_range_info_raw if domain
22852 (set_nonzero_bits): Call set_range_info_raw.
22853 * tree-ssanames.h (set_range_info_raw): New.
22855 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22858 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22859 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22860 of nonimmediate_operand and <store_mask_constraint> instead of m
22861 for the input operand. For V8FI iterator, always split if input
22862 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22863 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22864 <store_mask_predicate> instead of register_operand and
22865 <store_mask_constraint> instead of v for the input operand. Make
22866 sure both operands aren't MEMs for if not <mask_applied>.
22868 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22870 * lto-wrapper.c (copy_file) Close both file descriptors before
22873 2017-06-30 Martin Liska <mliska@suse.cz>
22876 * multiple_target.c (create_dispatcher_calls): Make ifunc
22877 also for function that don't have calls or are not referenced.
22879 2017-06-30 Richard Biener <rguenther@suse.de>
22881 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22882 analyze the first scalar stmt. Move vector type computation
22883 for the BB case here from ...
22884 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22885 live operation processing in the SLP case properly.
22887 2017-06-30 Richard Biener <rguenther@suse.de>
22889 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22891 2017-06-30 Martin Liska <mliska@suse.cz>
22894 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22895 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22897 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22899 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22902 2017-06-30 Richard Biener <rguenther@suse.de>
22904 PR tree-optimization/81249
22905 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22906 condition reduction result to original scalar type.
22908 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22910 * profile-count.h (enum profile_quality): Fix typos and whitespace
22913 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22915 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22916 type for branch probabilities.
22918 2017-06-29 Julian Brown <julian@codesourcery.com>
22919 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22921 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22922 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22923 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22924 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22926 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22928 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22929 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22930 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22931 CC usage from generic code to here.
22932 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22933 CC usage into the target macros.
22935 2017-06-29 Maya Rashish <coypu@sdf.org>
22937 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22940 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22942 * arm/arm-builtins.c: Include profile-count.h
22943 * except.c (sjlj_emit_function_enter): Use
22944 profile_probability::unlikely.
22946 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22948 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22949 and tocrel_offset be pointer args rather than implicitly using
22951 (legitimate_constant_pool_address_p, rs6000_emit_move,
22952 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22953 tocrel_offset and use in toc_relative_expr_p call.
22954 (print_operand, print_operand_address): Use static tocrel_base_oac
22955 and tocrel_offset_oac.
22956 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22959 2017-06-29 Maya Rashish <coypu@sdf.org>
22961 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22963 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22965 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22966 objects, take into account only the alignment of 'op0' and 'mode1' if
22969 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22971 * ccmp.c (ccmp_tree_comparison_p): New function.
22972 (ccmp_candidate_p): Update to use above function.
22973 (get_compare_parts): New function.
22974 (expand_ccmp_next): Update to use new functions.
22975 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22977 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22978 take mode as argument.
22979 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22980 * expr.c (expand_expr_real_1): Pass mode as argument.
22982 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22984 * combine.c (combine_instructions): Print insns to dump_file, together
22987 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22989 * asan.c (asan_emit_stack_protection): Update.
22990 (create_cond_insert_point): Update.
22991 * auto-profile.c (afdo_propagate_circuit): Update.
22992 * basic-block.h (struct edge_def): Turn probability to
22993 profile_probability.
22994 (EDGE_FREQUENCY): Update.
22995 * bb-reorder.c (find_traces_1_round): Update.
22996 (better_edge_p): Update.
22997 (sanitize_hot_paths): Update.
22998 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22999 (make_single_succ_edge): Update.
23000 (check_bb_profile): Update.
23001 (dump_edge_info): Update.
23002 (update_bb_profile_for_threading): Update.
23003 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23004 probabilitycount to 0.
23005 * cfgbuild.c (compute_outgoing_frequencies): Update.
23006 * cfgcleanup.c (try_forward_edges): Update.
23007 (outgoing_edges_match): Update.
23008 (try_crossjump_to_edge): Update.
23009 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23010 (expand_gimple_tailcall): Update.
23011 (construct_init_block): Use make_single_succ_edge.
23012 (construct_exit_block): Use make_single_succ_edge.
23013 * cfghooks.c (verify_flow_info): Update.
23014 (redirect_edge_succ_nodup): Update.
23015 (split_edge): Update.
23016 (account_profile_record): Update.
23017 * cfgloopanal.c (single_likely_exit): Update.
23018 * cfgloopmanip.c (scale_loop_profile): Update.
23019 (set_zero_probability): Remove.
23020 (duplicate_loop_to_header_edge): Update.
23021 * cfgloopmanip.h (loop_version): Update prototype.
23022 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23023 (force_nonfallthru_and_redirect): Update.
23024 (update_br_prob_note): Update.
23025 (rtl_verify_edges): Update.
23026 (purge_dead_edges): Update.
23027 (rtl_lv_add_condition_to_bb): Update.
23028 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23029 * cgraphunit.c (init_lowered_empty_function): Update.
23030 (cgraph_node::expand_thunk): Update.
23031 * cilk-common.c: Include profile-count.h
23032 * dojump.c (inv): Remove.
23033 (jumpifnot): Update.
23034 (jumpifnot_1): Update.
23035 (do_jump_1): Update.
23037 (do_jump_by_parts_greater_rtx): Update.
23038 (do_compare_rtx_and_jump): Update.
23039 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23040 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23041 * dwarf2cfi.c: Include profile-count.h
23042 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23043 (sjlj_emit_dispatch_table): Likewise.
23044 * explow.c: Include profile-count.h
23045 * expmed.c (emit_store_flag_force): Update.
23046 (do_cmp_and_jump): Update.
23047 * expr.c (compare_by_pieces_d::generate): Update.
23048 (compare_by_pieces_d::finish_mode): Update.
23049 (emit_block_move_via_loop): Update.
23050 (store_expr_with_bounds): Update.
23051 (store_constructor): Update.
23052 (expand_expr_real_2): Update.
23053 (expand_expr_real_1): Update.
23054 * expr.h (try_casesi, try_tablejump): Update prototypes.
23055 * gimple-pretty-print.c (dump_probability): Update.
23056 (dump_profile): New.
23057 (dump_gimple_label): Update.
23058 (dump_gimple_bb_header): Update.
23059 * graph.c (draw_cfg_node_succ_edges): Update.
23060 * hsa-gen.c (convert_switch_statements): Update.
23061 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23062 (find_if_case_1): Update.
23063 (find_if_case_2): Update.
23064 * internal-fn.c (expand_arith_overflow_result_store): Update.
23065 (expand_addsub_overflow): Update.
23066 (expand_neg_overflow): Update.
23067 (expand_mul_overflow): Update.
23068 (expand_vector_ubsan_overflow): Update.
23069 * ipa-cp.c (good_cloning_opportunity_p): Update.
23070 * ipa-split.c (split_function): Use make_single_succ_edge.
23071 * ipa-utils.c (ipa_merge_profiles): Update.
23072 * loop-doloop.c (add_test): Update.
23073 (doloop_modify): Update.
23074 * loop-unroll.c (compare_and_jump_seq): Update.
23075 (unroll_loop_runtime_iterations): Update.
23076 * lra-constraints.c (lra_inheritance): Update.
23077 * lto-streamer-in.c (input_cfg): Update.
23078 * lto-streamer-out.c (output_cfg): Update.
23079 * mcf.c (adjust_cfg_counts): Update.
23080 * modulo-sched.c (sms_schedule): Update.
23081 * omp-expand.c (expand_omp_for_init_counts): Update.
23082 (extract_omp_for_update_vars): Update.
23083 (expand_omp_ordered_sink): Update.
23084 (expand_omp_for_ordered_loops): Update.
23085 (expand_omp_for_generic): Update.
23086 (expand_omp_for_static_nochunk): Update.
23087 (expand_omp_for_static_chunk): Update.
23088 (expand_cilk_for): Update.
23089 (expand_omp_simd): Update.
23090 (expand_omp_taskloop_for_outer): Update.
23091 (expand_omp_taskloop_for_inner): Update.
23092 * omp-simd-clone.c (simd_clone_adjust): Update.
23093 * optabs.c (expand_doubleword_shift): Update.
23094 (expand_abs): Update.
23095 (emit_cmp_and_jump_insn_1): Update.
23096 (expand_compare_and_swap_loop): Update.
23097 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23098 * predict.c (predictable_edge_p): Update.
23099 (edge_probability_reliable_p): Update.
23100 (set_even_probabilities): Update.
23101 (combine_predictions_for_insn): Update.
23102 (combine_predictions_for_bb): Update.
23103 (propagate_freq): Update.
23104 (estimate_bb_frequencies): Update.
23105 (force_edge_cold): Update.
23106 * profile-count.c (profile_count::dump): Add missing space into dump.
23107 (profile_count::debug): Add newline.
23108 (profile_count::differs_from_p): Explicitly convert to unsigned.
23109 (profile_count::stream_in): Update.
23110 (profile_probability::dump): New member function.
23111 (profile_probability::debug): New member function.
23112 (profile_probability::differs_from_p): New member function.
23113 (profile_probability::differs_lot_from_p): New member function.
23114 (profile_probability::stream_in): New member function.
23115 (profile_probability::stream_out): New member function.
23116 * profile-count.h (profile_count_quality): Rename to ...
23117 (profile_quality): ... this one.
23118 (profile_probability): New.
23119 (profile_count): Update.
23120 * profile.c (compute_branch_probabilities): Update.
23121 * recog.c (peep2_attempt): Update.
23122 * sched-ebb.c (schedule_ebbs): Update.
23123 * sched-rgn.c (find_single_block_region): Update.
23124 (compute_dom_prob_ps): Update.
23125 (schedule_region): Update.
23126 * sel-sched-ir.c (compute_succs_info): Update.
23127 * stmt.c (struct case_node): Update.
23128 (do_jump_if_equal): Update.
23129 (get_outgoing_edge_probs): Update.
23130 (conditional_probability): Update.
23131 (emit_case_dispatch_table): Update.
23132 (expand_case): Update.
23133 (expand_sjlj_dispatch_table): Update.
23134 (emit_case_nodes): Update.
23135 * targhooks.c: Update.
23136 * tracer.c (better_p): Update.
23137 (find_best_successor): Update.
23138 * trans-mem.c (expand_transaction): Update.
23139 * tree-call-cdce.c: Update.
23140 * tree-cfg.c (gimple_split_edge): Upate.
23141 (move_sese_region_to_fn): Upate.
23142 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23143 * tree-eh.c (lower_resx): Upate.
23144 (cleanup_empty_eh_move_lp): Upate.
23145 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23146 * tree-inline.c (copy_edges_for_bb): Update.
23147 (copy_cfg_body): Update.
23148 * tree-parloops.c (gen_parallel_loop): Update.
23149 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23150 (gimple_gen_time_profiler): Update.
23151 * tree-ssa-dce.c (remove_dead_stmt): Update.
23152 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23153 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23154 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23155 (unloop_loops): Update.
23156 (try_peel_loop): Update.
23157 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23158 * tree-ssa-loop-split.c (connect_loops): Update.
23159 (split_loop): Update.
23160 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23161 (hoist_guard): Update.
23162 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23163 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23164 (value_replacement): Update.
23165 * tree-ssa-reassoc.c (branch_fixup): Update.
23166 * tree-ssa-tail-merge.c (replace_block_by): Update.
23167 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23168 (create_edge_and_update_destination_phis): Update.
23169 (compute_path_counts): Update.
23170 (recompute_probabilities): Update.
23171 (update_joiner_offpath_counts): Update.
23172 (freqs_to_counts_path): Update.
23173 (duplicate_thread_path): Update.
23174 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23175 (struct switch_conv_info): Update.
23176 (gen_inbound_check): Update.
23177 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23178 (vect_do_peeling): Update.
23179 (vect_loop_versioning): Update.
23180 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23181 (optimize_mask_stores): Update.
23182 * ubsan.c (ubsan_expand_null_ifn): Update.
23183 * value-prof.c (gimple_divmod_fixed_value): Update.
23184 (gimple_divmod_fixed_value_transform): Update.
23185 (gimple_mod_pow2): Update.
23186 (gimple_mod_pow2_value_transform): Update.
23187 (gimple_mod_subtract): Update.
23188 (gimple_mod_subtract_transform): Update.
23189 (gimple_ic): Update.
23190 (gimple_stringop_fixed_value): Update.
23191 (gimple_stringops_transform): Update.
23192 * value-prof.h: Update.
23194 2017-06-29 Carl Love <cel@us.ibm.com>
23196 * config/rs6000/rs6000-c.c: Add support for built-in functions
23197 vector signed int vec_signed (vector float);
23198 vector signed long long vec_signed (vector double);
23199 vector signed int vec_signed2 (vector double, vector double);
23200 vector signed int vec_signede (vector double);
23201 vector signed int vec_signedo (vector double);
23202 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23203 instruction generator.
23204 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23205 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23206 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23208 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23209 vunsignede_v2df): Add define_expands.
23210 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23211 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23212 VEC_UNSIGNEDO): Add definitions.
23213 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23214 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23215 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23216 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23217 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23218 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23219 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23220 vec_unsignede, vec_unsignedo): Add builtin defines.
23221 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23223 * doc/extend.texi: Update the built-in documentation file for the
23224 new built-in functions.
23226 2017-06-29 Richard Biener <rguenther@suse.de>
23228 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23229 reduction chains to LOOP_VINFO_REDUCTIONS.
23230 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23231 SLP reductions after processing reduction chains.
23233 2017-06-29 Nathan Sidwell <nathan@acm.org>
23235 * builtins.c (fold_builtin_FUNCTION): Use
23236 lang_hooks.decl_printable_name.
23238 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23240 PR middle-end/81194
23241 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23242 with only one label.
23243 * stmt.c (expand_case): Assert NCASES is greater than one.
23245 2017-06-29 Richard Biener <rguenther@suse.de>
23247 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23249 (group_case_labels): Likewise.
23250 (find_taken_edge): Push sanity checking on val to workers...
23251 (find_taken_edge_cond_expr): ... here
23252 (find_taken_edge_switch_expr): ... and here, handle cases
23253 with just a default label.
23254 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23255 (group_case_labels): Likewise.
23256 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23257 group_case_labels does anything cleanup the CFG again.
23259 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23261 PR tree-optimization/81196
23262 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23263 exit condition comparing two IVs.
23265 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23267 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23268 profile to the dummy entry at the end of the list of architectures.
23269 * config/arm/arm-cpu-cdata.h: Regenerated.
23271 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23272 Michael Collison <michael.collison@arm.com>
23275 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23277 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23278 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23279 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23280 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23281 of shift when the shift amount is masked with constant equal to
23282 the size of the mode.
23283 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23286 2017-06-29 Martin Liska <mliska@suse.cz>
23288 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23290 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23293 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23296 (edge_growth_cache_entry::edge_growth_cache_entry): New
23298 (reset_edge_growth_cache): Update to use constructor.
23300 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23302 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23303 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23304 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23306 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23308 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23309 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23311 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23313 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23314 (*-linux-uclibc*): Add t-uclibc tmake_file.
23315 * config/t-musl: New.
23316 * config/t-uclibc: New.
23318 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23320 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23322 (gen_comm_data): Emit architectural setting of arch_prof.
23323 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23325 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23326 (armv8-m.base, armv8-m.main): Likewise.
23327 * arm-protos.h (arm_build_target): Add profile field.
23328 (arch_option): Likewise.
23329 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23331 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23332 arm_active_target.profile.
23334 2017-06-28 Richard Biener <rguenther@suse.de>
23336 PR middle-end/81227
23337 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23338 TYPE_OVERFLOW_WRAPS.
23339 * match.pd (negate_expr_p): Likewise.
23340 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23341 fold_build2, not fold_binary.
23343 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23345 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23346 Convert memory address to Pmode.
23347 (aarch64_print_operand): Assert MEM operands are always Pmode.
23349 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23352 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23353 Remove redundant if.
23354 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23355 * config/arm/aarch-common-protos.h
23356 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23357 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23359 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23362 * multiple_target.c (create_dispatcher_calls): Set the default
23363 clone to be static, not public.
23365 2017-06-28 Richard Biener <rguenther@suse.de>
23367 * tree-vect-loop.c (vectorizable_reduction): Move special
23368 cond reduction IV var creation ...
23369 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23370 parameter. Use STMT_VINFO_VECTYPE.
23371 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23374 2017-06-28 Martin Liska <mliska@suse.cz>
23377 * ipa-visibility.c (non_local_p): Handle visibility.
23379 2017-06-28 Martin Liska <mliska@suse.cz>
23382 * common.opt: Add IntegerRange to various options.
23383 * opt-functions.awk (integer_range_info): New function.
23384 * optc-gen.awk: Add integer_range_info to cl_options struct.
23385 * opts-common.c (decode_cmdline_option): Handle
23386 CL_ERR_INT_RANGE_ARG.
23387 (cmdline_handle_error): Likewise.
23388 * opts.c (print_filtered_help): Show valid interval in
23389 when --help is provided.
23390 * opts.h (struct cl_option): Add range_min and range_max fields.
23391 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23393 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23395 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23396 (x * C EQ/NE y * C): New transformation.
23398 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23400 * genmultilib (combination_space): Accept '+' in option names.
23402 2017-06-28 Martin Liska <mliska@suse.cz>
23405 * asan.c (instrument_derefs): Bail out inner references
23406 that are hard register variables.
23408 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23411 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23412 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23414 2017-06-28 Richard Biener <rguenther@suse.de>
23416 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23417 (vect_get_slp_defs): Adjust.
23418 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23420 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23422 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23423 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23424 (vectorizable_reduction): Adjust.
23425 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23427 (vect_get_slp_defs): Likewise.
23428 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23429 (vectorizable_bswap): Adjust.
23430 (vectorizable_call): Likewise.
23431 (vectorizable_conversion): Likewise.
23432 (vectorizable_assignment): Likewise.
23433 (vectorizable_shift): Likewise.
23434 (vectorizable_operation): Likewise.
23435 (vectorizable_store): Likewise.
23436 (vectorizable_condition): Likewise.
23437 (vectorizable_comparison): Likewise.
23439 2017-06-28 Michael Collison <michael.collison@arm.com>
23442 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23444 (arm_gen_movmemqi): Removed unused variable 'i'.
23445 Convert 'for' loop into 'while' loop.
23446 (arm_expand_prologue): Remove last unnecessary set of insn.
23447 (thumb_pop): Remove unused variable 'pushed_words'.
23448 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23450 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23452 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23453 * config/s390/s390-protos.h: Add prototype of
23454 s390_rel_address_ok_p.
23455 * config/s390/s390.c (s390_got_symbol): New function.
23456 (s390_rel_address_ok_p): New function.
23457 (legitimize_pic_address): Use s390_rel_address_ok_p.
23458 (s390_load_got): Use s390_got_symbol.
23459 (s390_option_override): Issue error if
23460 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23461 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23463 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23465 2017-06-27 Andrew Pinski <apinski@cavium.com>
23467 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23468 (X * copysign (1.0, X)): New pattern.
23469 (X * copysign (1.0, -X)): New pattern.
23470 (copysign (-1.0, CST)): New pattern.
23472 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23474 * genmultilib (combination_space): Remove variable.
23475 Validate reuse rules against regular expression for any sequence
23476 of multilib options in any order.
23478 2017-06-27 Michael Collison <michael.collison@arm.com>
23480 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23481 call aarch64_split_simd_combine.
23482 * (aarch64_combine_internal<mode>): Delete pattern.
23483 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23484 Allow register and subreg operands.
23486 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23488 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23489 specific need, just fallback on defaults.
23490 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23492 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23493 Olivier Hainque <hainque@adacore.com>
23495 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23497 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23498 targets. Pick a default if no particular attempt applied.
23499 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23502 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23504 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23505 (x86_64-wrs-vxworks7): Likewise.
23507 2017-06-27 Marek Polacek <polacek@redhat.com>
23510 * ubsan.c (instrument_null): Check get_base_address's result for null.
23512 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23514 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23516 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23518 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23519 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23520 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23521 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23522 New function types.
23523 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23524 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23525 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23526 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23527 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23528 BUILT_IN_FEUPDATEENV): New builtins.
23529 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23530 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23531 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23532 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23534 (builtin_structptr_types): Adjust size.
23535 * tree.c (builtin_structptr_types): Add four entries.
23537 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23538 Olivier Hainque <hainque@adacore.com>
23540 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23541 (TLS_SYM): New local macro, forcing reference to __tls__ on
23542 link command lines for VxWorks 7 RTPs, triggering initialization
23544 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23545 OS features TLS support, true for RTPs on VxWorks 7.
23546 * config/vxworks.c (vxworks_override_options): Setup emutls
23549 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23551 * predict.c (test_prediction_value_range): Use -1U instead of -1
23552 to avoid narrowing conversion warning.
23553 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23554 to avoid narrowing conversion warning.
23555 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23557 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23559 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23561 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23562 64bit configurations.
23563 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23564 (SIZE_TYPE): Likewise.
23565 * config/vxworks.c (vxworks_emutls_var_fields): Use
23566 long_unsigned_type_node instead of unsigned_type_node as the offset
23567 field type, which is "pointer" mode in emutls.c.
23569 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23572 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23574 PR middle-end/81207
23575 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23576 gimple_vuse copying separately from gimple_vdef copying.
23578 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23580 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23583 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23584 Olivier Hainque <hainque@adacore.com>
23586 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23587 for all vxworks7 targets.
23588 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23589 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23590 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23591 variations for VX6/VX7 and 32/64bits later on in ...
23592 (VXWORKS_LIB_SPEC): Leverage new macros.
23593 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23594 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23596 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23598 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23599 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23601 2017-06-26 Carl Love <cel@us.ibm.com>
23603 * config/rs6000/rs6000-c.c: Add support for built-in functions
23604 vector bool char vec_reve (vector bool char);
23605 vector signed char vec_reve (vector signed char);
23606 vector unsigned char vec_reve (vector unsigned char);
23607 vector bool int vec_reve (vector bool int);
23608 vector signed int vec_reve (vector signed int);
23609 vector unsigned int vec_reve (vector unsigned int);
23610 vector bool long long vec_reve (vector bool long long);
23611 vector signed long long vec_reve (vector signed long long);
23612 vector unsigned long long vec_reve (vector unsigned long long);
23613 vector bool short vec_reve (vector bool short);
23614 vector signed short vec_reve (vector signed short);
23615 vector double vec_reve (vector double);
23616 vector float vec_reve (vector float);
23617 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23618 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23619 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23620 (altivec_vreve): New pattern.
23621 * config/rs6000/altivec.h (vec_reve): New define.
23622 * doc/extend.texi (vec_rev): Update the built-in documentation file
23623 for the new built-in functions.
23625 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23627 PR tree-optimization/71815
23628 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23630 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23632 (slsr_process_phi): Likewise.
23633 (replace_uncond_cands_and_profitable_phis): Don't replace a
23634 multiply candidate with a stride of 1 (copy or cast).
23635 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23637 (lowest_cost_path): Likewise.
23638 (total_savings): Likewise.
23640 2017-06-26 Richard Biener <rguenther@suse.de>
23643 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23644 Use def_builtin_pure for all gather builtins.
23646 2017-06-26 Richard Biener <rguenther@suse.de>
23648 PR tree-optimization/81203
23649 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23650 non-dominating BBs.
23652 2017-06-26 Marek Polacek <polacek@redhat.com>
23655 * doc/invoke.texi: Document -Wmultistatement-macros.
23657 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23659 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23660 arm_neon_ok_no_float_abi effective target.
23662 2017-06-26 Richard Biener <rguenther@suse.de>
23664 PR tree-optimization/80928
23665 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23666 (copy_bbs): Set BB_DUPLICATED flag early.
23667 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23669 (execute_on_shrinking_pred): Likewise.
23670 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23671 BB_DUPLICATED blocks.
23672 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23673 iterate over all PHIs considering removal of *gsi.
23675 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23677 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23680 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23682 * config/rs6000/rs6000-string.c: (expand_block_clear,
23683 do_load_for_compare, select_block_compare_mode,
23684 compute_current_alignment, expand_block_compare,
23685 expand_strncmp_align_check, expand_strn_compare,
23686 expand_block_move, rs6000_output_load_multiple)
23687 Move functions related to string/block move/compare
23688 to a separate file.
23689 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23690 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23691 for this function which is now used in two files.
23692 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23693 * config.gcc: Add rs6000-string.o to extra_objs for
23694 targets powerpc*-*-* and rs6000*-*-*.
23696 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23699 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23700 32-bit, since indexed is not valid for DImode.
23701 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23702 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23703 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23704 (define_peephole2 for Altivec d-form store): Likewise.
23707 * multiple_target.c (create_dispatcher_calls): Only create the
23708 dispatcher call if the function is the default clone of a
23709 versioned function.
23711 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23713 PR middle-end/80902
23714 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23715 a call, force the call to not be a tail call.
23717 2017-06-23 Jeff Law <law@redhat.com>
23719 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23722 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23724 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23725 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23726 handling for early expansion of vector shifts (sl,sr,sra,rl).
23727 (builtin_function_type): Add vector shift right instructions
23728 to the unsigned argument list.
23730 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23732 rtl-optimizatoin/79286
23733 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23734 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23735 trap. PIC register plus a const unspec without offset can never trap.
23737 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23739 * tree.h (builtin_structptr_type): New type.
23740 (builtin_structptr_types): Declare new array.
23741 * tree.c (builtin_structptr_types): New array.
23742 (free_lang_data, build_common_tree_nodes): Use it.
23744 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23747 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23750 2017-06-22 Matt Turner <mattst88@gmail.com>
23752 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23753 Lake models to skylake case. Assume skylake for unknown
23754 models with clflushopt.
23756 2017-06-22 Jeff Law <law@redhat.com>
23758 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23759 frame sizes that do not satisfy aarch64_uimm12_shift.
23761 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23763 * profile-count.h (apply_probability,
23764 apply_scale, probability_in): Fix checks for zero.
23766 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23768 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23769 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23771 2016-06-22 Richard Biener <rguenther@suse.de>
23773 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23774 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23775 REDUC_MAX_EXPR support.
23776 (vectorizable_reduction): Likewise.
23777 (vect_create_epilog_for_reduction): Likewise.
23779 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23781 * match.pd (A / (1 << B) -> A >> B): New.
23782 * generic-match-head.c: Include optabs-tree.h.
23783 * gimple-match-head.c: Likewise.
23784 * optabs-tree.h (target_supports_op_p): New.
23785 * optabs-tree.c (target_supports_op_p): New.
23787 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23789 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23790 $gcc_cv_ld --help output.
23791 (gcc_cv_ld_demangle): Likewise.
23792 (gcc_cv_ld_eh_frame_hdr): Likewise.
23793 (gcc_cv_ld_pie): Likewise.
23794 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23795 (gcc_cv_ld_buildid): Likewise.
23796 (gcc_cv_ld_sysroot): Likewise.
23797 (ld_bndplt_support): Likewise.
23798 (ld_pushpopstate_support): Likewise.
23799 * configure: Regenerate.
23800 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23802 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23805 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23806 operands indexes to avoid gap between operands and match_dups.
23808 2017-06-21 Andrew Pinski <apinski@cavium.com>
23810 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23811 Increment Arith_shift and Arith_shift_reg by 1.
23812 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23814 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23815 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23816 (aarch64_strip_extend): Add new argument and test for it.
23817 (aarch64_cheap_mult_shift_p): New function.
23818 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23819 add a cost if it is true.
23820 Update calls to aarch64_strip_extend.
23821 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23823 2017-06-21 Andrew Pinski <apinski@cavium.com>
23825 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23827 (thunderxt88): Likewise.
23828 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23829 (thunderx_prefetch_tune): New variable.
23830 (thunderx2t99_prefetch_tune): Update for the correct values.
23831 (thunderxt88_tunings): New variable.
23832 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23833 generic_prefetch_tune.
23834 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23836 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23838 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23839 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23840 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23841 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23842 (aarch64_atomic_cas<mode>, GPI): Likewise.
23844 2017-06-21 Martin Liska <mliska@suse.cz>
23846 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23847 statements on cold and hot labels.
23848 * predict.c (tree_estimate_probability_bb): Remove the
23849 prediction from this place.
23851 2017-06-21 Martin Liska <mliska@suse.cz>
23853 PR tree-optimization/79489
23854 * gimplify.c (maybe_add_early_return_predict_stmt): New
23856 (gimplify_return_expr): Call the function.
23857 * predict.c (tree_estimate_probability_bb): Remove handling
23859 * predict.def: Update comment about early return predictor.
23860 * gimple-predict.h (is_gimple_predict): New function.
23861 * predict.def: Change default value of early return to 66.
23862 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23864 * passes.def: Put pass_strip_predict_hints to the beginning of
23867 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23869 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23870 FUNCTION_DECL declarations.
23871 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23873 (dwaf2out_decl): Likewise.
23874 * godump.c (go_early_global_decl): Skip call to the real debug hook
23875 for FUNCTION_DECL declarations.
23876 * passes.c (rest_of_decl_compilation): Skip call to the
23877 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23878 -fdump-go-spec is passed.
23880 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23882 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23883 Reorder for compactness.
23884 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23885 (def_builtin_pure, def_builtin_pure2): New functions.
23886 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23888 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23890 * match.pd (nop_convert): New predicate.
23891 ((A +- CST1) +- CST2): Allow some NOP conversions.
23893 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23896 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23897 with ctors/dtors if GOVD_SHARED is set.
23899 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23901 * config/aarch64/aarch64.md (movti_aarch64):
23902 Emit mov rather than orr.
23903 (movtf_aarch64): Likewise.
23904 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23905 Emit mov rather than orr.
23907 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23909 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23910 Swap alternatives, make integer dup more expensive.
23912 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23914 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23915 Return true for non-tls symbols.
23917 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23919 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23920 (cortex-a75): Likewise.
23921 (cortex-a75.cortex-a55): Likewise.
23922 * config/aarch64/aarch64-tune.md: Regenerate.
23923 * doc/invoke.texi (-mtune): Document new values for -mtune.
23925 2017-06-21 Tom de Vries <tom@codesourcery.com>
23927 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23928 stack_size feature.
23929 (Effective-Target Keywords, Other attributes): Suggest using
23930 dg-add-options stack_size feature to get stack limit in stack_size
23931 effective target documentation.
23933 2017-06-21 Julian Brown <julian@codesourcery.com>
23934 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23936 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23937 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23938 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23940 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23941 attribute type list for neon_multiply.
23942 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23943 attribute type list for neon_multiply.
23944 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23945 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23946 attribute type list for neon_multiply.
23947 * config/arm/types.md (crypto_pmull): Add.
23948 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23949 attribute type list.
23951 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23953 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23956 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23958 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23959 to make sure not to dereference a NULL cost_classes_ptr pointer.
23961 2017-06-20 Carl Love <cel@us.ibm.com>
23963 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23964 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23965 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23966 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23967 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23968 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23969 VMULOSW): New enum "unspec" values.
23970 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23971 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23972 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23973 altivec_vmulosw): New patterns.
23974 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23975 VMULOSW): Add definitions.
23977 2017-06-20 Julia Koval <julia.koval@intel.com>
23979 * config/i386/i386.c: Fix rounding expand for new pattern.
23980 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23982 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23984 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23985 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23987 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23989 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23992 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23994 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23995 architecture, then by implementer ID.
23996 * config/aarch64/aarch64-tune.md: Regenerate.
23998 2017-06-20 Richard Biener <rguenther@suse.de>
24000 PR middle-end/81097
24001 * fold-const.c (split_tree): Fold to type before negating.
24003 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24005 * diagnostic-show-locus.c
24006 (selftest::test_fixit_deletion_affecting_newline): New function.
24007 (selftest::diagnostic_show_locus_c_tests): Call it.
24009 2017-06-20 Andreas Schwab <schwab@suse.de>
24012 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24015 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24017 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24018 __ARM_FEATURE_COPROC according to support.
24020 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24022 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24023 Rewritten to avoid overflow for > 32-bit pointers.
24026 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24027 by removing enum keyword.
24028 (ubsan_type_descriptor): Likewise. Formatting fix.
24031 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24032 splitter): Require TARGET_SSE2 in the condition.
24034 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24037 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24038 for doing vector set of SFmode on ISA 3.0.
24039 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24040 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24042 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24043 SFmode value into a V4SF variable that was extracted from another
24044 V4SF variable without converting the element to double precision
24045 and back to single precision vector format.
24046 (vsx_insert_extract_v4sf_p9_2): Likewise.
24048 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24050 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24051 in UWHI to avoid undefined overflow.
24054 * ubsan.h (enum ubsan_encode_value_phase): New.
24055 (ubsan_encode_value): Change second argument to
24056 enum ubsan_encode_value_phase with default value of
24057 UBSAN_ENCODE_VALUE_GENERIC.
24058 * ubsan.c (ubsan_encode_value): Change second argument to
24059 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24060 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24061 create_tmp_var_raw instead of create_tmp_var and use a
24063 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24064 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24065 ubsan_encode_value callers.
24068 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24069 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24070 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24072 2017-06-19 Richard Biener <rguenther@suse.de>
24074 PR middle-end/81118
24075 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24076 estimates if we changed anything.
24078 2017-06-19 Richard Biener <rguenther@suse.de>
24080 PR tree-optimization/80887
24081 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24082 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24083 simplified lookups, then reset mprts_hook.
24084 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24086 (try_to_simplify): Likewise.
24088 2017-06-19 Martin Liska <mliska@suse.cz>
24091 * gimplify.c (gimplify_switch_expr):
24092 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24094 2017-06-19 Martin Liska <mliska@suse.cz>
24096 * doc/install.texi: Document that PGO runs in 4 stages.
24098 2017-06-19 Martin Liska <mliska@suse.cz>
24101 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24102 to dispatcher function name.
24103 * multiple_target.c (replace_function_decl): New function.
24104 (create_dispatcher_calls): Redirect both edges and references.
24106 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24108 * profile-count.c (profile_count::dump): Dump quality.
24109 (profile_count::differs_from_p): Update for unsigned val.
24110 * profile-count.h (profile_count_quality): New enum.
24111 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24113 2017-06-19 Richard Biener <rguenther@suse.de>
24115 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24116 struct function as arg.
24117 (estimate_numbers_of_iterations): Export overload with loop arg.
24118 (free_numbers_of_iterations_estimates_loop): Use an overload of
24119 free_numbers_of_iterations_estimates instead.
24120 * tree-cfg.c (remove_bb): Adjust.
24121 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24122 * tree-parloops.c (gen_parallel_loop): Likewise.
24123 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24125 (tree_unroll_loops_completely): Likewise.
24126 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24127 Use an overload instead and export.
24128 (estimated_loop_iterations): Adjust.
24129 (max_loop_iterations): Likewise.
24130 (likely_max_loop_iterations): Likewise.
24131 (estimate_numbers_of_iterations): Take struct function as arg
24133 (loop_exits_before_overflow): Adjust.
24134 (free_numbers_of_iterations_estimates_loop): Use an overload.
24135 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24136 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24138 2017-06-19 Richard Biener <rguenther@suse.de>
24141 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24142 RANGE_EXPR conservatively.
24144 2017-06-16 Carl Love <cel@us.ibm.com>
24146 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24147 definitions for vec_float, vec_float2, vec_floato,
24148 vec_floate built-ins.
24149 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24150 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24152 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24153 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24154 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24155 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24156 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24157 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24158 vec_floato): Add builtin defines.
24159 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24160 Update the built-in documentation file for the new built-in
24163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24165 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24166 (mthumb): Mark as the negative of -marm.
24168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24170 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24172 (ARM Options, -mtune): Document that this accepts the same
24173 extension options as -mcpu.
24174 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24176 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24178 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24179 permitted extensions.
24181 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24183 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24184 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24185 (armv8-m.main): Add option +nodsp.
24186 * config/arm/arm-cpu-cdata.h: Regenerated.
24188 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24190 * config/arm/t-fuchsia: New file.
24191 * config.gcc (arm*-*-fuchsia*): Use it.
24193 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24195 * config/arm/t-symbian: Rewrite for new option infrastructure.
24197 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24199 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24200 (MULTILIB_REQUIRED): Likewise.
24202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24204 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24205 (MULTILIB_RESUE): Likewise.
24206 (MULTILIB_MATCHES): Likewise.
24207 (MULTLIB_REQUIRED): Likewise.
24209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24211 * config/arm/t-rtems: Rewrite for new option framework.
24213 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24215 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24216 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24217 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24218 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24219 * config/arm/t-multilib: ... here.
24220 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24221 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24222 armv7-a and armv8*-a when A-profile libraries have not been built.
24223 * config/arm/t-rmprofile: Rewrite.
24225 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24227 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24228 with a backslash. Remove the backslash after substituting unescaped
24230 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24232 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24234 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24235 the driver to pass through the default setting of -mfloat-abi.
24236 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24238 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24239 (all_feat_combs): New rule.
24240 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24242 * config/arm/t-aprofile: Rewrite.
24244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24246 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24247 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24248 fpu is not specified by the user/command-line.
24249 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24250 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24251 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24252 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24253 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24254 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24256 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24258 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24259 * config/arm/t-arm-elf: Rewritten.
24261 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24263 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24264 have some floating-point instructions.
24265 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24266 (TARGET_MAYBE_HARD_FLOAT): New macro.
24267 * config/arm/arm-builtins.c (arm_init_builtins): Use
24268 TARGET_MAYBE_HARD_FLOAT.
24269 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24271 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24273 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24274 (configargs.h): Include it.
24275 (arm_print_hint_for_fpu_option): New function.
24276 (arm_parse_fpu_option): New function.
24277 (candidate_extension): New class.
24278 (arm_canon_for_multilib): New function.
24279 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24280 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24281 (ARCH_CANONICAL_SPECS): New macro.
24282 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24284 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24286 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24287 are set after handling multilib fragments. Set target_cpu_default2
24290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24292 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24294 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24295 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24297 (check_arch): Likewise.
24298 * config/arm/arm.c (arm_configure_build_target): Handle
24299 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24300 options in the default.
24302 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24304 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24305 when an option is an alias of another.
24306 * config/arm/parsecpu.awk (optalias): New parser token.
24307 (gen_comm_data): Mark non-alias options as such. Emit entries
24308 for extension aliases.
24309 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24310 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24311 (armv6kz, armv6zk, armv6t2): Likewise.
24312 (armv7): Make vfpv3-d16 an alias.
24313 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24315 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24316 Sort in canonical order.
24317 (armv8-a): Sort in canonical order.
24318 (armv8.1-a, armv8.2-a): Likewise.
24319 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24321 (cortex-a9): Sort in canonical order.
24322 * config/arm/arm.c (selftests.h): Include it.
24323 (arm_test_cpu_arch_data): New function.
24324 (arm_run_self_tests): New function.
24325 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24326 (targetm): Move declaration to the end of the file.
24327 * arm-cpu-cdata.h: Regenerated.
24329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24331 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24332 call to target_mode_check describing the type of option passed.
24333 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24334 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24335 arm_parse_cpu_option_name to match parameters against list of
24337 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24338 arm_arch_core_flags data structure.
24339 * config/arm/arm-cpu_cdata.h: Regenerated.
24341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24343 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24345 (arm_print_hint_for_cpu_option): Likewise.
24346 (arm_print_hint_for_arch_option): Likewise.
24347 (arm_parse_cpu_option_name): Likewise.
24348 (arm_parse_arch_option_name): Likewise.
24349 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24350 of entries in the all_fpus list.
24351 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24352 (arm_parse_cpu_option_name): Declare.
24353 (arm_parse_arch_option_name): Declare.
24354 (arm_parse_option_features): Declare.
24355 (arm_intialize_isa): Declare.
24356 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24358 (gen_comm_data): ... here. Make definitions non-static.
24359 * config/arm/arm-cpu-data.h: Regenerated.
24360 * config/arm/arm-cpu-cdata.h: Regenerated.
24362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24364 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24365 (cpu_arch_extension): New structure.
24366 (cpu_arch_option, arch_option, cpu_option): New structures.
24367 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24368 architecture types.
24369 (gen_data): Generate new format data tables.
24370 * config/arm/arm.c (cpu_tune): New structure.
24371 (cpu_option, processors): Delete.
24372 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24373 (arm_print_hint_for_cpu_option): ... this and ...
24374 (arm_print_hint_for_arch_option): ... this.
24375 (arm_parse_arch_cpu_name): Delete. Replace with ...
24376 (arm_parse_cpu_option_name): ... this and ...
24377 (arm_parse_arch_option_name): ... this.
24378 (arm_unrecognized_feature): Change type of target parameter to
24380 (arm_parse_arch_cpu_features): Delete. Replace with ...
24381 (arm_parse_option_features): ... this.
24382 (arm_configure_build_target): Rework to use new configuration data
24384 (arm_print_tune_info): Rework for new configuration data tables.
24385 * config/arm/arm-cpu-data.h: Regenerated.
24386 * config/arm/arm-cpu.h: Regenerated.
24388 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24390 * Makefile.in (OBJS): Move sbitmap.o from here ...
24391 (OBJS-libcommon): ... to here.
24393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24395 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24396 (ISA_ALL_CRYPTO): New macro.
24397 (ISA_ALL_SIMD): New macro
24398 (ISA_ALL_FP): New macro.
24399 * config/arm/arm.c (fpu_bitlist): Update initializer.
24400 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24402 (arm9e): Add fpu. Add option for nofp
24403 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24404 (arm926ej-s, arm1026ej-s): Likewise.
24405 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24406 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24407 neon-fp16, neon-vfpv4, nofp and nosimd.
24408 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24409 (cortex-a8): Add fpu. Add option for nofp.
24410 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24411 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24412 (cortex-r4f): Add fpu.
24413 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24414 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24416 (cortex-r8): Likewise.
24417 (cortex-m4): Add fpu. Add option for nofp.
24418 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24419 (cortex-a17.cortex-a7): Likewise.
24420 (cortex-a32): Add fpu. Add options for crypto and nofp.
24421 (cortex-a35, cortex-a53): Likewise.
24422 (cortex-a57): Add fpu. Add option for crypto.
24423 (cortex-a72, cortex-a73): Likewise.
24424 (exynos-m1): Likewise.
24425 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24426 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24427 (cortex-m33): Add fpu. Add option for nofp.
24428 * config/arm/arm-cpu-cdata.h: Regenerated
24429 * config/arm/arm-cpu-data.h: Regenerated.
24431 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24433 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24434 (armv5te, armv5tej): Likewise.
24435 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24436 (armv7): Add options fp and vfpv3-d16.
24437 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24438 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24440 (armv7ve): Likewise.
24441 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24442 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24443 (armv8-a): Add nocrypto option.
24444 (armv8.1-a, armv8.2-a): Likewise.
24445 (armv8-m.main): add options fp, fp.dp and nofp.
24447 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24449 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24451 (armv8-a+crc): Delete.
24452 (armv8.1-a): Add options simd, crypto and nofp.
24453 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24454 (armv8.2-a+fp16): Delete.
24455 (armv8-m.main): Add option dsp.
24456 (armv8-m.main+dsp): Delete.
24457 (cortex-a8): Add fpu. Add nofp option.
24458 (cortex-a9): Add fpu. Add nofp and nosimd options.
24459 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24460 link to main cpu and architecture data structures.
24461 (gen_comm_data): Only put isa attributes from the main architecture
24463 (option): New statement for architecture and CPU entries.
24464 * arm.c (struct cpu_option): New structure.
24465 (struct processors): Add entry for options.
24466 (arm_unrecognized_feature): New function.
24467 (arm_parse_arch_cpu_name): Ignore any characters after the first
24469 (arm_parse_arch_cpu_feature): New function.
24470 (arm_configure_build_target): Separate out any CPU and architecture
24471 features and parse separately. Don't error out if -mfpu=auto is
24472 used with only an architecture string.
24473 (arm_print_asm_arch_directives): New function.
24474 (arm_file_start): Call it.
24475 * config/arm/arm-cpu-cdata.h: Regenerated.
24476 * config/arm/arm-cpu-data.h: Likewise.
24477 * config/arm/arm-tables.opt: Likewise.
24479 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24481 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24482 assembler when it is not -mfpu=auto.
24484 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24486 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24487 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24488 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24489 (ASM_CPU_SPEC): Rewrite.
24490 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24491 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24492 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24493 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24494 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24495 copied string is NUL-terminated. Also strip any characters prefixed
24497 (arm_rewrite_selected_arch): New function.
24498 (arm_rewrite_march): New function.
24500 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24502 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24503 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24504 (march, mcpu, mtune): Convert to string-based options.
24505 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24506 (arm_parse_arch_cpu_name): New function.
24507 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24508 identify selected architecture or CPU.
24509 (arm_option_save): New function.
24510 (TARGET_OPTION_SAVE): Redefine.
24511 (arm_option_restore): Restore string options.
24512 (arm_option_print): Print string options.
24514 2017-06-16 Martin Sebor <msebor@redhat.com>
24516 PR tree-optimization/80933
24517 PR tree-optimization/80934
24518 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24519 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24520 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24521 (gimple_fold_builtin): Call them.
24523 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24525 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24526 as unlikely; update profile.
24528 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24530 * predict.c (force_edge_cold): Handle declaring edges impossible
24533 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24535 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24537 (try_unroll_loop_completely): Fix reporting.
24539 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24541 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24543 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24546 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24547 if given a non-constant argument for an intrinsic which requires a
24550 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24552 * profile.c (compare_freqs): New function.
24553 (branch_prob): Sort edge list.
24554 (find_spanning_tree): Assume that the list is priority sorted.
24556 2017-06-16 Richard Biener <rguenther@suse.de>
24558 PR tree-optimization/81090
24559 * passes.def (pass_record_bounds): Remove.
24560 * tree-pass.h (make_pass_record_bounds): Likewise.
24561 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24562 make_pass_record_bounds): Likewise.
24563 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24564 not free niter estimates at the beginning but at the end.
24565 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24567 2017-06-16 Richard Biener <rguenther@suse.de>
24569 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24570 initializer to workaround ICE in host GCC 4.8.
24572 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24574 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24576 (clone_inlined_nodes): Update.
24578 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24580 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24581 prefetch settings, and enable prefetching by default at -O3.
24583 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24585 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24586 Set flag_prefetch_loop_arrays according to tuning data.
24588 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24590 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24591 New tune structure.
24592 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24593 [Unrelated to main purpose of the patch] Place the pointer field last
24594 to enable type checking errors when tune structure are wrongly merged.
24595 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24596 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24597 (thunderx2t99_prefetch_tune): New tune constants.
24598 (tune_params *_tunings): Update all tunings (no functional change).
24599 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24600 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24601 from tunings structures.
24603 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24606 * ubsan.c (instrument_null): Add T argument, use it instead
24607 of computing it based on IS_LHS.
24608 (instrument_object_size): Likewise.
24609 (pass_ubsan::execute): Adjust instrument_null and
24610 instrument_object_size callers to pass gimple_get_lhs or
24611 gimple_assign_rhs1 result to it. Use instrument_null instead of
24612 calling get_base_address and instrument_mem_ref. Handle
24613 aggregate call arguments for object-size sanitization.
24615 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24617 PR tree-optimization/81089
24618 * tree-vrp.c (is_masked_range_test): Validate operands of
24621 2017-06-15 Martin Sebor <msebor@redhat.com>
24624 * dumpfile.c (dump_register): Avoid calling memset to initialize
24625 a class with a default ctor.
24626 * gcc.c (struct compiler): Remove const qualification.
24627 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24628 * hash-table.h: Ditto.
24629 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24631 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24632 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24634 * params.h (struct param_info): Make struct members non-const.
24635 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24636 with default initialization.
24637 * vec.h (vec_copy_construct, vec_default_construct): New helper
24639 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24640 with vec_copy_construct.
24641 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24642 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24643 * doc/invoke.texi (-Wclass-memaccess): Document.
24645 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24647 * emit-rtl.h (is_leaf): Update comment about local
24648 register allocator.
24650 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24653 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24654 for a variable to have a section before checking if the section has a
24656 Set section to.persistent if persistent attribute is set.
24657 Warn if .persistent attribute is used on an automatic variable.
24659 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24661 PR rtl-optimization/80474
24662 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24664 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24666 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24669 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24671 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24672 (casesi): Emit bounds checking as RTL.
24673 (casesi_internal_mips16_<mode>): Remove bounds checking.
24675 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24677 * config/xtensa/xtensa.c (xtensa_option_override): Append
24678 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24679 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24680 xtensa_doloop_hooks): Define unconditionally.
24681 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24683 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24684 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24685 for it in xtensa_option_override.
24686 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24687 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24689 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24691 * doc/cppopts.texi: Document '-' special value to -MF.
24693 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24695 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24696 (cortex_a53_fconst): Likewise.
24697 (cortex_a53_fpmul): Likewise.
24698 (cortex_a53_f_load_64): Likewise.
24699 (cortex_a53_f_load_many): Likewise.
24700 (cortex_a53_advsimd_alu): Likewise.
24701 (cortex_a53_advsimd_alu_q): Likewise.
24702 (cortex_a53_advsimd_mul): Likewise.
24703 (cortex_a53_advsimd_mul_q): Likewise.
24704 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24705 Add missing fmul, r2f_cvt and fconst cases.
24707 2017-06-14 Richard Biener <rguenther@suse.de>
24709 PR middle-end/81088
24710 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24712 (fold_binary_loc): When associating do not treat pre-existing
24713 TREE_OVERFLOW on literal constants as a reason to allow
24714 TREE_OVERFLOW on associated literal constants.
24716 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24718 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24719 (MASK_FEATURES): New macro.
24720 * config/sparc/sparc.c (sparc_option_override): Remove the special
24721 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24723 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24725 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24726 a division of 0 if non-call exceptions are enabled.
24728 2017-06-14 Andrew Pinski <apinski@cavium.com>
24729 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24732 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24733 Improve vector initialization code gen for only variable case.
24735 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24737 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24739 2017-06-14 Richard Biener <rguenther@suse.de>
24741 PR tree-optimization/81083
24742 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24745 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24747 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24748 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24749 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24750 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24751 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24752 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24754 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24756 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24757 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24759 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24761 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24763 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24765 * config/rs6000/t-rtems: Don't handle SPE.
24767 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24769 * config/rs6000/t-linux: Don't handle SPE.
24771 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24773 * config/rs6000/eabispe.h: Delete file.
24775 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24777 * config/rs6000/t-spe: Delete file.
24779 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24781 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24782 (rs6000_legitimate_offset_address_p): Return false for anything in
24783 V2SImode or V2SFmode.
24785 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24787 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24788 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24789 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24791 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24792 (rs6000_legitimate_offset_address_p): Ditto.
24793 (rs6000_emit_move): Ditto.
24794 (rs6000_init_builtins): Remove V4HI_type_node.
24796 2017-06-13 Martin Liska <mliska@suse.cz>
24799 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24800 (gate_asan): Likewise.
24801 * asan.h (asan_no_sanitize_address_p): Remove the function.
24802 (sanitize_flags_p): New function.
24803 * builtins.def: Fix coding style.
24804 * common.opt: Use renamed enum value.
24805 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24806 * doc/extend.texi: Document no_sanitize attribute.
24807 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24808 to SANITIZE_UNDEFINED_NONDEFAULT.
24809 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24810 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24811 Use sanitize_flags_p.
24812 * gimplify.c (gimplify_function_tree): Likewise.
24813 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24814 * opts.c (parse_no_sanitize_attribute): New function.
24815 (common_handle_option): Use renamed enum value.
24816 * opts.h (parse_no_sanitize_attribute): Declare.
24817 * tree.c (sanitize_flags_p): New function.
24818 * tree.h: Declared here.
24819 * tsan.c: Use sanitize_flags_p.
24820 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24821 (instrument_mem_ref): Likewise.
24822 (instrument_bool_enum_load): Likewise.
24823 (do_ubsan_in_current_function): Remove the function.
24824 (pass_ubsan::execute): Use sanitize_flags_p.
24825 * ubsan.h: Remove do_ubsan_in_current_function
24826 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24827 (dump_function_to_file): Use it here.
24829 2017-06-13 Martin Jambor <mjambor@suse.cz>
24831 PR tree-optimization/80803
24832 PR tree-optimization/81063
24833 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24834 (propagate_subaccesses_across_link): Enqueue subtree whenever
24835 necessary instead of relying on the caller.
24837 2017-06-13 Martin Jambor <mjambor@suse.cz>
24839 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24840 that have a first_link.
24841 (sort_and_splice_var_accesses): Do not check first_link before
24843 (subtree_mark_written_and_enqueue): Likewise.
24844 (propagate_all_subaccesses): Likewise and do not stop at first
24845 parent with a first_link.
24847 2017-06-13 Martin Jambor <mjambor@suse.cz>
24849 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24852 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24854 * match.pd: New pattern.
24856 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24858 * tree-vrp.c (is_masked_range_test): New function.
24859 (register_edge_assert_for): Determine ranges for
24862 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24864 PR tree-optimization/67328
24865 * fold-const.c (maskable_range_p): New function.
24866 (build_range_check): Generate bittests if possible.
24868 2017-06-13 Martin Liska <mliska@suse.cz>
24870 * gimple-pretty-print.c (dump_probability): Add new argument.
24871 (dump_edge_probability): Dump both probability and count.
24872 (dump_gimple_label): Likewise.
24873 (dump_gimple_bb_header): Likewise.
24875 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24878 * config/avr/avr-devices.c: Fix indentation.
24879 * config/avr/gen-avr-mmcu-specs.c: Dito.
24881 2017-06-13 Richard Biener <rguenther@suse.de>
24883 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24884 instead get vector type from stmt_info.
24885 (vectorizable_reduction): Adjust. Remove dead code.
24887 2017-06-13 Richard Biener <rguenther@suse.de>
24889 PR middle-end/81065
24890 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24891 case of C * (x * C2 + C3).
24892 (fold_addr_of_array_ref_difference): Properly fold index difference.
24894 2017-06-12 David S. Miller <davem@davemloft.net>
24897 * config/sparc/sparc.md (return expander): Emit frame blockage if
24898 function uses alloca.
24900 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24902 * combine.c (make_field_assignment): Check len rather than the mode
24903 precision when calling force_to_mode.
24905 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24907 Support multilibs and devices that see flash in RAM address range.
24910 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24911 (avr_mcu_t) <flash_pm_offset>: New field.
24912 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24913 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24914 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24915 (AVR_TINY_PM_OFFSET): Remove macro.
24916 * config/avr/avr.opt (-mshort-calls): New option.
24917 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24918 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24919 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24920 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24921 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24922 instead of avr_arch->have_jmp_call.
24923 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24924 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24925 avr_arch->flash_pm_offset to define.
24926 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24927 new field flash_pm_offset. Add entry for avrxmega3.
24928 (avr_texinfo): Add entry for avrxmega3.
24929 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24930 attiny212, attiny214,
24931 attiny412, attiny414, attiny416, attiny417,
24932 attiny814, attiny816, attiny817,
24933 attiny1614, attiny1616, attiny1617,
24934 attiny3214, attiny3216, attiny3217.
24935 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24936 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24937 (avr_print_operand_address) [AVR_TINY]: Same.
24938 (avr_asm_init_sections) <readonly_data_section>: Only patch
24939 callback if avr_arch->flash_pm_offset = 0.
24940 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24941 for rodata if avr_arch->flash_pm_offset != 0.
24942 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24943 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24944 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24945 (m_options): Append opt_rcall.
24946 (m_dirnames): Append dir_rcall.
24947 * config/avr/t-multilib: Regenerate.
24949 * configure.ac [target=avr]: Check whether avrxmega3 default
24950 linker description file works as needed.
24951 * configure: Regenerate.
24952 * doc/avr-mmcu.texi: Regenerate.
24953 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24954 <__AVR_ARCH__>: Document avrxmega3 and 103.
24955 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24956 <__AVR_SHORT_CALLS__>: Document it.
24957 <__AVR_PM_BASE_ADDRESS__>: Document it.
24958 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24959 (AVR Variable Attributes) <progmem>: Document this is
24960 not needed for avrxmega3.
24961 (AVR Named Address Spaces) <__flash>: Dito.
24963 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24965 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24967 2017-06-12 Doug Rupp <rupp@adacore.com>
24969 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24970 Append vxworks-stdint.h to the tm_file list.
24971 * config/vxworks-stdint.h: New file.
24973 2017-06-12 Martin Liska <mliska@suse.cz>
24975 PR tree-optimization/81041
24976 * tree-profile.c (gimple_gen_ic_func_profiler):
24977 Create an extra BB in profile-generate
24978 (gimple_gen_time_profiler): Likewise.
24980 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24982 PR tree-optimization/81003
24983 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24984 (update_range_test): Use it instead of force_gimple_operand_gsi.
24986 2017-06-12 Richard Biener <rguenther@suse.de>
24988 PR tree-optimization/81053
24989 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24990 with backedge value not defined in loop. Simplify def stmt
24993 2017-06-11 Tom de Vries <tom@codesourcery.com>
24996 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24998 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24999 nvptx_cannot_force_const_mem.
25001 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25003 * opts.c (finish_options): Move test for flag_split_stack after
25004 it has been initialized.
25006 2017-06-11 Jason Merrill <jason@redhat.com>
25008 * tree.h (id_equal): New.
25009 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25010 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25011 instead of strcmp of IDENTIFIER_POINTER.
25013 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25015 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25016 (mark_all_inlined_calls_cdtor): Fix formating.
25017 (inline_transform): Rescale profile before inlining.
25019 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25021 * cgraph.h (cgraph_edge::clone): Update prototype.
25022 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25023 (cgraph_node::create_clone): Update.
25024 (cgraph_node::create_version_clone): Update.
25025 * tree-inline.c (copy_bb): Update.
25026 (expand_call_inline): Update.
25028 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25030 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25031 factored out from ...
25032 (rs6000_emit_prologue): ... here.
25034 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25036 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25037 factored out from ...
25038 (rs6000_emit_prologue): ... here.
25040 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25042 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25044 (handle_missing_profiles): Fix computation of tp_first_run.
25045 (counts_to_freqs): Do not touch freqs when count is 0.
25047 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25049 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25052 2017-06-10 Tom de Vries <tom@codesourcery.com>
25054 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25055 attributes): Document signal effective target.
25057 2017-06-10 Tom de Vries <tom@codesourcery.com>
25059 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25060 Document effective target stack_size.
25062 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25064 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25065 to the edit_context if they can be auto-applied.
25067 2017-06-9 Ian Lance Taylor <iant@golang.org>
25069 * opts.c (finish_options): If -fsplit-stack, disable implicit
25070 -forder-blocks-and-partition.
25071 * doc/invoke.texi (Optimize Options): Document that when using
25072 -fsplit-stack -forder-blocks-and-partition is not implicitly
25075 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25077 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25078 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25079 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25080 * builtins.def (abort, trap, unreachable): Declare cold.
25081 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25082 * tree-core.h (ECF_COLD): New.
25083 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25084 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25086 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25088 * predict.c (unlikely_executed_stmt_p): Cleanup.
25090 2017-06-09 Richard Biener <rguenther@suse.de>
25092 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25093 model if the ref is always written to.
25095 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25097 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25099 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25101 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25104 2017-06-09 Tom de Vries <tom@codesourcery.com>
25107 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25108 "target cannot support label values" when encountering LABEL_REF.
25110 2017-06-09 Martin Liska <mliska@suse.cz>
25112 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25113 (gimple_gen_ic_func_profiler): Emit direct comparison
25114 of __gcov_indirect_call_callee with NULL.
25115 (gimple_gen_time_profiler): Change probability from
25116 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25118 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25120 * profile.c (edge_gcov_counts): Turn to pointer.
25121 (compute_branch_probabilities, compute_branch_probabilities): Update.
25122 (branch_prob): Do not clear edge_gcov_count.
25123 * profile.h (edge_gcov_counts): Turn to pointer.
25124 (edge_gcov_count): Update.
25126 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25128 * gimple.h (gimple_check_failed): Mark cold.
25130 2017-06-09 Richard Biener <rguenther@suse.de>
25132 PR tree-optimization/66623
25133 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25134 refactor check_reduction into two parts, properly computing
25135 whether we have to check reduction validity for outer loop
25138 2017-06-09 Richard Biener <rguenther@suse.de>
25140 PR tree-optimization/79483
25141 * graphite-scop-detection.c (order): New global.
25142 (get_order): Compute bb to order mapping that satisfies code
25143 generation constraints.
25144 (cmp_pbbs): New helper.
25145 (build_scops): Start domwalk at entry block, sort generated
25148 2017-06-09 Richard Biener <rguenther@suse.de>
25150 PR middle-end/81007
25151 * ipa-polymorphic-call.c
25152 (ipa_polymorphic_call_context::restrict_to_inner_class):
25153 Skip FIELD_DECLs with error_mark_node type.
25154 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25157 2017-06-09 Martin Liska <mliska@suse.cz>
25159 * predict.c (struct branch_predictor): New struct.
25160 (test_prediction_value_range): New test.
25161 (predict_c_tests): New function.
25162 * selftest-run-tests.c (selftest::run_tests): Run the function.
25163 * selftest.h: Declare new tests.
25165 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25168 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25169 gen_add3_insn did not fail.
25170 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25171 r0, construct that number in a temporary reg and add that reg to r0.
25172 If asked to put the result in r0 as well, fail.
25174 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25176 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25177 for early expansion of vec_eqv.
25179 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25181 PR middle-end/81005
25182 * ubsan.c (instrument_null): Avoid pointless code temporary.
25183 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25185 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25189 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25192 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25195 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25197 * predict.c (unlikely_executed_edge_p): Move ahead.
25198 (probably_never_executed_edge_p): Use it.
25200 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25202 PR middle-end/79988
25203 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25204 gimple_call_builtin_p call.
25206 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25208 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25209 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25210 rtl_check_failed_type2, rtl_check_failed_code1,
25211 rtl_check_failed_code2, rtl_check_failed_code_mode,
25212 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25213 rtvec_check_failed_bounds, rtl_check_failed_flag,
25214 _fatal_insn_not_found, _fatal_insn): Likewise.
25215 * tree.h (tree_contains_struct_check_failed,
25216 tree_check_failed, tree_not_check_failed,
25217 tree_class_check_failed, tree_range_check_failed,
25218 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25219 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25220 tree_operand_check_failed, omp_clause_check_failed,
25221 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25224 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25226 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25227 flag_branch_probabilities.
25228 * ipa-inline.c (edge_badness): Likewise.
25229 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25230 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25231 * predict.c (maybe_hot_frequency_p): Likewise.
25232 (probably_never_executed): Likewise.
25233 * sched-ebb.c (schedule_ebbs): Likewise.
25234 * sched-rgn.c (find_single_block_region): Likewise.
25235 * tracer.c (tail_duplicate): Likewise.
25237 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25239 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25240 longer requires x_flag_profile_use.
25242 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25244 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25245 instead of flag_reorder_blocks_and_partition.
25246 * dbxout.c (dbxout_function_end): Likewise.
25247 * dwarf2out.c (gen_subprogram_die): Likewise.
25248 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25249 * hw-doloop.c (reorg_loops): Likewise.
25250 * varasm.c (assemble_start_function,
25251 assemble_end_function): Likewise.
25252 (decide_function_section): Do not check for
25253 flag_reorder_blocks_and_partition.
25255 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25257 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25259 (chkp_get_hard_register_fake_addr_expr): Ditto.
25260 (chkp_build_addr_expr): Add check for hard reg case.
25261 (chkp_parse_array_and_component_ref): Ditto.
25262 (chkp_find_bounds_1): Ditto.
25263 (chkp_process_stmt): Don't generate bounds store for
25266 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25268 * predict.c (maybe_hot_bb_p): Do not check profile status.
25269 (maybe_hot_edge_p): Likewise.
25270 (probably_never_executed): Check for zero counts even if profile
25272 (unlikely_executed_edge_p): New function.
25273 (unlikely_executed_stmt_p): New function.
25274 (unlikely_executed_bb_p): New function.
25275 (set_even_probabilities): Use unlikely predicates.
25276 (combine_predictions_for_bb): Likewise.
25277 (predict_paths_for_bb): Likewise.
25278 (predict_paths_leading_to_edge): Likewise.
25279 (determine_unlikely_bbs): New function.
25280 (estimate_bb_frequencies): Use it.
25281 (compute_function_frequency): Use zero counts even if profile is
25283 * profile-count.h: Fix typo.
25285 2017-08-08 Julia Koval <julia.koval@intel.com>
25287 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25288 _mm512_mask_cvtsepi16_storeu_epi8,
25289 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25290 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25291 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25292 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25293 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25294 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25295 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25296 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25297 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25298 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25299 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25300 __builtin_ia32_pmovuswb256mem_mask,
25301 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25302 __builtin_ia32_pmovwb512mem_mask): New builtins.
25304 2017-08-08 Julia Koval <julia.koval@intel.com>
25306 PR target/73350,80862
25307 * config/i386/subst.md (round): Fix round pattern.
25308 * config/i386/i386.c (ix86_erase_embedded_rounding):
25309 Fix erasing rounding for the fixed pattern.
25311 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25313 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25315 2017-06-08 Martin Liska <mliska@suse.cz>
25317 PR gcov-profile/80911
25318 * gcov.c (block_info::block_info): New constructor.
25320 2017-06-07 Carl Love <cel@us.ibm.com>
25322 * config/rs6000/rs6000-c: The return type of the following
25323 built-in functions was implemented as int not long long. Fix sign
25324 of return value for the unsigned version of vec_mulo and vec_mule.
25325 vector unsigned long long vec_bperm (vector unsigned long long,
25326 vector unsigned char)
25327 vector signed long long vec_mule (vector signed int,
25329 vector unsigned long long vec_mule (vector unsigned int,
25330 vector unsigned int)
25331 vector signed long long vec_mulo (vector signed int,
25333 vector unsigned long long vec_mulo (vector unsigned int,
25334 vector unsigned int)
25335 * doc/extend.texi: Fix the documentation for the built-in
25338 2017-06-07 Carl Love <cel@us.ibm.com>
25341 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25344 2017-06-07 Carl Love <cel@us.ibm.com>
25346 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25347 support, Generate doublehv for signed int/float for BE case only.
25349 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25351 * doc/invoke.texi (mcx16): Rewrite.
25353 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25355 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25356 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25357 *mov<mode>_softfloat, and an anonymous splitter): Use
25358 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25360 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25362 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25364 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25365 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25366 (rs6000_debug_reg_global): Adjust.
25367 (rs6000_init_hard_regno_mode_ok): Adjust.
25368 (rs6000_dbx_register_number): Adjust.
25369 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25370 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25371 Remove SPE_ACC and SPEFSCR.
25372 (REG_ALLOC_ORDER): Ditto.
25373 (FRAME_POINTER_REGNUM): Change to 111.
25374 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25375 (REG_CLASS_NAMES): Ditto.
25376 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25377 (REGISTER_NAMES): Ditto.
25378 (ADDITIONAL_REG_NAMES): Ditto.
25379 (rs6000_reg_names): Ditto.
25380 * config/rs6000/rs6000.md: Renumber some register number
25383 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25385 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25387 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25388 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25390 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25391 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25392 Delete the SPE high registers.
25393 (REG_ALLOC_ORDER): Ditto.
25394 (enum reg_class): Remove SPE_HIGH_REGS.
25395 (REG_CLASS_NAMES): Ditto.
25396 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25397 (REGISTER_NAMES): Ditto.
25398 (rs6000_reg_names): Ditto.
25399 * doc/tm.texi.in: Remove SPE as example.
25400 * doc/tm.texi: Regenerate.
25402 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25404 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25405 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25406 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25407 * config/rs6000/rs6000.md (type): Remove "brinc".
25409 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25411 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25412 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25413 * config/rs6000/linuxspe.h: Delete file.
25414 * config/rs6000/rs6000.md: Don't include spe.md.
25415 * config/rs6000/spe.h: Delete file.
25416 * config/rs6000/spe.md: Delete file.
25417 * config/rs6000/t-rs6000: Remove spe.md.
25419 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25421 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25422 (reg_or_none500mem_operand): Delete.
25423 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25424 instead of reg_or_none500mem_operand.
25426 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25428 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25429 handling of SPE flags.
25430 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25432 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25434 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25436 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25437 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25438 paired_divv2sf3): Similar.
25439 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25440 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25441 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25443 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25444 Rename the paired_* instruction patterns.
25445 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25447 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25448 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25449 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25450 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25451 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25452 PAIRED_VECTOR_MODE.
25453 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25454 (spe_func_has_64bit_regs_p): Delete.
25455 (spe_expand_predicate_builtin): Delete.
25456 (spe_expand_evsel_builtin): Delete.
25457 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25458 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25459 (invalid_e500_subreg): Delete.
25460 (rs6000_legitimize_address): Always force_reg op2 as well, for
25461 paired single memory accesses.
25462 (rs6000_member_type_forces_blk): Delete.
25463 (rs6000_spe_function_arg): Delete.
25464 (rs6000_expand_unop_builtin): Delete SPE handling.
25465 (rs6000_expand_binop_builtin): Ditto.
25466 (spe_expand_stv_builtin): Delete.
25467 (bdesc_2arg_spe): Delete.
25468 (spe_expand_builtin): Delete.
25469 (spe_expand_predicate_builtin): Delete.
25470 (spe_expand_evsel_builtin): Delete.
25471 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25472 (spe_init_builtins): Delete.
25473 (spe_func_has_64bit_regs_p): Delete.
25474 (savres_routine_name): Delete "info" parameter. Adjust callers.
25475 (rs6000_emit_stack_reset): Ditto.
25476 (rs6000_dwarf_register_span): Delete.
25477 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25478 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25479 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25481 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25483 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25484 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25485 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25486 mulv2sf3, divv2sf3): Delete expanders.
25488 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25490 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25492 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25494 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25495 * config/rs6000/rs6000.c: Ditto.
25497 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25499 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25500 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25501 comparison_operator.
25503 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25505 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25506 * config/rs6000/rs6000.opt: Ditto.
25507 * config/rs6000/t-rtems: Ditto.
25509 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25511 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25512 TARGET_E500_SINGLE by 0, simplify.
25513 * config/rs6000/rs6000.c: Ditto.
25514 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25515 (spe_build_register_parallel): Delete.
25516 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25517 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25518 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25519 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25520 (E500_CONVERT): Delete.
25521 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25523 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25525 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25526 * config/rs6000/dfp.md: Ditto.
25527 (negdd2, *negdd2_fpr): Merge.
25528 (absdd2, *absdd2_fpr): Merge.
25529 (negtd2, *negtd2_fpr): Merge.
25530 (abstd2, *abstd2_fpr): Merge.
25531 * config/rs6000/e500.h: Delete file.
25532 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25533 TARGET_FPRS by 1 and simplify.
25534 * config/rs6000/rs6000-c.c: Ditto.
25535 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25536 TARGET_DF_SPE by 0.
25537 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25539 * config/rs6000/rs6000.md: Ditto.
25540 (floatdidf2, *floatdidf2_fpr): Merge.
25541 (move_from_CR_gt_bit): Delete.
25542 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25543 (E500_CR_IOR_COMPARE): Delete.
25544 (All patterns that require !TARGET_FPRS): Delete.
25545 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25547 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25549 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25551 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25553 * graphds.c (add_edge): Intitialize edge's attached data.
25554 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25555 pointer parameter. Call pointed function on each edge during
25556 graph traversing. Skip traversing the edge when the function
25558 (graphds_dfs, graphds_scc): Ditto.
25559 (for_each_edge): New parameter. Pass the new parameter to callback
25561 * graphds.h (skip_edge_callback): New function pointer type.
25562 (graphds_dfs, graphds_scc): New function pointer parameter.
25563 (graphds_edge_callback, for_each_edge): New parameter.
25565 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25567 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25568 out code checking if runtime alias check is possible to below ...
25569 Call the new function.
25570 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25571 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25573 2017-06-07 Marek Polacek <polacek@redhat.com>
25576 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25577 TYPE_OVERFLOW_WRAPS check.
25579 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25581 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25582 if versioning is required.
25583 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25584 peeling with the check for versioning.
25586 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25588 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25589 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25590 Set true to new parameter if new ssa variable is defined.
25591 (vect_gen_vector_loop_niters): Refactor. Set range information
25592 for the new vector loop bound variable.
25593 (vect_do_peeling): Ditto.
25595 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25597 * tree-affine.c (ssa.h): Include header file.
25598 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25599 has wrapping overflow behavior.
25601 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25603 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25605 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25607 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25608 (tree_to_aff_combination): ... here.
25610 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25612 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25613 reg_pressure model function.
25614 (ivopts_global_cost_for_size): Delete.
25615 (determine_set_costs, iv_ca_recount_cost): Call new model function
25616 ivopts_estimate_reg_pressure.
25618 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25620 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25621 expensive than udiv. Remove floating point cases from mod.
25623 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25625 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25626 Increase idiv cost.
25628 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25630 * config/aarch64/aarch64.md
25631 (copysignsf3): Fix mask generation.
25633 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25635 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25637 (class dump_manager): Add register_dumps method.
25638 * dumpfile.c: Include langhooks.h.
25639 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25640 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25641 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25642 (dump_manager::dump_register): Start with 512 entries instead of 32.
25643 (dump_manager::register_dumps): New method.
25644 * toplev.c (general_init): Instead of invoking register_dumps
25645 langhook, invoke register_dumps method on the dump manager.
25646 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25649 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25651 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25652 Say that # requires an associated define_split to exist, and that
25653 the define_split must be suitable for use after register allocation.
25655 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25657 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25658 (compute_outgoing_frequencies): Also initialize zero counts.
25659 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25660 around loops; preserve more of profile when nothing changes.
25662 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25664 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25666 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25667 * config/arm/arm-cpu-cdata.h: Regenerate.
25668 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25669 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25670 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25671 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25672 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25674 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25676 * config/arm/t-rmprofile: Likewise.
25677 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25679 2017-06-06 David S. Miller <davem@davemloft.net>
25682 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25683 blockage if function uses alloca.
25685 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25687 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25688 New "uid" fields to hold pretty-print IDs of group and ref.
25689 Memory references are now identified as <group_id>:<ref_id>
25690 instead of using [random] addresses.
25691 (dump_mem_details): Simplify, no functional change.
25692 (dump_mem_ref): Simplify and make output more concise.
25693 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25694 (find_or_create_group): Initialize group uid.
25695 (record_ref): Initialize ref uid. Improve debug output.
25696 (prune_group_by_reuse, should_issue_prefetch_p,)
25697 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25698 (mark_nontemporal_store, determine_loop_nest_reuse):
25699 Improve debug output.
25701 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25703 * dbgcnt.def (prefetch): New debug counter.
25704 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25705 (schedule_prefetches): Stop issueing prefetches if debug counter
25708 2017-06-06 Tom de Vries <tom@codesourcery.com>
25710 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25711 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25713 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25715 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25716 Use aarch64_reg_or_zero predicate for operand 4.
25717 (aarch64_compare_and_swap<mode> define_insn_and_split):
25718 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25719 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25721 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25723 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25724 (arm_compute_save_core_reg_mask): This.
25725 (thumb1_compute_save_reg_mask): Rename into ...
25726 (thumb1_compute_save_core_reg_mask): This.
25727 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25728 (arm_compute_frame_layout): Likewise.
25730 2017-06-06 Richard Biener <rguenther@suse.de>
25732 PR tree-optimization/80974
25733 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25734 keep or clear leaders SSA info.
25736 2017-06-06 Tom de Vries <tom@codesourcery.com>
25738 * config/nvptx/nvptx.c (split_mode_p): New function.
25739 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25741 2017-06-06 Tom de Vries <tom@codesourcery.com>
25743 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25745 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25748 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25751 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25753 * shrink-wrap.c (handle_simple_exit): Update profile.
25754 (try_shrink_wrapping): Upate profile.
25756 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25758 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25759 (tree_guess_outgoing_edge_probabilities): New.
25760 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25761 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25763 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25765 * ipa-split.c (split_function): Initialize return bb profile.
25767 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25769 * profile.c (compute_branch_probabilities): Also initialize
25770 EXIT_BLOCK profile.
25772 2017-06-06 Richard Biener <rguenther@suse.de>
25774 PR tree-optimization/80928
25775 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25776 (vect_analyze_loop_operations): Properly guard analysis for
25778 (vect_transform_loop): Likewise.
25779 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25780 (vect_model_induction_cost): Do not cost for pure SLP.
25781 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25782 of induction in inner loop vectorization.
25783 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25784 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25785 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25787 (vect_analyze_slp_cost_1): Cost induction.
25788 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25789 (vect_get_slp_vect_defs): Likewise.
25790 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25791 (vect_transform_stmt): Handle SLP reductions.
25792 * tree-vectorizer.h (vectorizable_induction): Adjust.
25794 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25796 * config/rs6000/rs6000.c (make_resolver_func): Update
25797 init_lowered_empty_function call.
25799 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25801 * doc/invoke.texi: Document the -fprofile-abs-path option.
25802 * common.opt (fprofile-abs-path): New option.
25803 * gcov-io.h (gcov_write_filename): Declare.
25804 * gcov-io.c (gcov_write_filename): New function.
25805 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25806 * profile.c (output_location): Likewise.
25808 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25810 * shring-wrap.c: Revert accidental commit.
25812 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25814 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25816 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25818 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25820 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25821 profile in callgraph edge.
25822 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25823 (apply_scale): Likewise.
25824 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25825 Also scale profile when inlining function with zero profile.
25826 (initialize_cfun): Update exit block profile even when it is zero.
25827 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25828 when profile is read.
25830 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25832 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25833 (CLONE_*): New constants to define the processors we can generate
25834 code for with the target_clone attribute.
25835 (rs6000_clone_map): New array to identify which clone processors
25836 the current program is running on.
25837 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25838 target_clone attribute.
25839 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25840 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25841 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25842 (cpu_expand_builtin): Add support for target_clone attribute.
25843 (rs6000_valid_attribute_p): Allow "default" attribute.
25844 (get_decl_name): New debug function to simplify printing the
25845 current function name in debugging statements.
25846 (rs6000_clone_priority): New functions to support the target_clone
25847 attribute, and be able to generate code to switch between ISA 2.05
25848 through ISA 3.0 (power6 through power9).
25849 (rs6000_compare_version_priority): Likewise.
25850 (rs6000_get_function_versions_dispatcher): Likewise.
25851 (make_resolver_func): Likewise.
25852 (add_condition_to_bb): Likewise.
25853 (dispatch_function_versions): Likewise.
25854 (rs6000_generate_version_dispatcher_body): Likewise.
25855 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25856 (fusion_gpr_load_p): Fix a spacing issue.
25857 * doc/extend.texi (Common Function Attributes): Document that the
25858 PowerPC supports the target_clone attribute.
25860 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25862 * config/arm/arm.h: explain F symbol found in description of ARM
25863 register allocation in its legend.
25865 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25867 * config/mips/frame-header-opt.c: Include profile-count.h.
25868 * config/riscv/riscv.c: Include profile-count.h
25870 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25872 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25874 (sm_set_flag_if_changed): Add bbs field.
25875 (execute_sm_if_changed_flag_set): Pass BBS.
25876 (execute_sm): Update.
25878 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25880 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25883 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25885 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25886 (peephole2): New peephole2 to emit the above.
25887 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25889 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25891 * config/aarch64/aarch64.c (define_peephole2 above
25892 *sub_<shift>_<mode>): New peephole.
25894 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25896 * config/i386/i386.c (make_resolver_func): Update.
25897 * Makefile.in: Add profile-count.h and profile-count.o
25898 * auto-profile.c (afdo_indirect_call): Update to new API.
25899 (afdo_set_bb_count): Update.
25900 (afdo_propagate_edge): Update.
25901 (afdo_propagate_circuit): Update.
25902 (afdo_calculate_branch_prob): Update.
25903 (afdo_annotate_cfg): Update.
25904 * basic-block.h: Include profile-count.h
25905 (struct edge_def): Turn count to profile_count.
25906 (struct basic_block_def): Likewie.
25907 (REG_BR_PROB_BASE): Move to profile-count.h
25908 (RDIV): Move to profile-count.h
25909 * bb-reorder.c (max_entry_count): Turn to profile_count.
25910 (find_traces): Update.
25911 (rotate_loop):Update.
25912 (connect_traces):Update.
25913 (sanitize_hot_paths):Update.
25914 * bt-load.c (migrate_btr_defs): Update.
25915 * cfg.c (RDIV): Remove.
25916 (init_flow): Use alloc_block.
25917 (alloc_block): Uninitialize count.
25918 (unchecked_make_edge): Uninitialize count.
25919 (check_bb_profile): Update.
25920 (dump_edge_info): Update.
25921 (dump_bb_info): Update.
25922 (update_bb_profile_for_threading): Update.
25923 (scale_bbs_frequencies_int): Update.
25924 (scale_bbs_frequencies_gcov_type): Update.
25925 (scale_bbs_frequencies_profile_count): New.
25926 * cfg.h (update_bb_profile_for_threading): Update.
25927 (scale_bbs_frequencies_profile_count): Declare.
25928 * cfgbuild.c (compute_outgoing_frequencies): Update.
25929 (find_many_sub_basic_blocks): Update.
25930 * cfgcleanup.c (try_forward_edges): Update.
25931 (try_crossjump_to_edge): Update.
25932 * cfgexpand.c (expand_gimple_tailcall): Update.
25933 (construct_exit_block): Update.
25934 * cfghooks.c (verify_flow_info): Update.
25935 (dump_bb_for_graph): Update.
25936 (split_edge): Update.
25937 (make_forwarder_block): Update.
25938 (duplicate_block): Update.
25939 (account_profile_record): Update.
25940 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25941 (get_estimated_loop_iterations): Update.
25942 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25943 (single_likely_exit): Update.
25944 * cfgloopmanip.c (scale_loop_profile): Update.
25946 (set_zero_probability): Update.
25947 (lv_adjust_loop_entry_edge): Update.
25948 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25949 (purge_dead_edges): Update.
25950 (rtl_account_profile_record): Update.
25951 * cgraph.c (cgraph_node::create): Uninitialize count.
25952 (symbol_table::create_edge): Uninitialize count.
25953 (cgraph_update_edges_for_call_stmt_node): Update.
25954 (cgraph_edge::dump_edge_flags): Update.
25955 (cgraph_node::dump): Update.
25956 (cgraph_edge::maybe_hot_p): Update.
25957 * cgraph.h: Include profile-count.h
25958 (create_clone), create_edge, create_indirect_edge): Update.
25959 (cgraph_node): Turn count to profile_count.
25960 (cgraph_edge0: Likewise.
25961 (make_speculative, clone): Update.
25962 (create_edge): Update.
25963 (init_lowered_empty_function): Update.
25964 * cgraphclones.c (cgraph_edge::clone): Update.
25965 (duplicate_thunk_for_node): Update.
25966 (cgraph_node::create_clone): Update.
25967 * cgraphunit.c (cgraph_node::analyze): Update.
25968 (cgraph_node::expand_thunk): Update.
25969 * final.c (dump_basic_block_info): Update.
25970 * gimple-streamer-in.c (input_bb): Update.
25971 * gimple-streamer-out.c (output_bb): Update.
25972 * graphite.c (print_global_statistics): Update.
25973 (print_graphite_scop_statistics): Update.
25974 * hsa-brig.c: Include basic-block.h.
25975 * hsa-dump.c: Include basic-block.h.
25976 * hsa-gen.c (T sum_slice): Update.
25977 (convert_switch_statements):Update.
25978 * hsa-regalloc.c: Include basic-block.h.
25979 * ipa-chkp.c (chkp_produce_thunks): Update.
25980 * ipa-cp.c (struct caller_statistics): Update.
25981 (init_caller_stats): Update.
25982 (gather_caller_stats): Update.
25983 (ipcp_cloning_candidate_p): Update.
25984 (good_cloning_opportunity_p): Update.
25985 (get_info_about_necessary_edges): Update.
25986 (dump_profile_updates): Update.
25987 (update_profiling_info): Update.
25988 (update_specialized_profile): Update.
25989 (perhaps_add_new_callers): Update.
25990 (decide_about_value): Update.
25991 (ipa_cp_c_finalize): Update.
25992 * ipa-devirt.c (struct odr_type_warn_count): Update.
25993 (struct decl_warn_count): Update.
25994 (struct final_warning_record): Update.
25995 (possible_polymorphic_call_targets): Update.
25996 (ipa_devirt): Update.
25997 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25998 * ipa-icf.c (sem_function::merge): Update.
25999 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26000 * ipa-inline.c (compute_uninlined_call_time): Update.
26001 (compute_inlined_call_time): Update.
26002 (want_inline_small_function_p): Update.
26003 (want_inline_self_recursive_call_p): Update.
26004 (edge_badness): Update.
26005 (lookup_recursive_calls): Update.
26006 (recursive_inlining): Update.
26007 (inline_small_functions): Update.
26008 (dump_overall_stats): Update.
26009 (dump_inline_stats): Update.
26010 * ipa-profile.c (ipa_profile_generate_summary): Update.
26011 (ipa_propagate_frequency): Update.
26012 (ipa_profile): Update.
26013 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26014 * ipa-utils.c (ipa_merge_profiles): Update.
26015 * loop-doloop.c (doloop_modify): Update.
26016 * loop-unroll.c (report_unroll): Update.
26017 (unroll_loop_runtime_iterations): Update.
26018 * lto-cgraph.c (lto_output_edge): Update.
26019 (lto_output_node): Update.
26020 (input_node): Update.
26021 (input_edge): Update.
26022 (merge_profile_summaries): Update.
26023 * lto-streamer-in.c (input_cfg): Update.
26024 * lto-streamer-out.c (output_cfg): Update.
26025 * mcf.c (create_fixup_graph): Update.
26026 (adjust_cfg_counts): Update.
26027 (sum_edge_counts): Update.
26028 * modulo-sched.c (sms_schedule): Update.
26029 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26030 * predict.c (maybe_hot_count_p): Update.
26031 (probably_never_executed): Update.
26032 (dump_prediction): Update.
26033 (combine_predictions_for_bb): Update.
26034 (propagate_freq): Update.
26035 (handle_missing_profiles): Update.
26036 (counts_to_freqs): Update.
26037 (rebuild_frequencies): Update.
26038 (force_edge_cold): Update.
26039 * predict.h: Include profile-count.h
26040 (maybe_hot_count_p, counts_to_freqs): UPdate.
26041 * print-rtl-function.c: Do not include cfg.h
26042 * print-rtl.c: Include basic-block.h
26043 * profile-count.c: New file.
26044 * profile-count.h: New file.
26045 * profile.c (is_edge_inconsistent): Update.
26046 (correct_negative_edge_counts): Update.
26047 (is_inconsistent): Update.
26048 (set_bb_counts): Update.
26049 (read_profile_edge_counts): Update.
26050 (compute_frequency_overlap): Update.
26051 (compute_branch_probabilities): Update; Initialize and deinitialize
26053 (branch_prob): Update.
26054 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26055 (edge_gcov_count): New.
26056 (bb_gcov_count): New.
26057 * shrink-wrap.c (try_shrink_wrapping): Update.
26058 * tracer.c (better_p): Update.
26059 * trans-mem.c (expand_transaction): Update.
26060 (ipa_tm_insert_irr_call): Update.
26061 (ipa_tm_insert_gettmclone_call): Update.
26062 * tree-call-cdce.c: Update.
26063 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26064 (gimple_duplicate_sese_tail): Update.
26065 (gimple_account_profile_record): Update.
26066 (execute_fixup_cfg): Update.
26067 * tree-inline.c (copy_bb): Update.
26068 (copy_edges_for_bb): Update.
26069 (initialize_cfun): Update.
26070 (freqs_to_counts): Update.
26071 (copy_cfg_body): Update.
26072 (expand_call_inline): Update.
26073 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26074 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26075 (try_unroll_loop_completely): Update.
26076 (try_peel_loop): Update.
26077 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26078 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26079 * tree-ssa-loop-split.c (connect_loops): Update.
26080 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26081 * tree-ssa-reassoc.c (branch_fixup): Update.
26082 * tree-ssa-tail-merge.c (replace_block_by): Update.
26083 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26084 (compute_path_counts): Update.
26085 (update_profile): Update.
26086 (recompute_probabilities): Update.
26087 (update_joiner_offpath_counts): Update.
26088 (estimated_freqs_path): Update.
26089 (freqs_to_counts_path): Update.
26090 (clear_counts_path): Update.
26091 (ssa_fix_duplicate_block_edges): Update.
26092 (duplicate_thread_path): Update.
26093 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26094 (struct switch_conv_info): Update.
26095 * tree-tailcall.c (decrease_profile): Update.
26096 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26097 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26098 * value-prof.c (check_counter): Update.
26099 (gimple_divmod_fixed_value): Update.
26100 (gimple_mod_pow2): Update.
26101 (gimple_mod_subtract): Update.
26102 (gimple_ic_transform): Update.
26103 (gimple_stringop_fixed_value): Update.
26104 * value-prof.h (gimple_ic): Update.
26106 2017-06-02 Carl Love <cel@us.ibm.com>
26108 * config/rs6000/rs6000-c: Add support for built-in functions
26109 vector double vec_doublee (vector signed int);
26110 vector double vec_doublee (vector unsigned int);
26111 vector double vec_doublee (vector float);
26112 vector double vec_doubleh (vector signed int);
26113 vector double vec_doubleh (vector unsigned int);
26114 vector double vec_doubleh (vector float);
26115 vector double vec_doublel (vector signed int);
26116 vector double vec_doublel (vector unsigned int);
26117 vector double vec_doublel (vector float);
26118 vector double vec_doubleo (vector signed int);
26119 vector double vec_doubleo (vector unsigned int);
26120 vector double vec_doubleo (vector float);.
26121 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26122 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26124 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26125 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26126 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26128 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26129 vec_doublel, vec_doubleh.
26130 * doc/extend.texi: Update the built-in documentation file for the
26131 new built-in functions.
26133 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26136 * ipa-inline-analysis.c (free_growth_caches): Set
26137 edge_removal_hook_holder to NULL after removing it.
26139 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26141 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26142 comparision with zero.
26144 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26145 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26146 for early expansion of vec_min and vec_max builtins.
26147 (builtin_function_type): Add min/max unsigned variants to those
26148 identified as having unsigned arguments.
26150 2017-06-02 Olivier Hainque <hainque@adacore.com>
26152 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26154 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26156 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26157 Use VALL_F16 iterator rather than VALL.
26159 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26161 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26162 Emit CBNZ inside loop when doing a strong exchange and comparing
26163 against zero. Generate the CC flags after the loop.
26165 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26167 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26168 (dl_section_ref): New.
26169 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26170 On AIX, append an expression to subtract the size of the
26171 section length to dl_section_ref.
26173 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26175 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26176 for early expansion of vector absolute builtins.
26178 2017-06-02 Richard Biener <rguenther@suse.de>
26180 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26181 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26183 2017-06-02 Richard Biener <rguenther@suse.de>
26185 PR tree-optimization/80948
26186 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26187 stmt order as well.
26189 2017-06-02 Richard Biener <rguenther@suse.de>
26191 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26193 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26194 for inductions as relevant.
26196 2017-06-02 Richard Biener <rguenther@suse.de>
26198 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26199 (vectorizable_induction): ... this. Remove dead code.
26201 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26203 * builtins. (expand_builtin_alloca): Remove second parameter and
26204 infer its value from the first parameter instead.
26205 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26207 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26209 PR rtl-optimization/80903
26210 * loop-doloop.c (add_test): Unshare sequence.
26212 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26214 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26216 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26218 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26220 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26221 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26222 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26223 (xlogue_layout::get_stub_rtx): Make static.
26224 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26225 (xlogue_layout::compute_stub_managed_regs): Rename to...
26226 (xlogue_layout::count_stub_managed_regs): ...this.
26227 (xlogue_layout::is_stub_managed_reg): New function.
26228 (xlogue_layout::m_stub_names): Rename to...
26229 (xlogue_layout::s_stub_names): ...this, make static.
26230 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26231 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26232 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26233 xlogue_layout::s_stub_names): Instantiate statics.
26234 (stub_managed_regs): Remove.
26235 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26236 (disable_call_ms2sysv_xlogues): Rename to...
26237 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26238 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26240 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26241 change after reload_completed.
26242 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26244 (ix86_expand_prologue): Likewise.
26245 (ix86_expand_epilogue): Likewise.
26246 (ix86_expand_split_stack_prologue): Likewise.
26247 (ix86_compute_frame_layout): Remove frame parameter ...
26248 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26249 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26251 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26252 (ix86_frame): Move from here ...
26253 * config/i386/i386.h (ix86_frame): ... to here.
26254 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26255 complete ix86_frame data structure instead. Remove some_ld_name.
26257 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26259 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26260 symbols that hold a DECL_VALUE_EXPR.
26262 2017-06-01 Martin Jambor <mjambor@suse.cz>
26264 PR tree-optimization/80898
26265 * tree-sra.c (process_subtree_disqualification): Removed.
26266 (disqualify_candidate): Do not acll
26267 process_subtree_disqualification.
26268 (subtree_mark_written_and_enqueue): New function.
26269 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26270 RHS has been disqualified and re-queue LHS if necessary. Apart
26271 from that, ignore disqualified RHS.
26273 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26275 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26276 address fetch for z10 or later.
26278 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26280 * config/arc/arc.md (tst_movb): Add guard when splitting.
26282 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26284 * config/arc/arc.c (arc_can_eliminate): Test against
26285 arc_frame_pointer_needed.
26287 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26289 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26290 to prevent store reordering.
26291 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26292 (type): Add block type.
26293 (stack_tie): Define special instruction to be used in
26296 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26298 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26299 constraint. It is not valid for the pattern.
26300 (noncommutative_binary_comparison): Likewise.
26302 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26304 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26307 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26309 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26310 be used by the reg-alloc.
26312 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26314 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26315 reg-alloc when having mul64 or mul32x16 instructions.
26316 (mulsidi3): Likewise.
26317 (umulsidi3): Likewise.
26318 (mulsi32x16): New pattern.
26319 (mulsi64): Likewise.
26320 (mulsidi64): Likewise.
26321 (umulsidi64): Likewise.
26322 (MUL32x16_REG): Define.
26323 (mul64_600): Use MUL32x16_REG.
26324 (mac64_600): Likewise.
26325 (umul64_600): Likewise.
26326 (umac64_600): Likewise.
26328 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26330 * config/arc/arc.md (mulsi3_700): Make it commutative.
26332 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26334 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26336 (*sign_extendsidi2_insn): Likewise for movstosw.
26338 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26340 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26341 the type of the input discriminant value. Convert the
26342 discriminant value of signedness vary.
26344 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26346 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26347 Add to -Wall section.
26349 2017-06-01 Richard Biener <rguenther@suse.de>
26351 PR middle-end/66313
26352 * fold-const.c (fold_plusminus_mult_expr): If the factored
26353 factor may be zero use a wrapping type for the inner operation.
26354 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26355 and handle moved defs.
26356 (process_assignment): Properly guard the unary op case. Return a
26357 tri-state indicating that moving the stmt before the call may allow
26358 to continue. Pass through to_move.
26359 (find_tail_calls): Handle moving unrelated defs before
26362 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26365 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26366 splitter result in the canonical way.
26368 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26370 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26371 also for 32bit target. Update insn attributes.
26372 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26374 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26376 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26377 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26378 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26379 (_mm_maskz_min_ss): New intrinsics.
26381 2017-05-31 Martin Liska <mliska@suse.cz>
26383 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26384 Change comment style to one we normally use.
26385 (vectorizable_reduction): Likewise.
26386 (vectorizable_induction): Likewise.
26387 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26388 (vectorizable_call): Likewise.
26389 (vectorizable_simd_clone_call): Likewise.
26390 (vectorizable_conversion): Likewise.
26391 (vectorizable_assignment): Likewise.
26392 (vectorizable_shift): Likewise.
26393 (vectorizable_operation): Likewise.
26394 (vectorizable_store): Likewise.
26395 (vectorizable_load): Likewise.
26396 * tree-vectorizer.h: Likewise.
26398 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26400 * passes.c (emergency_dump_function): New.
26401 * tree-pass.h (emergency_dump_function): Declare.
26402 * plugin.c (plugins_internal_error_function): Remove.
26403 * plugin.h (plugins_internal_error_function): Remove declaration.
26404 * toplev.c (internal_error_function): New static function. Use it...
26405 (general_init): ...here.
26407 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26409 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26410 (arc_rtx_costs): Add costs for new patterns.
26411 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26412 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26414 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26416 * tree-ssa-strlen.c (get_next_strinfo): New function.
26417 (get_stridx_plus_constant): Use it.
26418 (zero_length_string): Likewise.
26419 (adjust_related_strinfos): Likewise.
26420 (adjust_last_stmt): Likewise.
26422 2017-05-31 Richard Biener <rguenther@suse.de>
26425 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26426 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26428 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26430 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26431 loop_vinfo argument and use of dependence distance vectors.
26432 Check instead whether the two references differ only in their
26433 initial value and assume that they have the same alignment if the
26434 difference is a multiple of the vector alignment.
26435 (vect_analyze_data_refs_alignment): Update call accordingly.
26437 2017-05-31 Martin Liska <mliska@suse.cz>
26440 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26442 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26444 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26445 (create_intersect_range_checks): Move from ...
26446 * tree-data-ref.c (create_intersect_range_checks_index)
26447 (create_intersect_range_checks): ... to here.
26448 (create_runtime_alias_checks): New function factored from ...
26449 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26450 here. Call above function.
26451 * tree-data-ref.h (create_runtime_alias_checks): New function.
26453 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26455 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26456 segment length for dr_b and compute it in wide_int.
26458 2017-05-31 Richard Biener <rguenther@suse.de>
26460 PR tree-optimization/80906
26461 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26462 and pass through iv_map.
26463 (copy_bb_and_scalar_dependences): Adjust.
26464 (translate_pending_phi_nodes): Likewise.
26465 (copy_loop_close_phi_args): Handle code-generating IVs instead
26468 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26470 * diagnostic-color.c (color_dict): Add "type-diff".
26471 (parse_gcc_colors): Update comment.
26472 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26473 -fdiagnostics-show-template-tree and -fno-elide-type.
26474 (GCC_COLORS): Add type-diff to example.
26476 (-fdiagnostics-show-template-tree): New.
26477 (-fno-elide-type): New.
26478 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26479 the pp_format_decoder callback. Call any m_format_postprocessor's
26481 (pretty_printer::pretty_printer): Initialize
26482 m_format_postprocessor.
26483 (pretty_printer::~pretty_printer): Delete any
26484 m_format_postprocessor.
26485 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26486 (class format_postprocessor): New class.
26487 (struct pretty_printer::format_decoder): Document the new parameters.
26488 (struct pretty_printer::m_format_postprocessor): New field.
26489 * tree-diagnostic.c (default_tree_printer): Update for new
26490 bool and const char ** params.
26491 * tree-diagnostic.h (default_tree_printer): Likewise.
26493 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26495 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26496 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26497 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26498 rs6000_gen_cell_microcode code.
26499 (rs6000_final_prescan_insn): Delete.
26500 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26501 "warn-cell-microcode" entries.
26502 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26503 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26504 throughout. Change cc_reg_not_micro_cr0_operand to
26505 cc_reg_not_cr0_operand throughout.
26506 (*extendhi<mode>2_noload): Delete.
26507 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26508 (mwarn-cell-microcode): Delete.
26509 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26510 -mgen-cell-microcode and -mwarn-cell-microcode.
26512 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26515 * config/i386/constraints.md (Yd): New constraint.
26517 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26518 and (?Yd, r) alternatives. Update insn attributes.
26519 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26520 and (?*Yd, r) alternatives. Update insn attributes.
26521 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26523 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26525 * gimplify.c (gimplify_modify_expr): Don't create a
26526 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26529 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26531 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26533 2017-05-30 Richard Biener <rguenther@suse.de>
26535 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26536 and reduc_def fields.
26537 (STMT_VINFO_REDUC_TYPE): New define.
26538 (STMT_VINFO_REDUC_DEF): Likewise.
26539 (vect_force_simple_reduction): Adjust prototype.
26540 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26541 (vect_is_simple_reduction): Remove check_reduction argument.
26542 (vect_force_simple_reduction): Adjust and set
26543 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26544 (vectorizable_reduction): Do not re-do reduction analysis
26545 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26546 * tree-parloops.c (gather_scalar_reductions): Adjust.
26548 2017-05-30 Richard Biener <rguenther@suse.de>
26550 PR middle-end/80901
26551 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26554 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26556 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26557 Introduce unknown_misalignment parameter and remove vf.
26558 (vect_peeling_hash_get_lowest_cost):
26559 Pass unknown_misalignment parameter.
26560 (vect_enhance_data_refs_alignment):
26561 Fix unsupportable data ref treatment.
26563 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26565 * tree-vect-data-refs.c (vect_get_data_access_cost):
26566 Workaround for SLP handling.
26567 (vect_enhance_data_refs_alignment):
26568 Compute costs for doing no peeling at all, compare to the best
26569 peeling costs so far and avoid peeling if cheaper.
26571 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26573 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26574 Return peeling info and set costs to zero for unlimited cost
26576 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26577 with unknown misalignment. Compute and costs for unknown
26578 misalignment, compare them to the costs for known misalignment
26579 and choose the cheapest for peeling.
26581 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26583 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26584 (vect_get_peeling_costs_all_drs): Create function.
26585 (vect_peeling_hash_get_lowest_cost):
26586 Use vect_get_peeling_costs_all_drs.
26587 (vect_peeling_supportable): Create function.
26588 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26590 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26592 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26593 DR_HAS_NEGATIVE_STEP.
26594 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26595 (vect_enhance_data_refs_alignment): Use.
26596 (vect_duplicate_ssa_name_ptr_info): Use.
26597 * tree-vectorizer.h (dr_misalignment): Use.
26598 (known_alignment_for_access_p): Use.
26600 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26603 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26605 (has_section_name): New function.
26607 2017-05-30 Martin Liska <mliska@suse.cz>
26610 * auto-profile.c (get_function_decl_from_block): Fix
26613 2017-05-30 Richard Biener <rguenther@suse.de>
26615 PR middle-end/80876
26616 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26618 2017-05-30 Martin Liska <mliska@suse.cz>
26620 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26621 * dumpfile.h (struct dump_file_info): Remove ctors.
26623 2017-05-30 Martin Liska <mliska@suse.cz>
26625 * predict.def: Fix GNU coding style.
26627 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26629 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26630 Mark 'to' argument with ATTRIBUTE_UNUSED.
26632 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26634 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26635 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26636 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26639 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26641 * doc/install.texi (Options specification): Restore entry of
26642 --enable-sjlj-exceptions.
26644 2017-05-27 Michael Eager <eager@eagercon.com>
26647 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26649 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26651 * config/microblaze/microblaze.h
26652 (FIXED_REGISTERS): Update in macro.
26653 (CALL_USED_REGISTERS): Update in macro.
26655 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26657 * doc/install.texi: Add links to macOS binary distributions.
26659 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26663 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26665 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26667 2017-05-26 Martin Liska <mliska@suse.cz>
26669 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26671 2017-05-26 Martin Liska <mliska@suse.cz>
26673 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26674 always leading ';; '.
26675 (dump_bb_info): Likewise.
26676 (brief_dump_cfg): Likewise.
26677 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26678 * dumpfile.c: Remove usage of TDF_VERBOSE.
26679 * dumpfile.h (enum dump_kind): Likewise.
26680 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26681 * print-tree.c (debug_verbose): Remove.
26682 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26683 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26684 * tree-diagnostic.c (default_tree_printer): Replace
26685 TDF_DIAGNOSTIC with TDF_SLIM.
26687 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26689 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26690 in parameter loop, rather than loop_vinfo.
26691 (create_intersect_range_checks): Ditto.
26692 (vect_create_cond_for_alias_checks): Update call to above functions.
26694 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26696 PR tree-optimization/80815
26697 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26698 for merging runtime alias checks. Handle negative DR_STEPs.
26700 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26702 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26704 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26705 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26706 out code pruning runtime alias checks.
26707 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26708 factored out from above.
26709 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26711 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26713 (prune_runtime_alias_test_list): New decalaration.
26715 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26717 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26718 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26719 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26720 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26721 (operator==, comp_dr_with_seg_len_pair): Ditto.
26722 (vect_prune_runtime_alias_test_list): Ditto.
26724 2017-05-26 Martin Liska <mliska@suse.cz>
26727 * params.def: Bound partial-inlining-entry-probability param.
26729 2017-05-26 Marek Polacek <polacek@redhat.com>
26732 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26735 2017-05-26 Richard Biener <rguenther@suse.de>
26737 PR tree-optimization/80842
26738 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26741 2017-05-26 Richard Biener <rguenther@suse.de>
26743 PR tree-optimization/80844
26744 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26746 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26748 * doc/md.texi (Machine Constraints): Update x86 family
26749 machine constraints section to match 'config/i386/constraints.md'.
26751 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26753 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26755 2017-05-25 Nathan Sidwell <nathan@acm.org>
26757 * doc/invoke.texi (--enable-languages): Update documentation.
26759 2017-05-25 Martin Liska <mliska@suse.cz>
26761 * dumpfile.c: Add TDF_FOLDING.
26762 * dumpfile.h (enum dump_kind): Likewise.
26763 * genmatch.c (dt_simplify::gen_1): Use it.
26765 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26767 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26769 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26771 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26772 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26774 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26776 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26777 * match.pd (X == C): Rewrite it here.
26778 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26779 with_certain_nonzero_bits2): New predicates.
26780 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26782 2017-05-24 Nathan Sidwell <nathan@acm.org>
26784 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26787 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26790 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26792 * config/powerpcspe: New port. Files are copied from the rs6000
26793 port, with "rs6000" in filenames replaced by "powerpcspe".
26795 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26797 PR rtl-optimization/80754
26798 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26800 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26802 * config/sparc/sparc.md (length): Return the correct value for -mflat
26803 sibcalls to match output_sibcall.
26805 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26809 * config/rs6000/rs6000.c (struct machine_function): Add new field
26811 (rs6000_get_separate_components): Init that field, use it.
26812 (rs6000_components_for_bb): Use the field.
26814 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26816 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26818 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26820 PR middle-end/80823
26821 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26823 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26826 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26827 address against address_operand predicate.
26828 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26830 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26832 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26833 parameters passed indirectly.
26835 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26837 * config/i386/i386.md (*movdi_internal): Remove SSE4
26838 alternative 18 (?r, *v). Update insn attributes.
26839 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26840 Update insn attributes.
26841 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26842 Update insn attributes.
26843 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26844 alternative 1 (r, v). Remove isa attribute.
26845 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26846 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26847 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26849 2017-05-23 Tom de Vries <tom@codesourcery.com>
26851 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26854 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26856 * cgraphunit.c (symbol_table::process_new_functions): Update.
26857 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26858 (inline_generate_summary): Rename to ...
26859 (ipa_fn_summary_generate): ... this one.
26860 (inline_read_summary): Rename to ...
26861 (ipa_fn_summary_read): ... this one.
26862 (inline_write_summary): Rename to ...
26863 (ipa_fn_summary_write): ... this one.
26864 (inline_free_summary): Rename to ...
26865 (ipa_free_fn_summary): ... this one.
26866 (pass_data_local_fn_summary, pass_local_fn_summary,
26867 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26868 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26869 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26870 make_pass_ipa_fn_summary): New.
26871 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26872 inline_write_summary, inline_free_summary): Remove.
26873 (ipa_free_fn_summary) : New.
26874 * ipa-inline.c (ipa_inline): Update.
26875 (pass_ipa_inline): Do not generate summaries.
26876 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26878 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26879 and add pass_ipa_fn_summary.
26880 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26882 (make_pass_inline_parameters): Remove.
26884 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26886 * omp-low.c (struct omp_context): Remove "default_kind" member.
26889 * omp-offload.c (execute_oacc_device_lower): Remove the
26890 parallelism dimensions function attributes for unparallelized
26891 OpenACC kernels constructs.
26893 2017-05-23 Martin Liska <mliska@suse.cz>
26895 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26897 (cgraph_edge::make_speculative): Likewise.
26898 (cgraph_edge::resolve_speculation): Likewise.
26899 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26900 (cgraph_node::dump): Likewise.
26901 * cgraph.h: Likewise.
26902 * cgraphunit.c (analyze_functions): Likewise.
26903 (symbol_table::compile): Likewise.
26904 * ipa-cp.c (print_all_lattices): Likewise.
26905 (determine_versionability): Likewise.
26906 (initialize_node_lattices): Likewise.
26907 (ipcp_verify_propagated_values): Likewise.
26908 (estimate_local_effects): Likewise.
26909 (update_profiling_info): Likewise.
26910 (create_specialized_node): Likewise.
26911 (perhaps_add_new_callers): Likewise.
26912 (decide_about_value): Likewise.
26913 (decide_whether_version_node): Likewise.
26914 (identify_dead_nodes): Likewise.
26915 (ipcp_store_bits_results): Likewise.
26916 * ipa-devirt.c (dump_targets): Likewise.
26917 (ipa_devirt): Likewise.
26918 * ipa-icf.c (sem_item::dump): Likewise.
26919 (sem_function::equals): Likewise.
26920 (sem_variable::equals): Likewise.
26921 (sem_item_optimizer::read_section): Likewise.
26922 (sem_item_optimizer::execute): Likewise.
26923 (congruence_class::dump): Likewise.
26924 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26925 (dump_inline_summary): Likewise.
26926 (estimate_node_size_and_time): Likewise.
26927 (inline_analyze_function): Likewise.
26928 * ipa-inline-transform.c (inline_call): Likewise.
26929 * ipa-inline.c (report_inline_failed_reason): Likewise.
26930 (want_early_inline_function_p): Likewise.
26931 (edge_badness): Likewise.
26932 (update_edge_key): Likewise.
26933 (inline_small_functions): Likewise.
26934 * ipa-profile.c (ipa_profile): Likewise.
26935 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26936 (ipa_make_edge_direct_to_target): Likewise.
26937 (remove_described_reference): Likewise.
26938 (ipa_impossible_devirt_target): Likewise.
26939 (propagate_controlled_uses): Likewise.
26940 (ipa_print_node_params): Likewise.
26941 (ipcp_transform_function): Likewise.
26942 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26943 (propagate_pure_const): Likewise.
26944 * ipa-reference.c (generate_summary): Likewise.
26945 (read_write_all_from_decl): Likewise.
26946 (propagate): Likewise.
26947 (ipa_reference_read_optimization_summary): Likewise.
26948 * ipa-utils.c (ipa_merge_profiles): Likewise.
26949 * ipa.c (walk_polymorphic_call_targets): Likewise.
26950 (symbol_table::remove_unreachable_nodes): Likewise.
26951 (ipa_single_use): Likewise.
26952 * passes.c (execute_todo): Likewise.
26953 * predict.c (drop_profile): Likewise.
26954 * symtab.c (symtab_node::get_dump_name): New function.
26955 (symtab_node::dump_name): Likewise.
26956 (symtab_node::dump_asm_name): Likewise.
26957 (symtab_node::dump_references): Likewise.
26958 (symtab_node::dump_referring): Likewise.
26959 (symtab_node::dump_base): Likewise.
26960 (symtab_node::debug_symtab): Likewise.
26961 * tree-sra.c (convert_callers_for_node): Likewise.
26962 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26963 * value-prof.c (init_node_map): Likewise.
26965 2017-05-23 Martin Liska <mliska@suse.cz>
26967 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26968 and symtab_node::debug_symtab to symbol_table::debug.
26969 * cgraphunit.c (analyze_functions): Use the renamed function.
26970 (symbol_table::compile): Likewise.
26971 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26972 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26973 * passes.c (execute_todo): Likewise.
26974 * symtab.c (symbol_table::dump): New function.
26975 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26977 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26979 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26980 that nonconst implies exec.
26982 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26984 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26985 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26986 (inline_edge_summary_vec): Turn into ...
26987 (ipa_call_summaries): ... this one.
26988 (redirect_to_unreachable, edge_set_predicate,
26989 evaluate_properties_for_edge, inline_summary_alloc,
26990 reset_ipa_call_summary, reset_inline_summary,
26991 inline_summary_t::duplicate): Update.
26992 (inline_edge_duplication_hook): Turn to ...
26993 (ipa_call_summary_t::duplicate): ... this one.
26994 (inline_edge_removal_hook): Turn to ...
26995 (ipa_call_summary_t::remove): ... this one.
26996 (dump_inline_edge_summary): Turn to ...
26997 (dump_ipa_call_summary): ... this one.
26998 (estimate_function_body_sizes): Update.
26999 (inline_update_callee_summaries): Update.
27000 (remap_edge_change_prob): Update.
27001 (remap_edge_summaries): Update.
27002 (inline_merge_summary): Update.
27003 (do_estimate_edge_time): Update.
27004 (inline_generate_summary): Update.
27005 (inline_read_section): Update.
27006 (inline_read_summary): Update.
27007 (inline_free_summary): Update.
27008 * ipa-inline.c (can_inline_edge_p): Update.
27009 (compute_inlined_call_time): Update.
27010 (want_inline_small_function_p): Update.
27011 (edge_badness): Update.
27012 (early_inliner): Update.
27013 * ipa-inline.h (inline_edge_summary): Turn to ...
27014 (ipa_call_summary): ... this one.
27015 (ipa_call_summary_t): New class.
27016 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27017 (ipa_call_summaries): New.
27018 (inline_edge_summary): Remove.
27019 (estimate_edge_growth): Update.
27020 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27021 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27022 * ipa-split.c (execute_split_functions): Update.
27023 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27025 2017-05-23 Tom de Vries <tom@codesourcery.com>
27027 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27028 attributes): Document rdrand effective target.
27030 2017-05-23 Tom de Vries <tom@codesourcery.com>
27032 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27033 attributes): Sort alphabetically.
27035 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27037 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27039 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27042 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27043 V2DF/V2DI splat into two separate patterns, one that handles
27044 registers, and the other that only handles memory. Drop support
27045 for splatting from a GPR on ISA 2.07 and then splitting the
27046 splat into direct move and splat.
27047 (vsx_splat_<mode>_reg): Likewise.
27048 (vsx_splat_<mode>_mem): Likewise.
27050 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27052 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27054 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27056 PR middle-end/80809
27057 * omp-low.c (finish_taskreg_remap): New function.
27058 (finish_taskreg_scan): If unit size of ctx->record_type
27059 is non-constant, unshare the size expression and replace
27060 decls in it with possible outer var refs.
27062 PR middle-end/80809
27063 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27064 GOVD_SHARED rather than GOVD_PRIVATE with it.
27065 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27066 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27068 PR middle-end/80853
27069 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27070 as last argument to build_outer_var_ref for pointer bases of array
27071 section reductions.
27073 2017-05-19 Martin Sebor <msebor@redhat.com>
27075 * print-tree.c (print_node): Print DECL_READ_P flag.
27077 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27079 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27080 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27081 * cgraph.c: Likewise.
27082 * cgraphunit.c: Likewise.
27083 * gengtype.c: Likewise.
27084 * ipa-cp.c: Likewise.
27085 * ipa-devirt.c: Likewise.
27086 * ipa-icf.c: Likewise.
27087 * ipa-predicate.c: Likewise.
27088 * ipa-profile.c: Likewise.
27089 * ipa-prop.c: Likewise.
27090 * ipa-split.c: Likewise.
27092 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27093 edge_predicate_pool, dump_inline_hints,
27094 inline_summary::account_size_time, redirect_to_unreachable,
27095 edge_set_predicate, set_hint_predicate,
27096 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27097 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27098 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27099 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27100 ipa_call_summary_t::remove, initialize_growth_caches,
27101 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27102 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27103 mark_modified, unmodified_parm_1, unmodified_parm,
27104 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27105 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27106 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27107 will_be_nonconstant_predicate, record_modified_bb_info,
27108 get_minimal_bb, record_modified, param_change_prob,
27109 phi_result_unknown_predicate, predicate_for_phi_result,
27110 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27111 estimate_function_body_sizes, compute_inline_parameters,
27112 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27113 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27114 inline_update_callee_summaries, remap_edge_change_prob,
27115 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27116 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27117 inline_analyze_function, inline_summary_t::insert,
27118 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27119 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27120 inline_free_summary): Move to ipa-fnsummary.h
27121 (predicate_t): Remove.
27122 * ipa-fnsummary.c: New file.
27123 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27124 (enum inline_hints_vals, inline_hints, agg_position_info,
27125 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27126 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27127 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27128 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27129 inline_read_summary, inline_write_summary, inline_free_summary,
27130 inline_analyze_function, initialize_inline_failed,
27131 inline_merge_summary, inline_update_overall_summary,
27132 compute_inline_parameters): Move to ipa-fnsummary.h
27133 * ipa-fnsummary.h: New file.
27134 * ipa-inline-transform.h: Include ipa-inline.h.
27135 * ipa-inline.c: LIkewise.
27137 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27139 * ipa-inline.c (edge_badness): Use inlined_time instead of
27140 inline_summaries->get.
27142 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27144 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27146 2017-05-22 Nathan Sidwell <nathan@acm.org>
27148 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27149 (fdump-lang): Document 'raw' option.
27150 * dumpfile.h (TDI_tu): Delete.
27151 * dumpfile.c (dump_files): Remove translation-unit.
27152 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27154 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27156 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27157 command option from $(AWK) call.
27158 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27159 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27160 [FORMAT]: Remove handling of variable.
27161 * config/avr/t-multilib: Regenerate.
27163 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27165 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27167 (dump_inline_summary): Do not print self_time.
27168 (estimate_function_body_sizes): Do not set self_time.
27169 (compute_inline_parameters): Likewise.
27170 (inline_read_section, inline_write_summary): Do not stream self_time.
27171 * ipa-inline.h (inline_summary): Drop self_time.
27173 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27175 * ipa-inline-analysis.c (account_size_time): Rename to ...
27176 (inline_summary::account_size_time): ... this one.
27177 (reset_ipa_call_summary): Turn to ...
27178 (ipa_call_summary::reset): ... this one.
27179 (reset_inline_summary): Turn to ...
27180 (inline_summary::reset): ... this one.
27181 (inline_summary_t::remove): Update.
27182 (inline_summary_t::duplicate): Update.
27183 (ipa_call_summary_t::remove): Update.
27184 (dump_inline_summary): Update.
27185 (estimate_function_body_sizes): Update.
27186 (compute_inline_parameters): Update.
27187 (estimate_node_size_and_time): Update.
27188 (inline_merge_summary): Update.
27189 (inline_update_overall_summary): Update.
27190 (inline_read_section): Update.
27191 (inline_write_summary): Update.
27192 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27193 add account_size_time and reset member functions.
27194 (ipa_call_summary): Add reset function.
27195 * ipa-predicate.h (predicate::operator &): Constify.
27197 2017-05-22 Richard Biener <rguenther@suse.de>
27199 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27201 2017-05-19 Jason Merrill <jason@redhat.com>
27203 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27205 2017-05-19 Marek Polacek <polacek@redhat.com>
27208 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27209 TYPE_OVERFLOW_WRAPS checks.
27211 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27213 * tree-core.h (enum omp_clause_default_kind): Add
27214 "OMP_CLAUSE_DEFAULT_PRESENT".
27215 * tree-pretty-print.c (dump_omp_clause): Handle it.
27216 * gimplify.c (enum gimplify_omp_var_data): Add
27217 "GOVD_MAP_FORCE_PRESENT".
27218 (gimplify_adjust_omp_clauses_1): Map it to
27219 "GOMP_MAP_FORCE_PRESENT".
27220 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27222 * gimplify.c (oacc_default_clause): Clarify.
27224 2017-05-19 Nathan Sidwell <nathan@acm.org>
27226 LANG_HOOK_REGISTER_DUMPS
27227 * toplev.c (general_init): Call register dump lang hook.
27228 * doc/invoke.texi: Document -fdump-lang option family.
27229 * dumpfile.c (dump_files): Remove class dump here.
27230 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27231 * dumpfile.h (tree_dump_index): Remove TDI_class.
27232 * langhooks-def.h (lhd_register_dumps): Declare.
27233 (LANG_HOOKS_REGISTER_DUMPS): Define.
27234 (LANG_HOOKS_INITIALIZER): Add it.
27235 * langhooks.c (lhd_register_dumps): Define.
27236 * langhooks.h (struct lang_hooks): Add register_dumps.
27238 2017-05-19 Nathan Sidwell <nathan@acm.org>
27240 * context.h (context::set_passes): New.
27241 * context.c (context::context): Do not create pass manager.
27242 * toplev.c (general_init): Create pass manager here.
27244 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27246 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27247 use this splitter if two add or or instructions would also work for
27248 the constant we want to generate.
27250 2017-05-19 Richard Biener <rguenther@suse.de>
27253 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27254 predicate evaluation.
27256 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27258 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27260 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27263 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27265 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27266 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27267 (inline_edge_summary_vec): Turn into ...
27268 (ipa_call_summaries): ... this one.
27269 (redirect_to_unreachable, edge_set_predicate,
27270 evaluate_properties_for_edge, inline_summary_alloc,
27271 reset_ipa_call_summary, reset_inline_summary,
27272 inline_summary_t::duplicate): Update.
27273 (inline_edge_duplication_hook): Turn to ...
27274 (ipa_call_summary_t::duplicate): ... this one.
27275 (inline_edge_removal_hook): Turn to ...
27276 (ipa_call_summary_t::remove): ... this one.
27277 (dump_inline_edge_summary): Turn to ...
27278 (dump_ipa_call_summary): ... this one.
27279 (estimate_function_body_sizes): Update.
27280 (inline_update_callee_summaries): Update.
27281 (remap_edge_change_prob): Update.
27282 (remap_edge_summaries): Update.
27283 (inline_merge_summary): Update.
27284 (do_estimate_edge_time): Update.
27285 (inline_generate_summary): Update.
27286 (inline_read_section): Update.
27287 (inline_read_summary): Update.
27288 (inline_free_summary): Update.
27289 * ipa-inline.c (can_inline_edge_p): Update.
27290 (compute_inlined_call_time): Update.
27291 (want_inline_small_function_p): Update.
27292 (edge_badness): Update.
27293 (early_inliner): Update.
27294 * ipa-inline.h (inline_edge_summary): Turn to ...
27295 (ipa_call_summary): ... this one.
27296 (ipa_call_summary_t): New class.
27297 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27298 (ipa_call_summaries): New.
27299 (inline_edge_summary): Remove.
27300 (estimate_edge_growth): Update.
27301 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27302 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27303 * ipa-split.c (execute_split_functions): Update.
27304 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27306 2017-05-19 Richard Biener <rguenther@suse.de>
27308 PR middle-end/80764
27309 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27311 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27313 * config/rs6000/rs6000.c (struct machine_function): Add field
27314 fpr_is_wrapped_separately.
27315 (rs6000_get_separate_components): Use 64 components. Handle the
27316 new FPR components.
27317 (rs6000_components_for_bb): Handle the FPR components.
27318 (rs6000_emit_prologue_components): Handle the FPR components.
27319 (rs6000_emit_epilogue_components): Handle the FPR components.
27320 (rs6000_set_handled_components): Handle the FPR components.
27321 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27322 that are already separately shrink-wrapped.
27323 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27324 that are already separately shrink-wrapped.
27326 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27329 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27332 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27333 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27334 cases where the register allocator uses a move and an offsettable
27335 memory operation to/from a FPR register on ISA 2.06/2.07.
27336 (define_peephole2 for Altivec d-form store): Likewise.
27338 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27341 * config/i386/mmx.md (*mov<mode>_internal): Enable
27342 alternatives 11, 12, 13 and 14 also for 32bit targets.
27343 Remove alternatives 15, 16, 17 and 18.
27344 * config/i386/sse.md (vec_concatv2di): Change
27345 alternative (!x, *y) to (x, ?!*Yn).
27347 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27349 * dumpfile.h (enum dump_kind): Remove stray comma.
27351 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27353 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27354 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27355 predicate::num_conditions
27356 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27357 (CHANGED): turn into predicate::changed.
27358 (agg_position_info): Move to ipa-predicate.h
27359 (add_condition, predicate::add_clause, predicate::operator &=,
27360 predicate::or_with, predicate::evaluate, predicate::probability,
27361 dump_condition, dump_clause, predicate::dump,
27362 predicate::remap_after_duplication, predicate::remap_after_inlining,
27363 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27364 (evaluate_conditions_for_known_args): Update.
27365 (set_cond_stmt_execution_predicate): Update.
27366 * ipa-inline.h: Include ipa-predicate.h
27367 (condition, inline_param_summary, conditions, agg_position_info,
27368 predicate): Move to ipa-predicate.h
27369 * ipa-predicate.c: New file.
27370 * ipa-predicate.h: New file.
27372 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27374 * final.c (leaf_function_p): Check we are not in a sequence.
27376 2017-05-18 Martin Liska <mliska@suse.cz>
27378 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27379 * dumpfile.c (dump_register): Use new enum dump_kind.
27380 (get_dump_file_name): Likewise.
27381 (dump_enable_all): Likewise.
27382 (dump_switch_p_1): Likewise.
27383 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27384 * dumpfile.h (enum dump_kind): New enum type.
27385 (struct dump_file_info): Create constructor and
27386 format fields and comments.
27387 * passes.c (pass_manager::register_one_dump_file):
27389 * statistics.c (statistics_early_init): Likewise.
27390 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27391 TDF_TREE with TDF_SLIM.
27392 (gather_memory_references_ref): Likewise.
27394 2017-05-18 Martin Liska <mliska@suse.cz>
27396 * vec.h (struct vnull): Use it.
27398 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27400 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27401 (true_predicate, false_predicate, true_predicate_p,
27402 false_predicate_p): Remove.
27403 (single_cond_predicate, not_inlined_predicate): Turn to member function
27405 (add_condition): Update.
27406 (add_clause): Turn to...
27407 (predicate::add_clause): ... this one; update; allow passing NULL
27409 (and_predicates): Turn to ...
27410 (predicate::operator &=): ... this one.
27411 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27412 (or_predicates): Turn to ...
27413 (predicate::or_with): ... this one.
27414 (evaluate_predicate): Turn to ...
27415 (predicate::evaluate): ... this one.
27416 (predicate_probability): Turn to ...
27417 (predicate::probability): ... this one.
27418 (dump_condition): Update.
27419 (dump_predicate): Turn to ...
27420 (predicate::dump): ... this one.
27421 (account_size_time): Update.
27422 (edge_set_predicate): Update.
27423 (set_hint_predicate): UPdate.
27424 (evaluate_conditions_for_known_args): Update.
27425 (evaluate_properties_for_edge): Update.
27426 (remap_predicate_after_duplication): Turn to...
27427 (predicate::remap_after_duplication): ... this one.
27428 (remap_hint_predicate_after_duplication): Update.
27429 (inline_summary_t::duplicate): UPdate.
27430 (dump_inline_edge_summary): Update.
27431 (dump_inline_summary): Update.
27432 (set_cond_stmt_execution_predicate): Update.
27433 (set_switch_stmt_execution_predicate): Update.
27434 (compute_bb_predicates): Update.
27435 (will_be_nonconstant_expr_predicate): Update.
27436 (will_be_nonconstant_predicate): Update.
27437 (phi_result_unknown_predicate): Update.
27438 (predicate_for_phi_result): Update.
27439 (array_index_predicate): Update.
27440 (estimate_function_body_sizes): Update.
27441 (estimate_node_size_and_time): Update.
27442 (estimate_ipcp_clone_size_and_time): Update.
27443 (remap_predicate): Rename to ...
27444 (predicate::remap_after_inlining): ... this one.
27445 (remap_hint_predicate): Update.
27446 (inline_merge_summary): Update.
27447 (inline_update_overall_summary): Update.
27448 (estimate_size_after_inlining): Update.
27449 (read_predicate): Rename to ...
27450 (predicate::stream_in): ... this one.
27451 (read_inline_edge_summary): Update.
27452 (write_predicate): Rename to ...
27453 (predicate::stream_out): ... this one.
27454 (write_inline_edge_summary): Update.
27455 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27456 (clause_t): Turn to uint32_t
27457 (predicate): Turn to class; implement constructor and operators
27459 (size_time_entry): Update.
27460 (inline_summary): Update.
27461 (inline_edge_summary): Update.
27463 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27465 * fold-const.c (fold_binary_loc): Move transformation...
27466 * match.pd (C - X CMP X): ... here.
27468 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27470 * config/sparc/sparc.c (sparc_option_override): Set function
27471 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27472 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27474 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27476 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27478 2017-05-18 Marek Polacek <polacek@redhat.com>
27481 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27482 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27483 gimple_assign_load_p.
27485 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27487 PR middle-end/80692
27488 * real.c (do_compare): Give decimal_do_compare preference over
27489 comparing just the signs.
27491 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27493 * doc/md.texi (Canonicalization of Instructions): Describe the
27494 canonical form of instructions that inherently set a condition
27497 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27499 PR middle-end/80775
27500 * tree-cfg.c: Move deletion of unreachable case statements to after
27501 the merging of consecutive case labels.
27503 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27505 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27506 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27507 restoring of callee-saved registers.
27509 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27511 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27512 * config/visium/visium.c (single_set_and_flags): Likewise.
27513 * config/visium/visium.md (Substitutions): Likewise.
27515 2017-05-17 Martin Liska <mliska@suse.cz>
27517 * cfg.c: Introduce dump_flags_t type and
27518 use it instead of int type.
27520 * cfghooks.c: Likewise.
27521 * cfghooks.h (struct cfg_hooks): Likewise.
27522 * cfgrtl.c: Likewise.
27523 * cfgrtl.h: Likewise.
27524 * cgraph.c (cgraph_node::get_body): Likewise.
27525 * coretypes.h: Likewise.
27526 * domwalk.c: Likewise.
27527 * domwalk.h: Likewise.
27528 * dumpfile.c (struct dump_option_value_info): Likewise.
27529 (dump_enable_all): Likewise.
27530 (dump_switch_p_1): Likewise.
27531 (opt_info_switch_p): Likewise.
27532 * dumpfile.h (enum tree_dump_index): Likewise.
27533 (struct dump_file_info): Likewise.
27534 * genemit.c: Likewise.
27535 * generic-match-head.c: Likewise.
27536 * gengtype.c (open_base_files): Likewise.
27537 * gimple-pretty-print.c: Likewise.
27538 * gimple-pretty-print.h: Likewise.
27539 * graph.c (print_graph_cfg): Likewise.
27540 * graphite-scop-detection.c (dot_all_sese): Likewise.
27541 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27542 * loop-unroll.c (report_unroll): Likewise.
27543 * passes.c (pass_manager::register_one_dump_file): Likewise.
27544 * print-tree.c: Likewise.
27545 * statistics.c: Likewise.
27546 * tree-cfg.c: Likewise.
27547 * tree-cfg.h: Likewise.
27548 * tree-dfa.c: Likewise.
27549 * tree-dfa.h: Likewise.
27550 * tree-dump.c (dump_function): Likewise.
27551 * tree-dump.h (struct dump_info): Likewise.
27552 * tree-pretty-print.c: Likewise.
27553 * tree-pretty-print.h: Likewise.
27554 * tree-ssa-live.c: Likewise.
27555 * tree-ssa-live.h: Likewise.
27556 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27557 * tree-vect-loop.c: Likewise.
27558 * tree-vect-slp.c: Likewise.
27560 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27561 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27563 PR tree-optimization/80457
27564 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27565 of all arguments to a statement as scalar_to_vec operations.
27566 (vectorizable_call): Adjust call to vect_model_simple_cost for
27568 (vectorizable_conversion): Likewise.
27569 (vectorizable_assignment): Likewise.
27570 (vectorizable_shift): Likewise.
27571 (vectorizable_operation): Likewise.
27572 (vectorizable_comparison): Likewise.
27573 (vect_is_simple_cond): Record the def types for operands.
27574 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27575 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27576 for statement argument count.
27578 2017-05-16 Carl Love <cel@us.ibm.com>
27580 * config/rs6000/rs6000-c: Add support for built-in functions
27581 vector unsigned long long vec_bperm (vector unsigned long long,
27582 vector unsigned char)
27583 vector signed long long vec_mule (vector signed int,
27585 vector unsigned long long vec_mule (vector unsigned int,
27586 vector unsigned int)
27587 vector signed long long vec_mulo (vector signed int,
27589 vector unsigned long long vec_mulo (vector unsigned int,
27590 vector unsigned int)
27591 vector signed char vec_sldw (vector signed char,
27592 vector signed char,
27594 vector unsigned char vec_sldw (vector unsigned char,
27595 vector unsigned char,
27597 vector signed short vec_sldw (vector signed short,
27598 vector signed short,
27600 vector unsigned short vec_sldw (vector unsigned short,
27601 vector unsigned short,
27603 vector signed int vec_sldw (vector signed int,
27606 vector unsigned int vec_sldw (vector unsigned int,
27607 vector unsigned int,
27609 vector signed long long vec_sldw (vector signed long long,
27610 vector signed long long,
27612 vector unsigned long long vec_sldw (vector unsigned long long,
27613 vector unsigned long long,
27615 * config/rs6000/rs6000-c: Add support for built-in functions
27616 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27617 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27618 * doc/extend.texi: Update the built-in documentation for the
27619 new built-in functions.
27621 2017-05-16 Marek Polacek <polacek@redhat.com>
27625 * tree.c (save_expr): Don't fold the expression.
27627 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27629 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27630 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27631 and (?*y,m). Update insn attributes.
27633 2017-05-16 Martin Liska <mliska@suse.cz>
27635 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27636 flags argument of print_gimple_stmt, print_gimple_expr,
27637 print_generic_stmt and print_generic_expr.
27638 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27639 * coretypes.h: Likewise.
27640 * except.c (dump_eh_tree): Likewise.
27641 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27642 * gimple-pretty-print.h: Likewise.
27643 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27644 (backprop::push_to_worklist): Likewise.
27645 (backprop::pop_from_worklist): Likewise.
27646 (backprop::process_use): Likewise.
27647 (backprop::intersect_uses): Likewise.
27648 (note_replacement): Likewise.
27649 * gimple-ssa-store-merging.c
27650 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27651 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27652 (pass_store_merging::execute): Likewise.
27653 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27654 (ssa_base_cand_dump_callback): Likewise.
27655 (dump_incr_vec): Likewise.
27656 (replace_refs): Likewise.
27657 (replace_mult_candidate): Likewise.
27658 (create_add_on_incoming_edge): Likewise.
27659 (create_phi_basis): Likewise.
27660 (insert_initializers): Likewise.
27661 (all_phi_incrs_profitable): Likewise.
27662 (introduce_cast_before_cand): Likewise.
27663 (replace_one_candidate): Likewise.
27664 * gimplify.c (gimplify_expr): Likewise.
27665 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27666 (set_rename): Likewise.
27667 (rename_uses): Likewise.
27668 (copy_loop_phi_nodes): Likewise.
27669 (add_close_phis_to_merge_points): Likewise.
27670 (copy_loop_close_phi_args): Likewise.
27671 (copy_cond_phi_args): Likewise.
27672 (graphite_copy_stmts_from_block): Likewise.
27673 (translate_pending_phi_nodes): Likewise.
27674 * graphite-poly.c (print_pdr): Likewise.
27675 (dump_gbb_cases): Likewise.
27676 (dump_gbb_conditions): Likewise.
27677 (print_scop_params): Likewise.
27678 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27679 (build_cross_bb_scalars_use): Likewise.
27680 (gather_bbs::before_dom_children): Likewise.
27681 * hsa-dump.c (dump_hsa_immed): Likewise.
27682 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27683 (get_replacement_map): Likewise.
27684 * ipa-inline-analysis.c (dump_condition): Likewise.
27685 (estimate_function_body_sizes): Likewise.
27686 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27687 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27688 * ipa-prop.c (ipa_dump_param): Likewise.
27689 (ipa_print_node_jump_functions_for_edge): Likewise.
27690 (ipa_modify_call_arguments): Likewise.
27691 (ipa_modify_expr): Likewise.
27692 (ipa_dump_param_adjustments): Likewise.
27693 (ipa_dump_agg_replacement_values): Likewise.
27694 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27695 * ipa-pure-const.c (check_stmt): Likewise.
27696 (pass_nothrow::execute): Likewise.
27697 * ipa-split.c (execute_split_functions): Likewise.
27698 * omp-offload.c (dump_oacc_loop_part): Likewise.
27699 (dump_oacc_loop): Likewise.
27700 * trans-mem.c (tm_log_emit): Likewise.
27701 (tm_memopt_accumulate_memops): Likewise.
27702 (dump_tm_memopt_set): Likewise.
27703 (dump_tm_memopt_transform): Likewise.
27704 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27705 (print_loop): Likewise.
27706 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27707 (chkp_gather_checks_info): Likewise.
27708 (chkp_get_check_result): Likewise.
27709 (chkp_remove_check_if_pass): Likewise.
27710 (chkp_use_outer_bounds_if_possible): Likewise.
27711 (chkp_reduce_bounds_lifetime): Likewise.
27712 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27713 (chkp_mark_completed_bounds): Likewise.
27714 (chkp_register_incomplete_bounds): Likewise.
27715 (chkp_mark_invalid_bounds): Likewise.
27716 (chkp_maybe_copy_and_register_bounds): Likewise.
27717 (chkp_build_returned_bound): Likewise.
27718 (chkp_get_bound_for_parm): Likewise.
27719 (chkp_build_bndldx): Likewise.
27720 (chkp_get_bounds_by_definition): Likewise.
27721 (chkp_generate_extern_var_bounds): Likewise.
27722 (chkp_get_bounds_for_decl_addr): Likewise.
27723 * tree-chrec.c (chrec_apply): Likewise.
27724 * tree-data-ref.c (dump_data_reference): Likewise.
27725 (dump_subscript): Likewise.
27726 (dump_data_dependence_relation): Likewise.
27727 (analyze_overlapping_iterations): Likewise.
27728 * tree-inline.c (expand_call_inline): Likewise.
27729 (tree_function_versioning): Likewise.
27730 * tree-into-ssa.c (dump_defs_stack): Likewise.
27731 (dump_currdefs): Likewise.
27732 (dump_names_replaced_by): Likewise.
27733 (dump_update_ssa): Likewise.
27734 (update_ssa): Likewise.
27735 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27736 * tree-parloops.c (build_new_reduction): Likewise.
27737 (try_create_reduction_list): Likewise.
27738 (ref_conflicts_with_region): Likewise.
27739 (oacc_entry_exit_ok_1): Likewise.
27740 (oacc_entry_exit_single_gang): Likewise.
27741 * tree-pretty-print.h: Likewise.
27742 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27743 (get_scalar_evolution): Likewise.
27744 (add_to_evolution): Likewise.
27745 (get_loop_exit_condition): Likewise.
27746 (analyze_evolution_in_loop): Likewise.
27747 (analyze_initial_condition): Likewise.
27748 (analyze_scalar_evolution): Likewise.
27749 (instantiate_scev): Likewise.
27750 (number_of_latch_executions): Likewise.
27751 (gather_chrec_stats): Likewise.
27752 (final_value_replacement_loop): Likewise.
27753 (scev_const_prop): Likewise.
27754 * tree-sra.c (dump_access): Likewise.
27755 (disqualify_candidate): Likewise.
27756 (create_access): Likewise.
27757 (reject): Likewise.
27758 (maybe_add_sra_candidate): Likewise.
27759 (create_access_replacement): Likewise.
27760 (analyze_access_subtree): Likewise.
27761 (analyze_all_variable_accesses): Likewise.
27762 (sra_modify_assign): Likewise.
27763 (initialize_constant_pool_replacements): Likewise.
27764 (find_param_candidates): Likewise.
27765 (decide_one_param_reduction): Likewise.
27766 (replace_removed_params_ssa_names): Likewise.
27767 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27768 * tree-ssa-copy.c (dump_copy_of): Likewise.
27769 (copy_prop_visit_cond_stmt): Likewise.
27770 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27771 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27772 (record_equivalences_from_stmt): Likewise.
27773 * tree-ssa-dse.c (compute_trims): Likewise.
27774 (delete_dead_call): Likewise.
27775 (delete_dead_assignment): Likewise.
27776 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27777 (forward_propagate_into_cond): Likewise.
27778 (pass_forwprop::execute): Likewise.
27779 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27780 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27782 (move_computations_worker): Likewise.
27783 (execute_sm): Likewise.
27784 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27785 (remove_exits_and_undefined_stmts): Likewise.
27786 (remove_redundant_iv_tests): Likewise.
27787 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27788 (adjust_iv_update_pos): Likewise.
27789 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27790 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27791 (value_replacement): Likewise.
27792 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27793 * tree-ssa-pre.c (print_pre_expr): Likewise.
27794 (get_representative_for): Likewise.
27795 (create_expression_by_pieces): Likewise.
27796 (insert_into_preds_of_block): Likewise.
27797 (eliminate_insert): Likewise.
27798 (eliminate_dom_walker::before_dom_children): Likewise.
27799 (eliminate): Likewise.
27800 (remove_dead_inserted_code): Likewise.
27801 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27802 * tree-ssa-reassoc.c (get_rank): Likewise.
27803 (eliminate_duplicate_pair): Likewise.
27804 (eliminate_plus_minus_pair): Likewise.
27805 (eliminate_not_pairs): Likewise.
27806 (undistribute_ops_list): Likewise.
27807 (eliminate_redundant_comparison): Likewise.
27808 (update_range_test): Likewise.
27809 (optimize_range_tests_var_bound): Likewise.
27810 (optimize_vec_cond_expr): Likewise.
27811 (rewrite_expr_tree): Likewise.
27812 (rewrite_expr_tree_parallel): Likewise.
27813 (linearize_expr): Likewise.
27814 (break_up_subtract): Likewise.
27815 (linearize_expr_tree): Likewise.
27816 (attempt_builtin_powi): Likewise.
27817 (attempt_builtin_copysign): Likewise.
27818 (transform_stmt_to_copy): Likewise.
27819 (transform_stmt_to_multiply): Likewise.
27820 (dump_ops_vector): Likewise.
27821 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27822 (print_scc): Likewise.
27823 (set_ssa_val_to): Likewise.
27824 (visit_reference_op_store): Likewise.
27825 (visit_use): Likewise.
27826 (sccvn_dom_walker::before_dom_children): Likewise.
27827 (run_scc_vn): Likewise.
27828 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27830 (expr_hash_elt::print): Likewise.
27831 (const_and_copies::pop_to_marker): Likewise.
27832 (const_and_copies::record_const_or_copy_raw): Likewise.
27833 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27834 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27835 (dump_predicates): Likewise.
27836 (find_uninit_use): Likewise.
27837 (warn_uninitialized_phi): Likewise.
27838 (pass_late_warn_uninitialized::execute): Likewise.
27839 * tree-ssa.c (verify_vssa): Likewise.
27840 (verify_ssa): Likewise.
27841 (maybe_optimize_var): Likewise.
27842 * tree-vrp.c (dump_value_range): Likewise.
27843 (dump_all_value_ranges): Likewise.
27844 (dump_asserts_for): Likewise.
27845 (register_edge_assert_for_2): Likewise.
27846 (vrp_visit_cond_stmt): Likewise.
27847 (vrp_visit_switch_stmt): Likewise.
27848 (vrp_visit_stmt): Likewise.
27849 (vrp_visit_phi_node): Likewise.
27850 (simplify_cond_using_ranges_1): Likewise.
27851 (fold_predicate_in): Likewise.
27852 (evrp_dom_walker::before_dom_children): Likewise.
27853 (evrp_dom_walker::push_value_range): Likewise.
27854 (evrp_dom_walker::pop_value_range): Likewise.
27855 (execute_early_vrp): Likewise.
27857 2017-05-16 Richard Biener <rguenther@suse.de>
27859 * dwarf2out.c (loc_list_from_tree_1): Do not create
27860 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27862 2017-05-16 Richard Biener <rguenther@suse.de>
27864 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27866 (note_variable_value_in_expr): If we resolved the decl ref
27867 do not push to the stack.
27869 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27871 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27872 operations in fast-math mode.
27873 (vaddq_f16): Likewise.
27874 (vmul_f16): Likewise.
27875 (vmulq_f16): Likewise.
27876 (vsub_f16): Likewise.
27877 (vsubq_f16): Likewise.
27878 * config/arm/neon.md (add<mode>3): New.
27880 (fma:<VH:mode>3): New. Also remove outdated comment.
27883 2017-05-16 Martin Liska <mliska@suse.cz>
27887 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27888 (odr_types_equivalent_p): Likewise.
27890 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27892 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27894 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27897 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27898 non-interunit SSE move alternatives with '?'.
27899 (zero-extendsidi peephole2): New peephole to skip intermediate
27900 general register in SSE zero-extend sequence.
27902 2017-05-15 Jeff Law <law@redhat.com>
27904 * reorg.c (relax_delay_slots): Create a new variable to hold
27905 the temporary target rather than clobbering TARGET_LABEL.
27907 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27908 missing argument to extract_bit_field call.
27909 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27911 2017-05-15 Martin Liska <mliska@suse.cz>
27914 * gcc.c (process_command): Do not allow empty argument of -o option.
27916 2017-05-15 Renlin Li <renlin.li@arm.com>
27918 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27919 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27920 * config/aarch64/constraints.md (Usf): Add long call check.
27921 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27922 (call_value): Likewise.
27923 (sibcall): Likewise.
27924 (sibcall_value): Likewise.
27926 (call_value_insn): New.
27927 (sibcall_insn): Update rtx pattern.
27928 (sibcall_value_insn): Likewise.
27929 (call_internal): Remove.
27930 (call_value_internal): Likewise.
27931 (sibcall_internal): Likewise.
27932 (sibcall_value_internal): Likewise.
27933 (call_reg): Likewise.
27934 (call_symbol): Likewise.
27935 (call_value_reg): Likewise.
27936 (call_value_symbol): Likewise.
27938 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27941 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27943 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27945 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27946 compatible with CCGOCmode and with CCZmode.
27948 2017-05-14 Martin Sebor <msebor@redhat.com>
27950 PR middle-end/77671
27951 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27952 (gimple_fold_builtin_snprintf): Same.
27953 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27954 (gimple_fold_builtin_snprintf): Same.
27955 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27956 of character types.
27957 (is_call_safe): New function.
27958 (try_substitute_return_value): Call it.
27959 (try_simplify_call): New function.
27960 (pass_sprintf_length::handle_gimple_call): Call it.
27962 2017-05-14 Martin Sebor <msebor@redhat.com>
27964 PR middle-end/80669
27965 * builtins.c (expand_builtin_stpncpy): Simplify.
27967 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27969 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27970 * config/i386/i386.h
27971 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27972 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27973 (struct machine_function): Add new members call_ms2sysv,
27974 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27975 (struct machine_frame_state): New fields sp_realigned and
27976 sp_realigned_offset.
27977 * config/i386/i386.c
27978 (enum xlogue_stub): New enum.
27979 (enum xlogue_stub_sets): New enum.
27980 (class xlogue_layout): New class.
27981 (struct ix86_frame): New fields stack_realign_allocate_offset,
27982 stack_realign_offset and outlined_save_offset. Modify comments to
27983 detail stack layout when using out-of-line stubs.
27984 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27985 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27986 -mcall-ms2sysv-xlogues.
27987 (stub_managed_regs): New static variable.
27988 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27989 registers managed by out-of-line stub.
27990 (disable_call_ms2sysv_xlogues): New function.
27991 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27992 m->call_ms2sysv when appropriate and compute frame layout for
27994 (sp_valid_at, fp_valid_at): New inline functions.
27995 (choose_basereg): New function.
27996 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27998 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27999 Use align parameter of choose_baseaddr to generated aligned SSE movs
28001 (pro_epilogue_adjust_stack): Modify to track
28002 machine_frame_state::sp_realigned.
28003 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28004 (ix86_nsaved_sseregs): Likewise.
28005 (ix86_emit_save_regs): Likewise.
28006 (ix86_emit_save_regs_using_mov): Likewise.
28007 (ix86_emit_save_sse_regs_using_mov): Likewise.
28008 (get_scratch_register_on_entry): Likewise.
28009 (gen_frame_set): New function.
28010 (gen_frame_load): Likewise.
28011 (gen_frame_store): Likewise.
28012 (emit_outlined_ms2sysv_save): Likewise.
28013 (emit_outlined_ms2sysv_restore): Likewise.
28014 (ix86_expand_prologue): Modify stack re-alignment code and call
28015 emit_outlined_ms2sysv_save when appropriate.
28016 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28017 parameter rtx_insn *insn, which allows the function to be used to only
28018 generate the notes.
28019 (ix86_expand_epilogue): Modify validity checks of frame and stack
28020 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28021 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28022 * config/i386/predicates.md
28023 (save_multiple): New predicate.
28024 (restore_multiple): Likewise.
28025 * config/i386/sse.md
28026 (save_multiple<mode>): New pattern.
28027 (save_multiple_realign<mode>): Likewise.
28028 (restore_multiple<mode>): Likewise.
28029 (restore_multiple_and_return<mode>): Likewise.
28030 (restore_multiple_leave_return<mode>): Likewise.
28031 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28033 2017-05-14 Julia Koval <julia.koval@intel.com>
28035 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28036 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28037 (__builtin_ia32_xsetbv): New builtins.
28038 * config/i386/i386.c (ix86_expand_special_args_builtin):
28040 (ix86_expand_builtin): Special expand for new intrinsics.
28041 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28042 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28043 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28045 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28047 * cfganal.c (inverted_post_order_compute): Change argument type
28049 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28050 * df-core.c (rest_of_handle_df_initialize): Adjust.
28051 (rest_of_handle_df_finish): Likewise.
28052 (df_analyze_1): Likewise.
28053 (df_analyze): Likewise.
28054 (loop_inverted_post_order_compute): Change argument to be a vec *.
28055 (df_analyze_loop): Adjust.
28056 (df_get_n_blocks): Likewise.
28057 (df_get_postorder): Likewise.
28058 * df.h (struct df_d): Change field to be a vec.
28059 * lcm.c (compute_laterin): Adjust.
28060 (compute_available): Likewise.
28061 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28062 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28063 * tree-ssa-pre.c (compute_antic): Likewise.
28065 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28067 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28068 (depth_first_search::depth_first_search): Change structure init
28069 function to this constructor.
28070 (depth_first_search::add_bb): Rename function to this member.
28071 (depth_first_search::execute): Likewise.
28072 (flow_dfs_compute_reverse_finish): Adjust.
28074 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28076 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28077 (longest_simple_path): Likewise.
28078 * shrink-wrap.c (spread_components): Likewise.
28079 (disqualify_problematic_components): Likewise.
28080 (emit_common_heads_for_components): Likewise.
28081 (emit_common_tails_for_components): Likewise.
28082 (insert_prologue_epilogue_for_components): Likewise.
28084 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28086 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28089 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28091 * df-core.c (df_set_blocks): Start using auto_bitmap.
28092 (df_compact_blocks): Likewise.
28093 * df-problems.c (df_rd_confluence_n): Likewise.
28094 * df-scan.c (df_insn_rescan_all): Likewise.
28095 (df_process_deferred_rescans): Likewise.
28096 (df_update_entry_block_defs): Likewise.
28097 (df_update_exit_block_uses): Likewise.
28098 (df_entry_block_bitmap_verify): Likewise.
28099 (df_exit_block_bitmap_verify): Likewise.
28100 (df_scan_verify): Likewise.
28101 * lra-constraints.c (lra_constraints): Likewise.
28102 (undo_optional_reloads): Likewise.
28103 (lra_undo_inheritance): Likewise.
28104 * lra-remat.c (calculate_gen_cands): Likewise.
28105 (do_remat): Likewise.
28106 * lra-spills.c (assign_spill_hard_regs): Likewise.
28107 (spill_pseudos): Likewise.
28108 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28109 (bitmap_set_subtract_values): Likewise.
28111 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28113 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28114 management with auto_bitmap.
28115 (fix_inter_tick): Likewise.
28116 (fix_recovery_deps): Likewise.
28117 * ira.c (add_store_equivs): Likewise.
28118 (find_moveable_pseudos): Likewise.
28119 (split_live_ranges_for_shrink_wrap): Likewise.
28120 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28121 (rtx_reuse_manager::seen_def_p): Likewise.
28122 (rtx_reuse_manager::set_seen_def): Likewise.
28123 * print-rtl.h (class rtx_reuse_manager): Likewise.
28125 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28127 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28129 (migrate_btr_def): Likewise.
28130 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28131 * df-core.c (loop_post_order_compute): Likewise.
28132 (loop_inverted_post_order_compute): Likewise.
28133 * hsa-common.h: Likewise.
28134 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28135 * init-regs.c (initialize_uninitialized_regs): Likewise.
28136 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28137 (inline_small_functions): Likewise.
28138 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28139 * ira.c (combine_and_move_insns): Likewise.
28140 (build_insn_chain): Likewise.
28141 * loop-invariant.c (find_invariants): Likewise.
28142 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28143 * predict.c (tree_predict_by_opcode): Likewise.
28144 (predict_paths_leading_to): Likewise.
28145 (predict_paths_leading_to_edge): Likewise.
28146 (estimate_loops_at_level): Likewise.
28147 (estimate_loops): Likewise.
28148 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28149 (spread_components): Likewise.
28150 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28151 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28152 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28153 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28154 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28155 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28156 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28157 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28158 (mark_threaded_blocks): Likewise.
28159 (thread_through_all_blocks): Likewise.
28160 * tree-ssa.c (verify_ssa): Likewise.
28161 (execute_update_addresses_taken): Likewise.
28162 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28164 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28166 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28168 (post_order_compute): Likewise.
28169 (inverted_post_order_compute): Likewise.
28170 (pre_and_rev_post_order_compute_fn): Likewise.
28172 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28174 * genrecog.c (int_set::int_set): Explicitly construct our
28175 auto_vec base class.
28176 * vec.h (auto_vec::auto_vec): New constructor.
28178 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28180 * bitmap.h (class auto_bitmap): New constructor taking
28181 bitmap_obstack * argument.
28183 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28185 * bitmap.h (class auto_bitmap): Change type of m_bits to
28186 bitmap_head, and adjust ctor / dtor and member operators.
28188 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28190 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28191 when returned register mode doesn't match original mode.
28193 2017-05-12 Jeff Law <law@redhat.com>
28194 Jakub Jelinek <jakub@redhat.com>
28196 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28197 we look for cc setter after the compare-elim changes.
28198 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28199 within the vector to match what compare-elim now expects.
28200 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28201 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28203 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28204 after the compare-elim changes.
28205 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28206 the vector to match what compare-elim now expects.
28207 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28208 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28209 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28210 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28211 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28213 * config/visium/visium.c (single_set_and_flags): Fix where
28214 we look for cc setter after the compare-elim changes.
28215 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28216 with the vector to match what compare-elim now expects.
28217 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28218 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28219 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28220 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28221 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28222 (neg<mode>2_insn_set_overflow): Likewise.
28224 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28226 PR middle-end/79794
28227 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28228 maybe_expand_insn call, set ops[0].target. If still set after call,
28229 set alt_rtl. Add extra arg to recursive calls.
28230 (extract_bit_field): Add alt_rtl argument. Pass to
28232 * expmed.h (extract_bit_field): Fix prototype.
28233 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28234 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28235 to extract_bit_field_calls.
28236 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28237 Pass alt_rtl to extract_bit_field calls.
28238 * calls.c (store_unaligned_arguments_into_psuedos)
28239 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28240 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28242 * optabs.h (struct expand_operand): Add target bitfield.
28244 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28246 * compare-elim.c (try_eliminate_compare): Canonicalize
28247 operation with embedded compare to
28248 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28249 (set (reg) (operation)].
28251 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28253 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28256 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28257 cost of adding a carry flag for ADC instruction.
28258 [case MINUS]: Ignore the cost of subtracting a carry flag
28259 for SBB instruction.
28261 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28263 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28265 * config/rs6000/bmiintrin.h: New file.
28266 * config/rs6000/bmi2intrin.h: New file.
28267 * config/rs6000/x86intrin.h: New file.
28269 2017-05-12 Jeff Law <law@redhat.com>
28271 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28274 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28276 PR middle-end/80707
28277 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28279 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28281 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28282 early expansion of vector divide builtins.
28283 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28284 builtins identified as having unsigned arguments.
28286 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28288 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28289 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28290 expansion of vector logical operations (and, andc, or, xor,
28293 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28295 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28296 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28298 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28300 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28301 early GIMPLE expansion of vector multiplies.
28303 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28305 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28306 TARGET_HAVE_MOVT conditional.
28307 (movt splitter): Likewise.
28309 2017-05-12 Richard Biener <rguenther@suse.de>
28311 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28312 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28313 Fold all stmts not inplace.
28315 2017-05-12 Richard Biener <rguenther@suse.de>
28317 PR tree-optimization/80713
28318 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28319 inserted_exprs bit for not removed stmts.
28321 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28323 PR middle-end/69921
28324 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28325 parallelized" attribute for parallelized OpenACC kernels.
28326 * omp-offload.c (execute_oacc_device_lower): Use it.
28328 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28329 Set "oacc kernels" attribute.
28330 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28331 parameter. Adjust all users.
28332 (oacc_fn_attrib_kernels_p): Remove function.
28333 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28334 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28335 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28336 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28337 assert "oacc kernels" attribute is set.
28339 2017-05-11 Carl Love <cel@us.ibm.com>
28341 * config/rs6000/rs6000-c: Add support for built-in functions
28342 vector unsigned char vec_popcnt (vector signed char)
28343 vector unsigned char vec_popcnt (vector unsigned char)
28344 vector unsigned short vec_popcnt (vector signed short)
28345 vector unsigned short vec_popcnt (vector unsigned short)
28346 vector unsigned int vec_popcnt (vector signed int)
28347 vector unsigned int vec_popcnt (vector unsigned int)
28348 vector unsigned long long vec_popcnt (vector signed long long)
28349 vector unsigned long long vec_popcnt (vector unsigned long long)
28350 vector signed long long vec_slo (vector signed long long,
28351 vector signed char)
28352 vector signed long long vec_slo (vector signed long long,
28353 vector unsigned char)
28354 vector unsigned long long vec_slo (vector unsigned long long,
28355 vector signed char)
28356 vector unsigned long long vec_slo (vector unsigned long long,
28357 vector unsigned char)
28358 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28359 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28360 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28361 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28362 * doc/extend.texi: Update the built-in documentation file for the
28363 new built-in functions.
28365 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28367 * attribs.h (sorted_attr_string): Move machine independent
28368 functions for target clone support from the i386 port to common
28369 code. Rename ix86_function_versions to common_function_versions.
28370 Rename make_name to make_unique_name.
28371 (common_function_versions): Likewise.
28372 (make_unique_name): Likewise.
28373 (make_dispatcher_decl): Likewise.
28374 (is_function_default_version): Likewise.
28375 * attribs.c (attr_strcmp): Likewise.
28376 (sorted_attr_string): Likewise.
28377 (common_function_versions): Likewise.
28378 (make_unique_name): Likewise.
28379 (make_dispatcher_decl): Likewise.
28380 (is_function_default_version): Likewise.
28381 * config/i386/i386.c (attr_strcmp): Likewise.
28382 (sorted_attr_string): Likewise.
28383 (ix86_function_versions): Likewise.
28384 (make_name): Likewise.
28385 (make_dispatcher_decl): Likewise.
28386 (is_function_default_version): Likewise.
28387 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28389 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28392 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28393 Account for direct move costs for vec_construct of integer
28396 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28399 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28400 (UNSPEC_STX_ATOMIC): Ditto.
28401 (loaddi_via_sse): New insn.
28402 (storedi_via_sse): Ditto.
28403 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28404 Update corresponding peephole2 patterns.
28405 (atomic_storedi_fpu): Ditto.
28407 2017-05-11 Julia Koval <julia.koval@intel.com>
28409 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28410 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28412 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28413 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28414 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28416 2017-05-11 Nathan Sidwell <nathan@acm.org>
28418 * graphite-poly.c: Include dumpfile.h.
28420 * dumpfle.h (dump_function): Declare here ...
28421 * tree-dump.h (dump_function): ... not here.
28422 * dumpfile.c: #include tree-cfg.h.
28423 (dump_function): Move here from ...
28424 * tree-dump.c (dump_function): ... here.
28425 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28426 * graphite-poly.c: Don't include tree-dump.h.
28427 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28428 * print-tree.c: Likewise.
28429 * stor-layout.c: Likewise.
28430 * tree-nested.c: Likewise.
28432 * dumpfile.c (dump_start): Use TDF_FLAGS.
28433 (dump_enable_all): Fix TDF_KIND check thinko.
28435 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28437 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28438 array entries to represent two legal parameterizations of the
28439 overloaded __builtin_cmpb function, as represented by the
28440 P6_OV_BUILTIN_CMPB constant.
28441 (altivec_resolve_overloaded_builtin): Add special case handling
28442 for the __builtin_cmpb function, as represented by the
28443 P6_OV_BUILTIN_CMPB constant.
28444 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28445 (BU_P6_64BIT_2): New macro.
28446 (BU_P6_OVERLOAD_2): New macro
28447 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28448 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28449 (CMPB): Add overload support to represent both 32-bit and 64-bit
28450 compare-bytes function.
28451 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28452 support for TARGET_CMPB.
28453 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28454 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28455 documentation of the __builtin_cmpb overloaded built-in function.
28457 2017-05-11 Richard Biener <rguenther@suse.de>
28459 PR tree-optimization/80705
28460 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28461 bases are not vectorizable.
28463 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28465 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28466 when counting register pressure.
28468 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28470 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28471 (iv_ca_more_deps): Renamed to ...
28472 (iv_ca_compare_deps): ... this.
28473 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28475 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28477 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28479 (determine_group_iv_costs): ... here.
28480 (find_inv_vars_cb): Record inv var if it's not recorded before.
28482 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28484 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28485 (get_shiftadd_cost): Ditto.
28487 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28489 * tree-ssa-address.c: Include header file.
28490 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28492 (add_to_parts): Refactor.
28493 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28494 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28497 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28499 PR tree-optimization/53090
28500 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28502 (extract_cond_operands): Detect condition with IV on both sides
28503 and return COMP_IV_EXPR_2.
28504 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28505 (rewrite_use_compare): Simplify by removing call to function
28506 extract_cond_operands.
28508 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28510 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28511 (extract_cond_operands): Detect condition comparing against non-
28512 invariant bound and return appropriate enum value.
28513 (find_interesting_uses_cond): Update use of extract_cond_operands.
28514 Handle its return value accordingly.
28515 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28517 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28519 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28520 nonlinear iv_use computation in loop invariant sensitive way.
28522 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28524 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28525 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28527 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28529 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28530 (dump_cand): Support iv_cand.inv_exprs.
28531 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28533 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28536 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28538 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28540 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28541 as local function. Include necessary header files.
28542 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28544 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28546 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28548 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28550 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28551 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28552 RSHIFT_EXPR and BIT_NOT_EXPR.
28554 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28556 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28557 (adjust_setup_cost): New parameter supporting round up adjustment.
28558 (struct address_cost_data): Delete.
28559 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28560 (split_address_cost, ptr_difference_cost): Delete.
28561 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28562 (struct ainc_cost_data): New struct.
28563 (get_address_cost_ainc): New function.
28564 (get_address_cost, get_computation_cost): Reimplement.
28565 (determine_group_iv_cost_address): Record inv_expr for all uses of
28567 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28568 (iv_ca_has_deps): Reimplemented to ...
28569 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28571 (iv_ca_extend): Call iv_ca_more_deps.
28573 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28575 * tree-ssa-address.c (struct mem_address): Move to header file.
28576 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28577 * tree-ssa-address.h (struct mem_address): Move from C file.
28578 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28580 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28582 * tree-affine.h (aff_combination_type): New interface.
28583 (aff_combination_zero_p): Remove static.
28584 (aff_combination_const_p): New interface.
28585 (aff_combination_singleton_var_p): New interfaces.
28587 2017-05-11 Richard Biener <rguenther@suse.de>
28589 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28590 Skip unreachable blocks and destinations.
28591 (eliminate): Move stmt removal and fixup ...
28592 (fini_eliminate): ... here. Skip inserted exprs.
28593 (pass_pre::execute): Move fini_pre after fini_eliminate.
28594 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28595 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28596 PRE to get rid of dead code that has invalid SSA form and
28597 split critical edges again.
28599 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28601 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28603 2017-05-11 Richard Biener <rguenther@suse.de>
28605 * passes.c (execute_function_todo): Verify loops if they are
28606 said to be up-to-date.
28607 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28608 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28610 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28613 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28614 handle calling assemble_external ourself.
28617 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28618 modes with zero size. Enhance comment.
28620 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28622 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28623 built-ins for vec_xl and vec_xst with short and char pointer
28626 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28628 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28629 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28630 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28631 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28632 (_mm_maskz_min_round_ss): New intrinsics.
28633 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28634 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28635 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28636 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28637 (__builtin_ia32_minss_mask_round): New builtins.
28638 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28639 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28640 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28642 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28643 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28645 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28648 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28650 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28651 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28652 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28653 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28654 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28655 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28656 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28657 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28658 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28659 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28660 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28661 (__builtin_ia32_mulss_mask_round): New builtins.
28662 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28663 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28664 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28666 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28667 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28669 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28672 2017-05-10 Julia Koval <julia.koval@intel.com>
28674 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28675 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28676 (_mm256_setr_m128i): New intrinsics.
28678 2017-05-10 Julia Koval <julia.koval@intel.com>
28680 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28681 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28682 (_mm_maskz_rcp14_ss): New intrinsics.
28683 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28684 (__builtin_ia32_rcp14ss_mask): New builtins.
28685 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28687 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28689 PR tree-optimization/51513
28690 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28691 (assert_unreachable_fallthru_edge_p): Use it.
28692 (group_case_labels_stmt): Likewise.
28693 * tree-cfg.h: Prototype it.
28694 * stmt.c: Include cfghooks.h and tree-cfg.h.
28695 (emit_case_dispatch_table) <gap_label>: New local variable.
28696 Use it to fill dispatch table gaps.
28697 Test for default_label before updating probabilities.
28698 (expand_case) <default_label>: Remove unneeded initialization.
28699 Test for unreachable default case statement and remove its edge.
28700 Set default_label accordingly.
28701 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28703 2017-05-10 Carl Love <cel@us.ibm.com>
28705 * config/rs6000/rs6000-c: Add support for built-in functions
28706 vector signed char vec_neg (vector signed char)
28707 vector signed short int vec_neg (vector short int)
28708 vector signed int vec_neg (vector signed int)
28709 vector signed long long vec_neg (vector signed long long)
28710 vector float vec_neg (vector float)
28711 vector double vec_neg (vector double)
28712 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28714 * config/rs6000/altivec.h: Add define for vec_neg
28715 * doc/extend.texi: Update the built-in documentation for the
28716 new built-in functions.
28718 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28720 PR tree-optimization/77644
28721 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28723 2017-05-10 Nathan Sidwell <nathan@acm.org>
28725 * dumpfile.h (TDI_lang_all): New.
28726 (TDF_KIND): New. Renumber others
28727 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28729 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28731 (get_dump_file_name): Adjust suffix generation.
28732 (dump_enable_all): Use TDF_KIND.
28733 * doc/invoke.texi (-fdump-lang-all): Document.
28735 * dumpfile.h: Tabify.
28737 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28740 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28741 Move member access before delete.
28743 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28745 * tree-inline.c (expand_call_inline): Split block at stmt
28748 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28751 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28752 are now unused after splitting mov{sf,sd}_hardfloat.
28753 (f32_lr2): Likewise.
28754 (f32_lm): Likewise.
28755 (f32_lm2): Likewise.
28756 (f32_li): Likewise.
28757 (f32_li2): Likewise.
28758 (f32_lv): Likewise.
28759 (f32_sr): Likewise.
28760 (f32_sr2): Likewise.
28761 (f32_sm): Likewise.
28762 (f32_sm2): Likewise.
28763 (f32_si): Likewise.
28764 (f32_si2): Likewise.
28765 (f32_sv): Likewise.
28766 (f32_dm): Likewise.
28767 (f32_vsx): Likewise.
28768 (f32_av): Likewise.
28769 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28770 For movsf, order stores so the VSX stores occur before the GPR
28771 store which encourages the register allocator to use a traditional
28772 FPR instead of a GPR. For movsd, order the stores so that the GPR
28773 store comes before the VSX stores to allow the power6 to work.
28774 This is due to the power6 not having a 32-bit integer store
28775 instruction from a FPR.
28776 (movsf_hardfloat): Likewise.
28777 (movsd_hardfloat): Likewise.
28779 2017-05-09 Martin Sebor <msebor@redhat.com>
28781 PR translation/80280
28782 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28785 PR translation/80280
28786 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28787 data member added in r247778.
28788 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28790 2017-05-09 Nathan Sidwell <nathan@acm.org>
28792 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28794 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28797 2017-05-09 Marek Polacek <polacek@redhat.com>
28799 * doc/invoke.texi: Fix typo.
28801 2017-05-09 Richard Biener <rguenther@suse.de>
28803 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28804 (vrp_val_is_min): Likewise.
28805 (set_value_range_to_value): Likewise.
28806 (set_value_range_to_nonnegative): Likewise.
28807 (gimple_assign_nonzero_p): Likewise.
28808 (gimple_stmt_nonzero_p): Likewise.
28809 (vrp_int_const_binop): Likewise. Remove unreachable case.
28810 (adjust_range_with_scev): Adjust comments.
28811 (compare_range_with_value): Likewise.
28812 (extract_range_from_phi_node): Likewise.
28813 (test_for_singularity): Likewise.
28815 2017-05-09 Richard Biener <rguenther@suse.de>
28817 * tree-vrp.c (get_single_symbol): Add assert that we don't
28818 get overflowed constants as invariant part.
28819 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28820 checks. Use wi::cmp instead of recursing for integer constants.
28821 (compare_values): Just ignore whether we assumed undefined
28822 overflow instead of failing the compare.
28823 (extract_range_for_var_from_comparison_expr): Add comment before the
28824 TREE_NO_WARNING sets.
28825 (test_for_singularity): Likewise.
28826 (extract_range_from_comparison): Do not disable optimization
28827 when we assumed undefined overflow.
28828 (extract_range_basic): Remove init of unused var.
28830 2017-05-09 Richard Biener <rguenther@suse.de>
28832 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28833 (extract_range_from_multiplicative_op_1): Adjust.
28834 (extract_range_from_binary_expr_1): Use int_const_binop.
28836 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28839 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28840 rs6000_store_data_bypass_p in seven define_bypass directives and
28841 in several comments.
28842 * config/rs6000/rs6000-protos.h: Add prototype for
28843 rs6000_store_data_bypass_p function.
28844 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28845 function implements slightly different (rs6000-specific) semantics
28846 than store_data_bypass_p, returning false rather than aborting
28847 with assertion error when arguments do not satisfy the
28848 requirements of store data bypass.
28849 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28850 rs6000_store_data_bypass_p.
28852 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28854 * config/xtensa/xtensa-protos.h
28855 (xtensa_initial_elimination_offset): New declaration.
28856 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28857 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28858 macro definition, add case for FRAME_POINTER_REGNUM when
28859 FRAME_GROWS_DOWNWARD.
28860 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28861 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28862 xtensa_initial_elimination_offset.
28864 2017-05-08 Nathan Sidwell <nathan@acm.org>
28866 * doc/invoke.texi: Alphabetize -fdump options.
28868 2017-05-08 Martin Sebor <msebor@redhat.com>
28870 PR translation/80280
28871 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28873 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28875 * target.def (compute_frame_layout): New optional target hook.
28876 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28877 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28878 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28880 * reload1.c (verify_initial_elim_offsets): Likewise.
28881 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28882 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28883 (arm_get_frame_offsets): Split up into this ...
28884 (arm_compute_frame_layout): ... and this function.
28886 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28888 * config/aarch64/constraints.md (Usa): New constraint.
28889 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28891 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28893 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28894 with_multilib_list after it has been checked.
28896 2017-05-08 Richard Biener <rguenther@suse.de>
28898 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28899 (bitmap_set_subtract_values): Likewise.
28901 2017-05-08 Richard Biener <rguenther@suse.de>
28903 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28904 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28906 (gimple_stmt_nonzero_warnv_p): Rename to ...
28907 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28909 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28910 (extract_range_basic): Adjust, do not disable propagation on
28911 strict overflow sensitive simplification.
28912 (vrp_visit_cond_stmt): Likewise.
28914 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28916 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28917 body size unconditionally.
28919 2017-05-07 Jeff Law <law@redhat.com>
28922 2017-05-06 Jeff Law <law@redhat.com>
28923 PR tree-optimization/78496
28924 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28927 PR tree-optimization/78496
28928 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28929 (simplify_stmt_using_ranges): Call it.
28930 (vrp_dom_walker::before_dom_children): Extract equivalences
28931 from an ASSERT_EXPR with an equality comparison against a
28934 2017-05-06 Jeff Law <law@redhat.com>
28936 PR tree-optimization/78496
28937 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28940 PR tree-optimization/78496
28941 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28942 (simplify_stmt_using_ranges): Call it.
28943 (vrp_dom_walker::before_dom_children): Extract equivalences
28944 from an ASSERT_EXPR with an equality comparison against a
28947 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28949 * lra-constraints.c (lra_copy_reg_equiv): New function.
28950 (split_reg): Use it to copy equivalence information from the
28951 original register to the spill register.
28953 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28955 PR rtl-optimization/75964
28956 * simplify-rtx.c (simplify_const_relational_operation): Remove
28957 invalid handling of comparisons of integer ABS.
28959 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28961 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28962 initialize to zero.
28963 (init_regs): Remove declaration.
28964 (function_arg_advance_32): Initialize error_p as boolean variable.
28966 2017-05-05 Nathan Sidwell <nathan@acm.org>
28968 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28969 lines. Use for (;;).
28971 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28973 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28974 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28975 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28976 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28977 VF=2 that require versioning.
28979 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28981 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28984 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28986 * diagnostic.h (diagnostic_override_option_index): Convert from
28987 macro to inline function.
28989 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28991 * diagnostic.c (last_module_changed_p): New function.
28992 (set_last_module): New function.
28993 (diagnostic_report_current_module): Convert macro usage to
28994 the above functions.
28995 * diagnostic.h (diagnostic_context::last_module): Strengthen
28996 from const line_map * to const line_map_ordinary *.
28997 (diagnostic_last_module_changed): Delete macro.
28998 (diagnostic_set_last_module): Delete macro.
29000 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29002 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29003 with diagnostic_report_diagnostic.
29004 (diagnostic_n_impl_richloc): Likewise.
29005 * diagnostic.h (report_diagnostic): Delete macro.
29006 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29007 with diagnostic_report_diagnostic.
29008 * substring-locations.c (format_warning_va): Likewise.
29010 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29012 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29013 save/restor of format_spec. Move option-printing code to...
29014 (print_option_information): ...this new function, and
29015 reimplement by simply printing to the pretty_printer,
29016 rather than appending to the format string.
29018 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29020 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29021 handling logic into...
29022 (update_effective_level_from_pragmas): ...this new function.
29024 2017-05-04 Andrew Waterman <andrew@sifive.com>
29026 * config/riscv/riscv.opt (mstrict-align): New option.
29027 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29028 (SLOW_UNALIGNED_ACCESS): Define.
29029 (riscv_slow_unaligned_access): Declare.
29030 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29032 (riscv_slow_unaligned_access): New variable.
29033 (rocket_tune_info): Set slow_unaligned_access to true.
29034 (optimize_size_tune_info): Set slow_unaligned_access to false.
29035 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29036 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29037 (riscv_option_override): Set riscv_slow_unaligned_access.
29038 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29040 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29042 * config/riscv/riscv.md: Unify indentation.
29044 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29049 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29051 (extendsi<mode>2): Add support for doing sign extension via
29052 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29053 don't have ISA 3.0 instructions.
29054 (extendsi<mode>2 splitter): Likewise.
29055 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29056 generate the normal insns since SImode can now go in vector
29057 registers. Disallow the special UNSPECs needed for previous
29058 machines to hide SImode being used. Add new insns
29059 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29060 (fix_trunc<mode>si2_stfiwx): Likewise.
29061 (fix_trunc<mode>si2_internal): Likewise.
29062 (fixuns_trunc<mode>si2): Likewise.
29063 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29064 (fctiw<u>z_<mode>_smallint): Likewise.
29065 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29066 of floating point to 32-bit integer from doing a direct move to
29067 the GPR registers to do a store.
29068 (fctiwz_<mode>): Break long line.
29070 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29072 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29073 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29074 (addr_list, addr_offset_valid_p): New.
29075 (split_address_groups): Check offset validity with above function.
29076 (gt-tree-ssa-loop-ivopts.h): Include header file.
29078 2017-05-05 Nathan Sidwell <nathan@acm.org>
29080 * config.gcc (arm*-*-*): Add missing 'fi'.
29082 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29084 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29085 included in -fopt-info does not matter.
29086 * doc/optinfo.texi (-fopt-info): Fix description of default
29087 behavour. Explicitly say order of options included in -fopt-info
29090 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29092 * config.gcc: Allow combinations of aprofile and rmprofile values for
29093 --with-multilib-list.
29094 * config/arm/t-multilib: New file.
29095 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29096 variables. Remove setting of ISA and floating-point ABI in
29097 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29098 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29099 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29100 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29102 * config/arm/t-rmprofile: Likewise except for the matches changes.
29103 * doc/install.texi (--with-multilib-list): Document the combination of
29104 aprofile and rmprofile values and warn about pitfalls in doing that.
29106 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29108 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29109 (movdi_aarch64): Likewise.
29111 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29113 PR tree-optimization/80632
29114 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29116 (build_arrays): Initialize it for virtual phis.
29117 (fix_phi_nodes): Use it for virtual phis.
29119 PR tree-optimization/80558
29120 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29121 [x, y] op z into [x op, y op z] for op & or | if conditions
29124 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29125 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29128 * config/arm/arm.md (use_literal_pool): Remove.
29129 (64-bit immediate split): No longer takes cost into consideration
29130 if arm_disable_literal_pool is enabled.
29131 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29132 used when arm_disable_literal_pool is enabled.
29133 (arm_max_const_double_inline_cost): Remove use of
29134 arm_disable_literal_pool.
29135 (push_minipool_fix): Add assert.
29136 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29137 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29138 (no_literal_pool_sf_immediate): New.
29140 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29142 PR tree-optimization/80613
29143 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29144 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29146 2017-05-05 Richard Biener <rguenther@suse.de>
29148 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29150 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29152 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29153 of this flag from insn conditions due to removal from r247495.
29155 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29157 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29159 (arm_early_store_addr_dep_ptr): Likewise.
29160 * config/arm/aarch-common-protos.h
29161 (arm_early_load_addr_dep_ptr): Add prototype.
29162 (arm_early_store_addr_dep_ptr): Likewise.
29163 * config/arm/cortex-a53.md: Add new bypasses.
29165 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29167 * tree.c (next_type_uid): Change type to unsigned.
29168 (type_hash_canon): Decrement back next_type_uid if
29169 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29170 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29173 2017-05-04 Martin Sebor <msebor@redhat.com>
29175 * builtins.c: Fix a trivial typo in a comment.
29177 PR middle-end/79234
29178 * builtins.c (check_sizes): Adjust to handle reading past the end.
29179 Avoid printing excessive upper bound of ranges. Use %E to print
29180 tree nodes instead of converting them to %wu.
29181 (expand_builtin_memchr): New function.
29182 (compute_dest_size): Rename...
29183 (compute_objsize): ...to this.
29184 (expand_builtin_memcpy): Adjust.
29185 (expand_builtin_mempcpy): Adjust.
29186 (expand_builtin_strcat): Adjust.
29187 (expand_builtin_strcpy): Adjust.
29188 (check_strncat_sizes): Adjust.
29189 (expand_builtin_strncat): Adjust.
29190 (expand_builtin_strncpy): Adjust and simplify.
29191 (expand_builtin_memset): Adjust.
29192 (expand_builtin_bzero): Adjust.
29193 (expand_builtin_memcmp): Adjust.
29194 (expand_builtin): Handle memcmp.
29195 (maybe_emit_chk_warning): Check strncat just once.
29197 2017-05-04 Martin Sebor <msebor@redhat.com>
29199 PR preprocessor/79214
29200 PR middle-end/79222
29201 PR middle-end/79223
29202 * builtins.c (check_sizes): Add inlining context and issue
29203 warnings even when -Wno-system-headers is set.
29204 (check_strncat_sizes): Same.
29205 (expand_builtin_strncat): Same.
29206 (expand_builtin_memmove): New function.
29207 (expand_builtin_stpncpy): Same.
29208 (expand_builtin): Handle memmove and stpncpy.
29210 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29212 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29213 which is not used any more.
29215 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29217 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29219 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29221 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29222 (cortexa53_tunings): Likewise.
29223 (cortexa57_tunings): Likewise.
29224 (cortexa72_tunings): Likewise.
29225 (cortexa73_tunings): Likewise.
29227 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29229 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29230 Set loop alignment to 8.
29232 2017-05-04 Martin Sebor <msebor@redhat.com>
29234 PR translation/80280
29235 * builtins.c (expand_builtin_object_size): Add missing quoting to
29236 %D and like directives.
29237 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29238 (hsa_type_for_tree_type): Same.
29239 (verify_function_arguments): Same.
29240 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29241 * varasm.c (get_section): Same.
29244 2017-05-04 Martin Sebor <msebor@redhat.com>
29246 PR translation/80280
29247 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29249 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29251 * config/aarch64/aarch64.c (generic_addrcost_table):
29252 Change HI/TI mode setting.
29254 2017-05-04 Martin Jambor <mjambor@suse.cz>
29256 PR tree-optimization/80622
29257 * tree-sra.c (comes_initialized_p): New function.
29258 (build_accesses_from_assign): Only set write lazily when
29259 comes_initialized_p is false.
29260 (analyze_access_subtree): Use comes_initialized_p.
29261 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29262 instead of testing for PARM_DECL.
29264 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29266 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29267 constraint on operand 0 to allow more general addressing modes.
29268 Adjust output template.
29269 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29271 * config/aarch64/aarch64-protos.h
29272 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29273 * config/aarch64/constraints.md (Dp): New address constraint.
29274 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29277 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29279 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29280 update use of estimate_ipcp_clone_size_and_time.
29281 (estimate_local_effects): Update use of
29282 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29283 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29284 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29285 Return nonspecialized time.
29287 2017-05-04 Richard Biener <rguenther@suse.de>
29289 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29290 for the last VUSE which def dominates the PHI. Directly call
29292 (get_continuation_for_phi_1): Remove.
29294 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29296 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29297 to explain the use of truncating division. Cap the number of
29298 iterations to the maximum given by nb_iterations_upper_bound,
29301 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29303 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29304 * configure: Regenerate.
29305 * config.in: Regenerate.
29306 * config/i386/driver-mingw32.c: new file.
29307 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29308 * config.host: Link driver-mingw32.o on MinGW host.
29309 * doc/install.texi: Document new --enable-mingw-wildcard configure
29312 2017-05-04 Marek Polacek <polacek@redhat.com>
29314 PR tree-optimization/80612
29315 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29317 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29318 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29320 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29321 (movt splitter): Likewise.
29322 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29323 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29324 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29325 block for Thumb-1 with MOVT.
29326 (thumb2_legitimate_address_p): Move code block ...
29327 (can_avoid_literal_pool_for_label_p): ... into this new function.
29328 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29330 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29331 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29332 "M-profile targets with the MOVT instruction".
29334 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29336 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29337 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29338 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29340 2017-05-04 Martin Liska <mliska@suse.cz>
29342 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29343 variable cond_code.
29345 2017-05-04 Richard Biener <rguenther@suse.de>
29347 * tree.c (array_at_struct_end_p): Handle arrays at struct
29348 end with flexarrays more conservatively. Refactor and treat
29349 arrays of arrays or aggregates more strict. Fix
29350 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29351 * tree.c (array_at_struct_end_p): Adjust prototype.
29352 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29353 * gimple-fold.c (get_range_strlen): Likewise.
29354 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29356 2017-05-04 Richard Biener <rguenther@suse.de>
29358 PR tree-optimization/31130
29359 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29361 (supports_overflow_infinity): Likewise.
29362 (is_negative_overflow_infinity): Likewise.
29363 (is_positive_overflow_infinity): Likewise.
29364 (is_overflow_infinity): Likewise.
29365 (stmt_overflow_infinity): Likewise.
29366 (overflow_infinity_range_p): Likewise.
29367 (usable_range_p): Remove as always returning true.
29368 (make_overflow_infinity): Remove.
29369 (negative_overflow_infinity): Likewise.
29370 (positive_overflow_infinity): Likewise.
29371 (avoid_overflow_infinity): Likewise.
29372 (set_value_range): Adjust accordingly.
29373 (set_value_range_to_nonnegative): Likewise, remove now unused
29374 overflow_infinity arg.
29375 (vrp_operand_equal_p): Adjust.
29376 (update_value_range): Likewise.
29377 (range_int_cst_singleton_p): Likewise.
29378 (operand_less_p): Likewise.
29379 (compare_values_warnv): Likewise.
29380 (extract_range_for_var_from_comparison_expr): Likewise.
29381 (vrp_int_const_binop): Likewise.
29382 (zero_nonzero_bits_from_vr): Likewise.
29383 (extract_range_from_multiplicative_op_1): Likewise.
29384 (extract_range_from_binary_expr_1): Likewise.
29385 (extract_range_from_unary_expr): Likewise.
29386 (extract_range_from_comparison): Likewise.
29387 (extract_range_basic): Likewise.
29388 (adjust_range_with_scev): Likewise.
29389 (compare_ranges): Likewise.
29390 (compare_range_with_value): Likewise.
29391 (dump_value_range): Likewise.
29392 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29394 (simplify_cond_using_ranges): Adjust.
29396 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29398 * brig-builtins.def: Added a builtin for class_f64.
29399 * builtin-types.def: Added a builtin type needed by class_f64.
29401 2017-05-03 Jason Merrill <jason@redhat.com>
29403 * timevar.def: Add TV_CONSTEXPR.
29405 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29407 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29409 2017-05-03 Martin Jambor <mjambor@suse.cz>
29411 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29412 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29413 * ipa-cp.c (ipcp_propagate_stage): Do not call
29414 ipa_update_after_lto_read.
29415 * ipa-inline.c (ipa_inline): Likewise.
29417 2017-05-03 Martin Jambor <mjambor@suse.cz>
29419 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29420 tag. Added a default constructor and a destructor.
29421 (ipa_edge_args_sum_t): New class;
29422 (ipa_edge_args_sum): Declare.
29423 (ipa_edge_args_vector): Remove declaration.
29424 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29425 (ipa_free_edge_args_substructures): Remove declaration.
29426 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29427 (ipa_edge_args_info_available_for_edge_p): Likewise.
29428 * ipa-prop.c (ipa_edge_args_vector): Removed.
29429 (edge_removal_hook_holder): Likewise.
29430 (edge_duplication_hook_holder): Likewise.
29431 (ipa_edge_args_sum): New variable.
29432 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29433 ipa_edge_args_vector.
29434 (ipa_free_edge_args_substructures): Likewise.
29435 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29436 ipa_edge_args_vector.
29437 (ipa_edge_removal_hook): Turned into method
29438 ipa_edge_args_sum_t::remove.
29439 (ipa_edge_duplication_hook): Turned into method
29440 ipa_edge_args_sum_t::duplicate.
29441 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29442 registering edge hooks.
29443 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29444 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29445 ipa_edge_args_sum instead of ipa_edge_args_vector.
29446 * ipa-profile.c (ipa_profile): Likewise.
29448 2017-05-03 Martin Jambor <mjambor@suse.cz>
29450 * symbol-summary.h (function_summary): New method exists.
29451 (function_summary::symtab_removal): Deallocate through release.
29452 (call_summary): New class.
29453 (gt_ggc_mx): New overload.
29454 (gt_pch_nx): Likewise.
29455 (gt_pch_nx): Likewise.
29457 2017-05-03 Jeff Law <law@redhat.com>
29459 PR tree-optimization/78496
29460 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29461 from simplify_cond_using_ranges. Split off code to walk
29462 backwards through casts into ...
29463 (simplify_cond_using_ranges_2): New function.
29464 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29465 (execute_vrp): After identifying jump threads, call
29466 simplify_cond_using_ranges_2.
29468 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29471 * ipa-inline.h (inline_summary): Add ctor.
29472 (create_ggc): Do not use ggc_cleared_alloc.
29474 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29475 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29477 * gcc.c (handle_braces): Support escaping in switch matching
29479 * doc/invoke.texi (Spec Files): Document it.
29480 Remove superfluous @code markup in items.
29482 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29484 * diagnostic-show-locus.c (struct column_range): New struct.
29485 (get_affected_columns): New function.
29486 (get_printed_columns): New function.
29487 (struct correction): New struct.
29488 (correction::ensure_capacity): New function.
29489 (correction::ensure_terminated): New function.
29490 (struct line_corrections): New struct.
29491 (line_corrections::~line_corrections): New dtor.
29492 (line_corrections::add_hint): New function.
29493 (layout::print_trailing_fixits): Reimplement in terms of the new
29495 (selftest::test_overlapped_fixit_printing): New function.
29496 (selftest::diagnostic_show_locus_c_tests): Call it.
29498 2017-05-03 Nathan Sidwell <nathan@acm.org>
29500 Canonicalize canonical type hashing
29501 * tree.h (type_hash_canon_hash): Declare.
29502 * tree.c (type_hash_list, attribute_hash_list): Move into
29503 type_hash_canon_hash.
29504 (build_type_attribute_qual_variant): Break out hash code calc into
29505 type_hash_canon_hash.
29506 (type_hash_canon_hash): New. Generic type hash computation.
29507 (build_range_type_1, build_array_type_1, build_function_type,
29508 build_method_type_directly, build_offset_type, build_complex_type,
29509 make_vector_type): Call it.
29511 2017-05-03 Richard Biener <rguenther@suse.de>
29513 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29514 When all DRs have unknown misaligned do not always peel
29515 when there is a store but apply the same costing model as if
29516 there were only loads.
29518 2017-05-03 Richard Biener <rguenther@suse.de>
29521 PR tree-optimization/80492
29522 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29523 compare_base_decls returning dont-know properly.
29525 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29527 * config/arm/iterators.md (CCSI): New mode iterator.
29528 (arch): New mode attribute.
29529 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29530 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29531 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29532 code iterator for success result mode.
29533 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29534 the corresponding new insn generators.
29536 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29539 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29540 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29542 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29544 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29545 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29546 (DDR_A): Wrap DDR argument in brackets.
29547 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29548 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29549 (DDR_REVERSED_P): Likewise.
29551 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29553 PR tree-optimization/79472
29554 * tree-switch-conversion.c (struct switch_conv_info): Add
29555 contiguous_range and default_case_nonstandard fields.
29556 (collect_switch_conv_info): Compute contiguous_range and
29557 default_case_nonstandard fields, don't clear final_bb if
29558 contiguous_range and only the default case doesn't have the required
29560 (check_all_empty_except_final): Set default_case_nonstandard instead
29561 of failing if contiguous_range and the default case doesn't have empty
29563 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29564 and only the default case doesn't have the required constants. Skip
29566 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29567 if default_case_nonstandard.
29568 (build_constructors): Build constant 1 just once. Assert that default
29569 values aren't inserted in between cases if contiguous_range. Skip
29571 (build_arrays): Skip virtual phis.
29572 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29573 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29574 Handle virtual phis.
29575 (gen_inbound_check): Handle default_case_nonstandard case.
29576 (process_switch): Adjust check_final_bb caller. Call
29577 gather_default_values with the first non-default case instead of
29578 default case if default_case_nonstandard.
29580 2017-05-02 Nathan Sidwell <nathan@acm.org>
29582 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29583 check. Fix formatting.
29585 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29587 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29588 errors when comparing specialized and unspecialized times.
29590 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29592 * diagnostic-show-locus.c
29593 (layout::should_print_annotation_line_p): Make private.
29594 (layout::print_annotation_line): Make private.
29595 (layout::annotation_line_showed_range_p): Make private.
29596 (layout::show_ruler): Make private.
29597 (layout::print_source_line): Make private. Pass in line and
29598 line_width, rather than calling location_get_source_line. Drop
29600 (layout::print_leading_fixits): New method.
29601 (layout::print_any_fixits): Rename to...
29602 (layout::print_trailing_fixits): ...this, and make private.
29603 Don't print newline fixits.
29604 (diagnostic_show_locus): Move logic for printing one row into...
29605 (layout::print_line): ...this new function. Move the
29606 location_get_source_line call and error-handling from
29607 print_source_line to here. Call print_leading_fixits, and rename
29608 print_any_fixits to print_trailing_fixits.
29609 (selftest::test_fixit_insert_containing_newline): Update now that
29610 newlines are partially supported.
29611 (selftest::test_fixit_insert_containing_newline_2): New test.
29612 (selftest::test_fixit_replace_containing_newline): Update comments.
29613 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29614 * edit-context.c (class added_line): New class.
29615 (class edited_line): Describe newline handling in comment.
29616 (edited_line::actually_edited_p): New method.
29617 (edited_line::print_content): Delete redundant decl.
29618 (edited_line::m_predecessors): New field.
29619 (edited_file::print_content): Call edited_line::print_content.
29620 (edited_file::print_diff): Update to support newlines.
29621 (edited_file::print_diff_hunk): Likewise.
29622 (edited_file::print_run_of_changed_lines): New function.
29623 (edited_file::print_diff_line): Convert to...
29624 (print_diff_line): ...this.
29625 (edited_file::get_effective_line_count): New function.
29626 (edited_line::edited_line): Initialize new field m_predecessors.
29627 (edited_line::~edited_line): Clean up m_predecessors.
29628 (edited_line::apply_fixit): Handle newlines.
29629 (edited_line::get_effective_line_count): New function.
29630 (edited_line::print_content): New function.
29631 (edited_line::print_diff_lines): New function.
29632 (selftest::test_applying_fixits_insert_containing_newline): New
29634 (selftest::test_applying_fixits_replace_containing_newline): New
29636 (selftest::insert_line): New function.
29637 (selftest::test_applying_fixits_multiple_lines): Add example of
29639 (selftest::edit_context_c_tests): Call the new tests.
29641 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29643 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29644 parameter cand. Update dump information.
29645 (get_computation_cost): Update uses.
29647 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29649 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29650 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29651 (get_computation_at, rewrite_use_address): Update use of
29652 get_computation_aff.
29654 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29656 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29657 (get_computation): Delete.
29658 (get_computation_cost): Implement like get_computation_cost_at.
29659 Use get_computation_at.
29660 (get_computation_cost_at): Delete.
29661 (rewrite_use_nonlinear_expr): Use get_computation_at.
29662 (rewrite_use_compare, remove_unused_ivs): Ditto.
29664 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29666 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29668 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29670 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29671 (ivopts_global_cost_for_size): Rename parameter and update uses.
29672 (iv_ca_recount_cost): Update uses.
29673 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29674 candidates seperately in n_invs and n_cands.
29675 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29677 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29679 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29680 (find_inv_vars_cb): New.
29681 (find_depends): Renamed to ...
29682 (find_inv_vars): ... this.
29683 (add_candidate_1, force_var_cost): Call find_inv_vars.
29684 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29686 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29688 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29689 inv_vars. Add inv_exprs.
29690 (struct iv_cand): Rename depends_on to inv_vars.
29691 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29692 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29693 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29694 (dump_cand): Dump inv_vars.
29695 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29696 (record_invariant, find_depends, add_candidate_1): Ditto.
29697 (set_group_iv_cost, force_var_cost): Ditto.
29698 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29699 (get_computation_cost_at, get_computation_cost): Ditto.
29700 (determine_group_iv_cost_generic): Ditto.
29701 (determine_group_iv_cost_address): Ditto.
29702 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29703 (determine_group_iv_costs): Ditto.
29704 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29705 (iv_ca_set_remove_invariants): Renamed to ...
29706 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29707 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29708 (iv_ca_set_add_invariants): Renamed to ...
29709 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29710 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29711 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29712 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29713 (create_new_ivs): Remove useless dump.
29715 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29717 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29719 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29720 (iv_ca_set_no_cp, create_new_iv): Ditto.
29722 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29724 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29726 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29728 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29729 function tree_check2.
29731 2017-05-02 Martin Liska <mliska@suse.cz>
29733 * doc/gcov.texi: Add missing preposition.
29734 * gcov.c (function_info::function_info): Properly fill up
29735 all member variables.
29737 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29739 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29741 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29743 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29745 2017-05-02 Martin Liska <mliska@suse.cz>
29748 * lto-streamer-in.c (lto_read_tree_1): Remove
29749 LTO_STREAMER_DEBUG.
29750 * lto-streamer.c (struct tree_hash_entry): Likewise.
29751 (struct tree_entry_hasher): Likewise.
29752 (tree_entry_hasher::hash): Likewise.
29753 (tree_entry_hasher::equal): Likewise.
29754 (lto_streamer_init): Likewise.
29755 (lto_orig_address_map): Likewise.
29756 (lto_orig_address_get): Likewise.
29757 (lto_orig_address_remove): Likewise.
29758 * lto-streamer.h: Likewise.
29759 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29760 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29762 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29764 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29765 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29766 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29767 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29768 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29769 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29770 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29771 (mm_maskz_sub_ss): New intrinsics.
29772 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29773 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29774 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29775 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29776 (__builtin_ia32_subss_mask_round): New builtins.
29777 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29778 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29779 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29781 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29782 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29784 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29787 2017-05-02 Martin Jambor <mjambor@suse.cz>
29789 PR tree-optimization/78687
29790 * tree-sra.c (access): New field parent.
29791 (process_subtree_disqualification): New function.
29792 (disqualify_candidate): Call it.
29793 (build_accesses_from_assign): Reset write flag if creating an
29795 (build_access_subtree): Fill in parent field and also prpagate
29796 down grp_write flag.
29797 (create_artificial_child_access): New parameter set_grp_write, set
29798 grp_write to its value.
29799 (propagate_subaccesses_across_link): Also propagate grp_write flag
29801 (propagate_all_subaccesses): Push the closest parent back to work
29802 queue if add_access_to_work_queue returned true.
29804 2017-05-02 Richard Biener <rguenther@suse.de>
29806 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29807 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29808 -fstrict-overflow documentation.
29809 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29810 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29811 flag_strict_overflow.
29812 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29813 * lto-opts.c (lto_write_options): Do not stream it.
29814 * lto-wrapper.c (merge_and_complain): Do not handle it.
29815 * opts.c (default_options_table): Do not set -fstrict-overflow.
29816 (finish_options): Likewise do not clear it when sanitizing.
29817 * simplify-rtx.c (simplify_const_relational_operation): Do not
29818 test flag_strict_overflow.
29820 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29822 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29823 using enabled attribute.
29824 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29825 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29826 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29827 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29828 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29829 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29830 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29831 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29832 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29833 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29835 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29837 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29839 2017-05-02 Richard Biener <rguenther@suse.de>
29841 PR tree-optimization/80591
29843 2017-04-10 Richard Biener <rguenther@suse.de>
29845 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29848 2017-05-02 Richard Biener <rguenther@suse.de>
29850 PR tree-optimization/80549
29851 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29852 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29853 headers if they do not have a preheader.
29855 2017-05-02 Martin Liska <mliska@suse.cz>
29858 * common.opt: Fix typo.
29859 * doc/invoke.texi: Likewise.
29861 2017-05-01 Jan Beulich <jbeulich@suse.com>
29863 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29864 swapping, add (x,x,m,x,n) alternative.
29866 2017-05-01 Nathan Sidwell <nathan@acm.org>
29868 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29869 unnecessary unadjusted_alignment check.
29871 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29874 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29876 * gimplify.c (gimplify_cilk_detach): New function.
29877 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29878 * tree-core.h: Document EXPR_CILK_SPAWN.
29879 * tree.h (EXPR_CILK_SPAWN): Define.
29881 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29883 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29884 to use new fixit_hint representation, using the "replace" logic.
29885 (get_line_span_for_fixit_hint): Likewise.
29886 (layout::print_any_fixits): Likewise.
29887 (selftest::test_one_liner_many_fixits): Rename to...
29888 (selftest::test_one_liner_many_fixits_1): ...this, and update
29889 comment and expected output to reflect that the multiple fix-it
29890 hints are now consolidated into one insertion.
29891 (selftest::test_one_liner_many_fixits_2): New test.
29892 (selftest::test_diagnostic_show_locus_one_liner): Update for
29894 (selftest::test_fixit_consolidation): Update for fix-it API
29896 * diagnostic.c (print_parseable_fixits): Likewise.
29897 * edit-context.c (edited_line::m_line_events): Convert from
29898 auto_vec <line_event *> to auto_vec <line_event>.
29899 (class line_event): Convert from abstract base class to a concrete
29900 class, taking over the role of replace_event.
29901 (class insert_event): Delete.
29902 (class replace_event): Rename to class line_event. Convert to
29904 (edit_context::add_fixits): Reimplement.
29905 (edit_context::apply_insert): Delete.
29906 (edit_context::apply_replace): Rename to...
29907 (edit_context::apply_fixit): ...this. Convert to half-open range.
29908 (edited_file::apply_insert): Delete.
29909 (edited_file::apply_replace): Rename to...
29910 (edited_file::apply_fixit): ...this.
29911 (edited_line::~edited_line): Drop deletion of events.
29912 (edited_line::apply_insert): Delete.
29913 (edited_line::apply_replace): Rename to...
29914 (edited_line::apply_fixit): ...this. Convert to half-open range.
29915 Update for change to type of m_line_events.
29916 * edit-context.h (edit_context::apply_insert): Delete.
29917 (edit_context::apply_replace): Rename to...
29918 (edit_context::apply_fixit): ...this.
29920 2017-05-01 Martin Sebor <msebor@redhat.com>
29922 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29925 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29928 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29929 __get_cpuid_max returns 0.
29930 (__get_cpuid_count): Ditto.
29932 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29934 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29935 replacement expression is another instance of one of its arguments.
29937 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29940 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29941 check for stack push/pop autoinc.
29942 * config/i386/i386.c (ix86_agi_dependent): Return false
29943 if the only reason why modified_in_p returned true is that
29944 addr is SP based and set_insn is a push or pop.
29946 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29948 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29951 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29954 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29955 (account_size_time): Use two predicates - exec_pred and
29957 (evaluate_conditions_for_known_args): Compute both clause and
29959 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29960 (inline_summary_t::duplicate): Update.
29961 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29963 (compute_inline_parameters): Likewise.
29964 (estimate_edge_size_and_time): Update caluclation of time.
29965 (estimate_node_size_and_time): Compute both time and nonspecialized
29967 (estimate_ipcp_clone_size_and_time): Update.
29968 (inline_merge_summary): Update.
29969 (do_estimate_edge_time): Update.
29970 (do_estimate_edge_size): Update.
29971 (do_estimate_edge_hints): Update.
29972 (inline_read_section, inline_write_summary): Stream both new predicates.
29973 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29975 (compute_inlined_call_time): Cleanup.
29976 (big_speedup_p): Update.
29977 (edge_badness): Update.
29978 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29979 (size_time_entry): Replace predicate by exec_predicate and
29980 nonconst_predicate.
29981 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29982 (estimate_edge_time): Return also nonspec_time.
29983 (reset_edge_growth_cache): Update.
29985 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29987 PR rtl-optimization/80491
29988 * ifcvt.c (noce_process_if_block): When looking for x setter
29989 with missing else_bb, don't check only the insn right before
29990 cond_earliest, but look for the last insn that x is modified in
29991 within the same bb.
29993 PR rtl-optimization/80491
29994 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29996 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29998 PR tree-optimization/80487
29999 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30001 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30003 PR tree-optimization/79697
30004 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30005 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30006 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30008 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30009 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30011 2017-04-28 Martin Sebor <msebor@redhat.com>
30013 PR tree-optimization/80523
30014 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30015 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30017 (maybe_warn, format_directive, parse_directive): Use new functions.
30018 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30020 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30022 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30024 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30026 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30027 target_header_dir): Set correctly.
30028 * configure: Regenerated.
30029 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30030 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30031 instead of SYSTEM_HEADER_DIR.
30033 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30035 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30036 (estimate_local_effects): Likewise.
30037 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30038 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30039 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30040 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30041 do_estimate_edge_time, estimate_edge_time): Likewise.
30042 * ipa-inline-analysis.c (estimate_node_size_and_time,
30043 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30044 (estimate_time_after_inlining): Remove.
30046 2017-04-28 Martin Liska <mliska@suse.cz>
30048 * doc/gcov.texi: Enhance documentation of gcov.
30050 2017-04-28 Martin Liska <mliska@suse.cz>
30052 * doc/gcov.texi: Sort options in alphabetic order.
30053 * doc/gcov-dump.texi: Likewise.
30054 * doc/gcov-tool.texi: Likewise.
30055 * gcov.c (print_usage): Likewise.
30056 * gcov-dump.c (print_usage): Likewise.
30057 * gcov-tool.c (print_merge_usage_message): Likewise.
30058 (print_rewrite_usage_message): Likewise.
30059 (print_overlap_usage_message): Likewise.
30061 2017-04-28 Martin Liska <mliska@suse.cz>
30063 PR gcov-profile/53915
30064 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30066 2017-04-28 Martin Liska <mliska@suse.cz>
30068 PR gcov-profile/79891
30069 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30070 is marked by compiler as living on a line.
30071 (get_cycles_count): Remove usage of the union.
30072 (output_intermediate_file): Likewise.
30073 (find_source): Fix GNU coding style.
30074 (accumulate_line_counts): Remove old non-all block mode.
30075 (output_lines): Remove usage of the union.
30076 * profile.c (output_location): Include all BBs, even if
30077 belonging to a same line (and file) as a previous BB.
30079 2017-04-28 Martin Liska <mliska@suse.cz>
30081 * gcov.c (process_args): Handle new argument 'w'.
30082 (read_graph_file): Assign ID to BBs.
30083 (output_branch_count): Display BB # if verbose flag is set.
30084 (output_lines): Likewise for arcs.
30085 (print_usage): Add '--verbose' option help.
30086 * doc/gcov.texi: Document --verbose (-w) option.
30088 2017-04-28 Martin Liska <mliska@suse.cz>
30090 * gcov.c (struct block_location_info): New struct.
30091 (process_file): Fill up the new structure.
30092 (read_graph_file): Replace usage of encoding by the newly added
30094 (add_line_counts): Likewise.
30095 (accumulate_line_counts): Remove usage of the union.
30096 (function_info::function_info): New function.
30097 (function_info::~function_info): Likewise.
30098 (process_file): Call delete instead of release_function.
30099 (release_function): Release the function.
30100 (release_structures): Call delete instead of release_function.
30101 (solve_flow_graph): Replace usage of num_blocks.
30102 (find_exception_blocks): Likewise.
30103 (output_lines): Fix GNU coding style.
30105 2017-04-28 Martin Liska <mliska@suse.cz>
30108 * coverage.c (coverage_remove_note_file): New function.
30109 * coverage.h: Declare the function.
30110 * toplev.c (finalize): Clean if an error has been seen.
30112 2017-04-28 Martin Liska <mliska@suse.cz>
30114 PR gcov-profile/80031
30115 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30116 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30117 * gcov.c (read_graph_file): Read just number of blocks.
30118 * profile.c (branch_prob): Do not stream 0 flags per a basic
30121 2017-04-28 Martin Liska <mliska@suse.cz>
30123 * gcov-dump.c (tag_*): Add new argument to declarations.
30124 (dump_gcov_file): Likewise.
30125 (tag_blocks): Add and use new argument depth.
30126 (tag_arcs): Likewise.
30127 (tag_lines): Likewise.
30128 (tag_counters): Likewise.
30129 (tag_summary): Likewise.
30130 (dump_working_sets): Use depth to do a proper indentation.
30132 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30135 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30136 * symtab.c (symtab_node::debug_symtab): Move definition here.
30138 2017-04-28 Richard Biener <rguenther@suse.de>
30140 * lto-streamer.h (LTO_major_version): Bump to 7.
30142 2017-04-28 Richard Biener <rguenther@suse.de>
30144 * tree-vrp.c (assert_info): New struct.
30145 (add_assert_info): New helper.
30146 (register_edge_assert_for_2): Refactor to add asserts to a vector
30148 (register_edge_assert_for_1): Likewise.
30149 (register_edge_assert_for): Likewise.
30150 (finish_register_edge_assert_for): New helper actually registering
30151 asserts where live on edge.
30152 (find_conditional_asserts): Adjust.
30153 (find_switch_asserts): Likewise.
30154 (evrp_dom_walker::try_find_new_range): Generalize.
30155 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30157 2017-04-27 Marek Polacek <polacek@redhat.com>
30160 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30161 arg10 and arg11 to itype.
30163 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30165 * doc/extend.texi (Object Size Checking): Improve grammar.
30167 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30170 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30171 that the logic for permitting reciprocal estimates matches that
30174 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30177 * tree.c (type_cache_hasher::equal): Only compare
30178 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30179 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30180 non-aggregate element types.
30181 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30182 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30184 2017-04-27 Richard Biener <rguenther@suse.de>
30186 PR middle-end/80533
30187 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30188 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30189 keeping a reference to a trailing array.
30191 2017-04-27 Richard Biener <rguenther@suse.de>
30193 PR middle-end/80539
30194 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30195 being in loop-closed SSA form conservatively.
30196 (chrec_fold_multiply_poly_poly): Likewise.
30198 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30200 PR middle-end/79665
30201 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30202 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30204 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30207 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30208 (aarch64_function_arg_alignment): Return unsigned int again, but still
30209 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30210 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30211 Don't emit -Wpsabi note.
30212 (aarch64_function_arg_boundary): Likewise.
30213 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30216 2017-04-26 Nathan Sidwell <nathan@acm.org>
30218 * tree.h (crc32_unsigned_n): Declare.
30219 (crc32_unsigned, crc32_unsigned): Make inline.
30220 * tree.c (crc32_unsigned_bits): Replace with ...
30221 (crc32_unsigned_n): ... this.
30222 (crc32_unsigned, crc32_byte): Remove.
30223 (crc32_string): Remove unnecessary braces.
30225 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30227 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30228 * ipa-inline-analysis.c (MAX_TIME): Remove.
30229 (account_size_time): Use sreal for time.
30230 (dump_inline_summary): Update.
30231 (estimate_function_body_sizes): Update.
30232 (estimate_edge_size_and_time): Update.
30233 (estimate_calls_size_and_time): Update.
30234 (estimate_node_size_and_time): Update.
30235 (inline_merge_summary): Update.
30236 (inline_update_overall_summary): Update.
30237 (estimate_time_after_inlining): Update.
30238 (inline_read_section): Update.
30239 (inline_write_summary): Update.
30240 * ipa-inline.c (compute_uninlined_call_time): Update.
30241 (compute_inlined_call_time): Update.
30242 (recursive_inlining): Update.
30243 (inline_small_functions): Update.
30244 (dump_overall_stats): Update.
30245 * ipa-inline.h: Include sreal.h.
30246 (size_time_entry): Turn time to sreal.
30247 (inline_summary): Turn self_time nad time to sreal.
30249 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30251 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30253 (sreal::stream_out, sreal::stream_in): New.
30254 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30256 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30258 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30261 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30264 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30266 (dimode_scalar_chain::compute_convert_gain): Ditto.
30267 (dimode_scalar_chain::make_vector_copies): Ditto.
30268 (dimode_scalar_chain::convert_reg): Ditto.
30269 (dimode_scalar_chain::convert_insn): Ditto.
30270 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30271 (VI248_AVX512BW_1): New mode iterator.
30272 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30273 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30276 2017-04-25 Martin Sebor <msebor@redhat.com>
30278 PR tree-optimization/80497
30279 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30280 constants are representable in HOST_WIDE_INT.
30281 (parse_directive): Ditto.
30283 2017-04-25 Martin Sebor <msebor@redhat.com>
30286 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30287 (new_zero_array): Adjust signature.
30288 (dom_info::dom_init): Used unsigned rather that size_t.
30289 (dom_info::dom_info): Same.
30291 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30292 Jakub Jelinek <jakub@redhat.com>
30295 * config/arm/arm.c: Include gimple.h.
30296 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30297 returns negative, increment ncrn only if it returned positive.
30298 (arm_needs_doubleword_align): Return int instead of bool,
30299 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30300 members, but if there is any such non-FIELD_DECL
30301 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30302 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30303 returns negative, increment nregs only if it returned positive.
30304 (arm_setup_incoming_varargs): Likewise.
30305 (arm_function_arg_boundary): Emit -Wpsabi note if
30306 arm_needs_doubleword_align returns negative, return
30307 DOUBLEWORD_ALIGNMENT only if it returned positive.
30309 2017-04-25 Marek Polacek <polacek@redhat.com>
30312 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30313 first argument to type.
30315 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30318 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30319 type checks to test for compatibility instead of equality.
30321 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30322 Jakub Jelinek <jakub@redhat.com>
30325 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30327 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30328 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30329 the alignment computation, but return their maximum in warn_alignment.
30330 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30331 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30333 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30334 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30337 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30339 * config/arc/simdext.md (dmpyh): Fix typo.
30341 2017-04-25 Richard Biener <rguenther@suse.de>
30343 PR tree-optimization/80492
30344 * alias.c (compare_base_decls): Handle registers with asm
30345 specification conservatively.
30346 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30347 compare_base_decls returning dont-know properly.
30349 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30351 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30352 (legitimate_offset_address_p): New function.
30353 (arc_legitimate_address_p): Use above function.
30355 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30357 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30359 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30361 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30362 ACCH registers whenever they are available.
30364 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30366 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30367 double regs fix when not used.
30369 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30371 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30373 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30374 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30376 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30378 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30379 when emitting short ADD instructions.
30381 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30383 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30385 (cmpsi_cc_c_insn): Likewise.
30386 (cbranchsi4_scratch): Compute proper instruction length using
30387 compact_hreg_operand.
30388 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30390 2017-04-25 Richard Biener <rguenther@suse.de>
30392 PR middle-end/80509
30393 * passes.c (pass_manager::pass_manager): Initialize
30394 m_name_to_pass_map.
30396 2017-04-25 Richard Biener <rguenther@suse.de>
30398 PR tree-optimization/79201
30399 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30401 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30404 * config/s390/vector.md: Split MEM->GPR vector moves for
30405 non-s_operand addresses.
30407 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30410 * config/s390/predicates.md (reload_const_wide_int_operand): New
30412 * config/s390/s390.md ("movti"): Remove d/P alternative.
30413 ("movti_bigconst"): New pattern definition.
30415 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30418 * s390-protos.h (s390_expand_cs_hqi): Removed.
30419 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30420 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30421 modes as well as CCZ1mode and CCZmode.
30422 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30423 signature of s390_emit_compare_and_swap.
30424 (s390_expand_cs_hqi): Likewise, make static.
30425 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30426 compare-and-swap, in some cases.
30427 (s390_expand_cs): Wrapper function.
30428 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30430 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30431 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30432 patterns for small and large integers. Forbid symref memory operands.
30433 Move expander to s390.c. Require cc register.
30434 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30435 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30436 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30437 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30438 symref memory operands. Remove CC mode and call s390_match_ccmode
30440 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30442 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30444 * config/s390/s390.md (define_peephole2): New peephole to help
30445 combining the load-and-test pattern with volatile memory.
30447 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30449 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30450 with CCZmode for TARGET_Z196.
30452 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30454 PR rtl-optimization/80501
30455 * combine.c (make_compound_operation_int): Set subreg_code to SET
30456 even for AND with mask of the sign bit of mode.
30458 PR rtl-optimization/80500
30459 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30460 sum's initial value.
30462 2017-04-25 Julian Brown <julian@codesourcery.com>
30463 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30465 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30467 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30469 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30471 2017-04-25 Julian Brown <julian@codesourcery.com>
30472 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30474 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30475 (thunderx2t99_sha): New Reservation.
30477 2017-04-25 Julian Brown <julian@codesourcery.com>
30478 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30480 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30481 type for 1-element load.
30483 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30485 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30487 2017-04-24 Martin Jambor <mjambor@suse.cz>
30489 PR tree-optimization/80293
30490 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30491 char arrays not totally scalarizable if it is false.
30492 (analyze_all_variable_accesses): Pass correct value in the new
30493 parameter. Add a statistics counter.
30495 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30497 PR middle-end/79931
30498 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30500 2017-04-24 Richard Biener <rguenther@suse.de>
30502 PR tree-optimization/80494
30503 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30504 out for complex types.
30506 2017-04-24 Richard Biener <rguenther@suse.de>
30508 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30509 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30510 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30511 (DFS): Adjust and never fail.
30512 (sccvn_dom_walker::fail): Remove.
30513 (sccvn_dom_walker::before_dom_children): Adjust.
30514 (run_scc_vn): Likewise and never fail.
30515 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30516 (pass_fre::execute): Likewise.
30518 2017-04-24 Richard Biener <rguenther@suse.de>
30520 PR tree-optimization/79725
30521 * tree-ssa-sink.c (statement_sink_location): Return whether
30522 failure reason was zero uses. Move that check later.
30523 (sink_code_in_bb): Deal with zero uses by removing the stmt
30526 2017-04-24 Richard Biener <rguenther@suse.de>
30529 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30530 pointer-based references.
30532 2017-04-24 Richard Biener <rguenther@suse.de>
30535 * pass_manager.h (pass_manager::operator new): Remove.
30536 (pass_manager::operator delete): Likewise.
30537 * passes.c (pass_manager::operator new): Remove.
30538 (pass_manager::operator delete): Likewise.
30539 (pass_manager::pass_manager): Zero individual pass members.
30541 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30544 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30545 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30546 Check "XEXP (src, 1)" operand here.
30547 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30548 Check "XEXP (src, 1)" operand here.
30549 (dimode_scalar_chain::make_vector_copies): Detect count register
30550 of a shift instruction. Zero extend count register from QImode
30551 to DImode to satisfy vector shift pattern count operand predicate.
30552 Substitute vector shift count operand with a DImode copy.
30553 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30556 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30558 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30559 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30560 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30561 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30562 (UNSPEC_NOREX_MEM): Remove definition.
30564 2017-04-21 Richard Biener <rguenther@suse.de>
30566 PR tree-optimization/79547
30567 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30568 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30569 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30570 without any constraints.
30572 2017-04-21 Richard Biener <rguenther@suse.de>
30574 PR tree-optimization/78847
30575 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30577 2017-04-21 Richard Biener <rguenther@suse.de>
30579 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30580 (build_distinct_type_copy): Likewise.
30581 (build_variant_type_copy): Likewise.
30582 * tree.c (build_qualified_type): Pass down mem-stat info.
30583 (build_distinct_type_copy): Likewise.
30584 (build_variant_type_copy): Likewise.
30586 2017-04-21 Richard Biener <rguenther@suse.de>
30588 PR tree-optimization/80237
30589 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30591 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30592 for a simplified result.
30594 2016-04-21 Richard Biener <rguenther@suse.de>
30596 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30597 sth as strict as a simple_iv but a chrec without symbols and an
30598 operand defined in the loop we are peeling (and not some subloop).
30599 (propagate_constants_for_unrolling): Propagate all constants.
30601 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30604 * config/i386/i386.c (print_reg): Remove assert for disalowed
30605 regno values, call output_operand_lossage instead.
30607 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30610 * config/i386/constraints.md (Yc): New register constraint.
30611 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30612 Use Yc constraint for alternative 2 of operand 0. Remove
30613 preferred_for_speed attribute.
30615 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30617 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30618 lastprivate clauses in SIMT case.
30620 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30622 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30624 2017-04-20 Richard Biener <rguenther@suse.de>
30626 PR tree-optimization/57796
30627 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30628 as N scalar stores.
30629 (vect_model_load_cost): Cost gathers as N scalar loads.
30631 2017-04-20 Richard Biener <rguenther@suse.de>
30633 * ggc-page.c (ggc_allocated_p): Rename to ...
30634 (safe_lookup_page_table_entry): ... this and return the lookup
30636 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30638 2017-04-20 Richard Biener <rguenther@suse.de>
30640 PR tree-optimization/80453
30641 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30642 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30643 from the conditions.
30644 (vn_phi_eq): Pass them down.
30645 (vn_phi_lookup): Record them.
30646 (vn_phi_insert): Likewise.
30648 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30650 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30651 uninitialized variable warning to avoid buffer overrun.
30653 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30656 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30657 is suppressed for '{ 0 }' in C.
30659 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30661 * BASE-VER: Set to 8.0.0.
30663 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30665 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30666 priority .init_array and .fini_array section with SECTION_NOTYPE
30669 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30671 PR middle-end/80423
30672 * tree.h (build_array_type): Add typeless_storage default argument.
30673 * tree.c (type_cache_hasher::equal): Also compare
30674 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30675 (build_array_type): Add typeless_storage argument, set
30676 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30678 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30679 (build_array_type): Likewise. Add typeless_storage argument.
30681 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30682 Jakub Jelinek <jakub@redhat.com>
30684 PR tree-optimization/80426
30685 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30686 operation on symbolic operands, also compute the overflow for the
30687 invariant part when the operation degenerates into a negation.
30689 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30692 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30693 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30696 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30698 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30701 * config/avr/avr.c (tree.h): Include it.
30702 (cgraph.h): Include it.
30703 (avr_encode_section_info): Don't warn for uninitialized progmem
30704 variable if it's just an alias.
30706 2017-04-19 Richard Biener <rguenther@suse.de>
30709 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30710 when needed by AutoPGO.
30712 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30715 * doc/lto.texi: Remove an extra 'that'.
30717 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30719 PR rtl-optimization/80429
30720 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30721 are only used in debug insns.
30723 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30724 Vladimir Makarov <vmakarov@redhat.com>
30726 * config/sparc/predicates.md (input_operand): Add comment. Return
30727 true for any memory operand when LRA is in progress.
30728 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30730 2017-04-18 Jeff Law <law@redhat.com>
30733 * mips.md ({return,simple_return}_internal): Do not overwrite
30736 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30738 PR tree-optimization/80443
30739 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30740 instead of adding 1, subtract -1 and similarly instead of subtracting
30743 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30745 PR rtl-optimization/80357
30746 * haifa-sched.c (tmp_bitmap): New variable.
30747 (model_recompute): Handle duplicate use records.
30748 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30749 (free_global_sched_pressure_data): Free it.
30751 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30754 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30755 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30756 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30757 instead of SYSTEM_HEADER_DIR.
30759 2017-04-18 Jeff Law <law@redhat.com>
30761 PR middle-end/80422
30762 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30763 predecessors after walking up the insn chain.
30765 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30768 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30769 sizetype type into debug info.
30771 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30774 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30775 unneeded test for TARGET_UPPER_REGS_SF.
30776 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30778 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30781 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30782 instead of gsi_after_labels.
30784 2017-04-18 Jeff Law <law@redhat.com>
30786 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30790 2017-04-13 Jeff Law <law@redhat.com>
30791 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30792 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30794 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30797 * config/avr/avr.c (intl.h): Include it.
30798 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30800 2017-04-18 Martin Liska <mliska@suse.cz>
30802 PR gcov-profile/78783
30803 * gcov-tool.c (gcov_output_files): Validate that destination
30804 file is either removed by the tool or by a user.
30806 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30807 Guy Benyei <guybe@mellanox.com>
30809 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30810 block, and do not negate it, the stored id is already negative.
30812 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30814 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30816 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30819 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30820 masks of options that should be turned off if the VSX vector
30821 options are turned off.
30822 (OTHER_P8_VECTOR_MASKS): Likewise.
30823 (OTHER_VSX_VECTOR_MASKS): Likewise.
30824 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30825 rs6000_disable_incompatible_switches to validate no type switches
30827 (rs6000_incompatible_switch): New function to disallow turning on
30828 other vector options if -mno-vsx, -mno-power8-vector, or
30829 -mno-power9-vector are specified.
30831 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30833 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30835 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30837 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30838 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30839 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30840 (ARG_POINTER_CFA_OFFSET): Likewise.
30842 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30844 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30845 conditions to take advantage of various optimizations.
30847 2017-04-13 Jeff Law <law@redhat.com>
30849 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30850 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30851 (zero_extendsidi2_dext): Likewise.
30853 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30856 * fold-const.c (fold_ternary_loc): Revert
30857 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30860 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30862 PR rtl-optimization/80343
30863 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30864 new scratch pseudo.
30866 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30869 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30870 to ubsan_encode_value.
30872 2017-04-13 Jeff Law <law@redhat.com>
30874 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30875 appearing in DEBUG_INSNs.
30877 2017-04-13 Martin Liska <mliska@suse.cz>
30879 PR gcov-profile/80413
30880 * gcov-io.c (gcov_write_string): Copy to buffer just when
30881 allocated size is greater than zero.
30883 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30886 * dwarf2out.c (decls_for_scope): Ignore declarations of
30887 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30889 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30892 * ipa-visibility.c (non_local_p): Fix typos.
30893 (localize_node): When localizing symbol in same comdat group,
30894 dissolve the group only when we know external symbols are going
30896 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30898 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30900 PR tree-optimization/79390
30901 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30902 order does not result in usable sequence, retry with reversed operand
30908 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30909 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30910 op0 instead of fold_convert_loc (loc, type, arg0).
30912 2017-04-12 Jeff Law <law@redhat.com>
30914 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30915 has a delay slot in the generated code.
30917 * config/cris/cris.md (cris_preferred_reload_class): Return
30918 GENNONACR_REGS rather than GENERAL_REGS.
30920 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30923 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30924 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30925 signedness of the result type.
30927 2017-04-12 Richard Biener <rguenther@suse.de>
30928 Jeff Law <law@redhat.com>
30930 PR tree-optimization/80359
30931 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30932 trim stores to TARGET_MEM_REFs.
30934 2017-04-12 Richard Biener <rguenther@suse.de>
30936 PR tree-optimization/79390
30937 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30938 threading case even more.
30940 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30943 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30944 for quad_address_p for TImode, instead of just not indexed_address.
30946 2017-04-12 Richard Biener <rguenther@suse.de>
30947 Bernd Edlinger <bernd.edlinger@hotmail.de>
30949 PR middle-end/79671
30950 * alias.c (component_uses_parent_alias_set_from): Handle
30951 TYPE_TYPELESS_STORAGE.
30952 (get_alias_set): Likewise.
30953 * tree-core.h (tree_type_common): Add typeless_storage flag.
30954 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30955 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30956 for types containing members with TYPE_TYPELESS_STORAGE.
30957 (place_field): Likewise.
30958 (layout_type): Likewise for ARRAY_TYPE.
30959 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30960 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30961 TYPE_TYPELESS_STORAGE.
30962 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30964 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30967 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30968 first argument to type.
30970 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30974 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30975 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30976 (rs6000_expand_binop_builtin): Likewise.
30977 (rs6000_expand_ternop_builtin): Likewise; also add missing
30978 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30979 vshasigma built-ins.
30980 * doc/extend.texi: Document that vec_xxpermdi's third argument
30981 must be a constant.
30983 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30985 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30986 Use shift_const cost parameter when calculating gain of STV shifts.
30988 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30990 PR rtl-optimization/70478
30991 * lra-constraints.c (process_alt_operands): Check memory for
30992 disfavoring memory insn operand.
30994 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30996 PR middle-end/80100
30997 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30998 left shift in unsigned HOST_WIDE_INT type.
31000 PR rtl-optimization/80385
31001 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31002 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31005 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31006 if they have any depend clauses.
31008 2017-04-11 Martin Liska <mliska@suse.cz>
31011 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31012 * ipa-split.c (split_function): Create a local comdat symbol
31013 if caller is in a comdat group.
31015 2017-04-11 Martin Liska <mliska@suse.cz>
31018 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31021 2017-04-11 Martin Sebor <msebor@redhat.com>
31023 PR middle-end/80364
31024 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31025 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31027 (directive::set_width, directive::set_precision, format_character):
31029 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31032 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31035 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31036 conflict, set target->arch_name instead of target->cpu_name.
31038 2017-04-11 Richard Biener <rguenther@suse.de>
31040 PR tree-optimization/80374
31041 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31042 build_zero_cst, remove fold_convertible_p check again.
31044 2017-04-11 Martin Liska <mliska@suse.cz>
31047 * ubsan.c (instrument_object_size): Do not instrument register
31050 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31053 * config/i386/i386-builtin-types.def
31054 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31055 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31056 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31057 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31058 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31059 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31060 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31061 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31062 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31063 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31064 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31065 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31066 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31067 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31068 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31069 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31070 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31071 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31072 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31073 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31074 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31075 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31076 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31077 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31078 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31079 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31080 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31081 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31082 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31083 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31084 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31085 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31086 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31087 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31088 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31089 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31090 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31091 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31092 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31093 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31094 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31095 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31096 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31097 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31098 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31099 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31101 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31102 flag to second_arg_count, handle 4 argument function type _COUNT
31103 aliases, handle second_arg_count on second argument rather than last.
31105 2017-04-10 Jeff Law <law@redhat.com>
31107 PR tree-optimization/80374
31108 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31109 record anything if we can not convert integer_zero_node to the
31112 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31115 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31116 Enhance special handling given to the TARGET_P9_MINMAX option in
31117 relation to certain other options.
31119 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31121 PR tree-optimization/80153
31122 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31123 remove POINTER_PLUS_EXPR's base part directly, rather than through
31126 2017-04-10 Richard Biener <rguenther@suse.de>
31127 Bin Cheng <bin.cheng@arm.com>
31129 PR tree-optimization/80153
31130 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31131 the first element of pointer type aff_tree. Build result expr in
31133 (add_elt_to_tree): Convert to type unconditionally. Remove other
31134 fold_convert calls.
31135 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31136 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31138 2017-04-10 Richard Biener <rguenther@suse.de>
31140 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31143 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31145 PR rtl-optimization/70478
31146 * lra-constraints.c (curr_small_class_check): New.
31147 (update_and_check_small_class_inputs): New.
31148 (process_alt_operands): Update curr_small_class_check. Disfavor
31149 alternative insn memory operands. Check available regs for small
31152 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31155 * config/mips/mips.opt (-mvirt): Update description.
31156 * doc/invoke.texi (-mvirt): Likewise.
31158 2017-04-10 Richard Biener <rguenther@suse.de>
31160 PR middle-end/80362
31161 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31162 looking for NEGATE_EXPR in -A / -B to A / B folding.
31164 2017-04-10 Martin Liska <mliska@suse.cz>
31166 PR gcov-profile/80224
31167 * gcov.c (print_usage): Fix usage string.
31168 (get_gcov_intermediate_filename): Remove.
31169 (output_gcov_file): Use both for normal and intermediate format.
31170 (generate_results): Do not initialize special file for
31171 intermediate format.
31173 2017-04-10 Richard Biener <rguenther@suse.de>
31175 PR tree-optimization/80304
31176 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31179 2017-04-10 Nathan Sidwell <nathan@acm.org>
31182 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31183 (rs6000_init_builtins): Use it.
31185 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31187 * config/arm/arm.md (<mrc>): Add mode to SET source.
31188 (<mrrc>): Likewise.
31190 2017-04-10 Richard Biener <rguenther@suse.de>
31192 PR middle-end/80344
31193 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31195 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31198 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31199 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31200 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31201 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31202 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31203 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31204 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31205 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31206 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31207 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31208 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31209 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31210 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31211 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31212 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31213 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31214 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31215 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31216 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31217 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31218 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31219 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31220 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31222 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31224 PR rtl-optimization/70478
31225 * lra-constraints.c: Reverse the last patch.
31227 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31229 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31230 Add comment for WCHAR_T.
31232 2017-04-08 Martin Liska <mliska@suse.cz>
31235 2017-04-07 Martin Liska <mliska@suse.cz>
31238 * ipa-split.c (split_function): Add function part to a same comdat
31241 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31244 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31246 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31248 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31249 (vec_pairing): ...to this.
31250 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31251 (rs6000_sched_init): Adjust for name change.
31252 (struct rs6000_sched_context): Likewise.
31253 (rs6000_init_sched_context): Likewise.
31254 (rs6000_set_sched_context): Likewise.
31256 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31262 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31264 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31265 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31266 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31268 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31270 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31272 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31274 PR rtl-optimization/70703
31275 * ira-color.c (update_conflict_hard_regno_costs): Use
31276 int64_t instead of HOST_WIDE_INT.
31278 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31280 PR rtl-optimization/70478
31281 * lra-constraints.c (process_alt_operands): Disfavor alternative
31282 insn memory operands.
31284 2017-04-07 Jeff Law <law@redhat.com>
31286 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31287 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31289 2017-04-07 Martin Liska <mliska@suse.cz>
31292 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31293 Show error message instead of an ICE.
31295 2017-04-07 Martin Liska <mliska@suse.cz>
31298 * ipa-split.c (split_function): Add function part to a same comdat
31301 2017-04-07 Richard Biener <rguenther@suse.de>
31303 PR middle-end/80341
31304 * tree.c (get_unwidened): Also handle ! for_type case for
31306 * convert.c (do_narrow): Split out from ...
31307 (convert_to_integer_1): ... here. Do not pass final truncation
31308 type to get_unwidened for TRUNC_DIV_EXPR.
31310 2017-04-07 Richard Biener <rguenther@suse.de>
31312 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31314 (aff_combination_const): Adjust.
31315 (aff_combination_scale): Likewise.
31316 (aff_combination_add_elt): Likewise.
31317 (aff_combination_add_cst): Likewise.
31318 (aff_combination_convert): Likewise.
31319 (add_elt_to_tree): Likewise. Remove unused argument.
31320 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31322 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31324 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31326 * config/arm/arm.c (arm_default_short_enums): Use
31327 ARM_DEFAULT_SHORT_ENUMS.
31328 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31330 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31333 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31334 members with redundant out-of-class redeclaration.
31336 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31339 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31340 * config/i386/i386.md (*zero_extendsidi2):
31341 Add (?*x,*x) and (?*v,*v) alternatives.
31343 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31346 * config/i386/i386.c (ix86_expand_builtin)
31347 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31348 mode from insn data. Convert operands to insn operand mode.
31349 Copy operands that don't satisfy insn predicate to a register.
31351 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31353 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31356 2017-04-06 Richard Biener <rguenther@suse.de>
31358 PR tree-optimization/80334
31359 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31360 preserve alignment of accesses.
31362 2017-04-06 Richard Biener <rguenther@suse.de>
31364 PR tree-optimization/80262
31365 * tree-sra.c (build_ref_for_offset): Preserve address-space
31367 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31368 Drop useless address-space information on MEM_REF offsets.
31370 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31372 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31374 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31376 PR rtl-optimization/70703
31377 * ira-color.c (update_conflict_hard_regno_costs): Use
31378 HOST_WIDE_INT instead of long.
31380 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31383 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31384 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31386 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31387 for x86_64 target. Handle -m3dnowa option.
31389 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31391 PR rtl-optimization/70703
31392 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31393 (update_conflict_hard_regno_costs): Use long instead of unsigned
31394 arithmetic for cost calculation.
31396 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31397 Bernd Edlinger <bernd.edlinger@hotmail.de>
31400 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31403 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31406 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31407 ptr_mode with Pmode throughout.
31408 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31409 into probe_stack_range and use DImode.
31411 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31414 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31415 call_eh_return is true.
31417 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31419 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31420 Initialize last_match_fntype_index.
31422 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31425 * tree-nvr.c: Include internal-fn.h.
31426 (pass_return_slot::execute): Ignore internal calls without
31429 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31430 Richard Biener <rguenther@suse.de>
31433 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31434 captures used multiple times, except for the last use.
31435 * generic-match-head.c: Include gimplify.h.
31437 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31439 PR tree-optimization/79390
31440 * target.h (struct noce_if_info): Declare.
31441 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31442 * target.def (noce_conversion_profitable_p): New target hook.
31443 * ifcvt.h (struct noce_if_info): New type, moved from ...
31444 * ifcvt.c (struct noce_if_info): ... here.
31445 (noce_conversion_profitable_p): Renamed to ...
31446 (default_noce_conversion_profitable_p): ... this. No longer
31448 (noce_try_store_flag_constants, noce_try_addcc,
31449 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31450 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31451 instead of noce_conversion_profitable_p.
31452 * config/i386/i386.c: Include ifcvt.h.
31453 (ix86_option_override_internal): Don't override
31454 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31455 (ix86_noce_conversion_profitable_p): New function.
31456 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31457 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31458 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31459 * doc/tm.texi: Regenerated.
31461 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31463 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31466 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31469 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31470 instructions for small multiply cores.
31472 2017-04-04 Jeff Law <law@redhat.com>
31474 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31476 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31477 that are not set by the loop over the elements.
31479 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31482 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31483 int mode, convert_modes it to mode as unsigned, otherwise use
31484 lowpart_subreg to mode rather than SImode.
31485 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31486 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31487 Use DImode instead of SImode for the shift count operand.
31488 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31491 2017-04-04 Richard Biener <rguenther@suse.de>
31493 PR middle-end/80281
31494 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31495 arithmetic done for the negate or the plus. Simplify.
31496 (A - (-B) -> A + B): Likewise.
31497 * fold-const.c (split_tree): Make sure to not negate pointers.
31499 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31501 PR rtl-optimization/60818
31502 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31503 a compare of comparisons with the thing compared if this results
31504 in a different machine mode.
31506 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31508 * alias.c (base_alias_check): Fix typo in comment.
31509 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31510 * cgraphunit.c (symbol_table::compile): Likewise.
31511 * collect2.c (maybe_run_lto_and_relink): Likewise.
31512 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31513 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31514 * config/avr/avr.c (avr_map_op_t): Likewise.
31515 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31516 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31517 * config/epiphany/epiphany.md (movcc): Likewise.
31518 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31519 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31521 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31522 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31523 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31524 * config/sh/sh.c (sh_rtx_costs): Likewise.
31525 * fold-const.c (fold_truth_andor): Likewise.
31526 * genautomata.c (collapse_flag): Likewise.
31527 * gengtype.h (struct type::u::s): Likewise.
31528 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31529 * input.c (FORMAT_AMOUNT): Likewise.
31530 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31531 (known_aggs_to_agg_replacement_list): Likewise.
31532 * ipa-inline-analysis.c: Likewise.
31533 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31534 * ipa-polymorphic-call.c
31535 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31536 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31537 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31539 * modulo-sched.c (apply_reg_moves): Likewise.
31540 * omp-expand.c (build_omp_regions_1): Likewise.
31541 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31542 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31543 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31544 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31545 * value-prof.c: Likewise.
31546 * var-tracking.c (val_reset): Likewise.
31548 2017-04-03 Richard Biener <rguenther@suse.de>
31550 PR tree-optimization/80275
31551 * fold-const.c (split_address_to_core_and_offset): Handle
31554 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31556 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31557 descriptors is at least equal to that of functions.
31559 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31561 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31563 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31566 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31567 (mov<IMOD4:mode>): New expander.
31568 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31570 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31572 PR rtl-optimization/79405
31573 * fwprop.c (propagations_left): New variable.
31574 (forward_propagate_into): Decrement it.
31575 (fwprop_init): Initialize it.
31576 (fw_prop): If the variable has reached zero, stop propagating.
31577 (fwprop_addr): Ditto.
31579 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31582 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31583 a FUNCTION_DECL, pass it as decl instead of origin to
31586 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31588 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31591 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31594 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31595 TARGET_VSX_SMALL_INTEGER.
31597 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31599 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31600 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31602 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31604 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31605 extraction from odd-numbered MSA register.
31607 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31609 PR middle-end/80173
31610 * expmed.c (store_bit_field_1): Don't attempt to create
31611 a word subreg out of hard registers wider than word if they
31612 have HARD_REGNO_NREGS of 1 for their mode.
31614 PR middle-end/80163
31615 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31616 conversions to integer types wider than word and pointer.
31619 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31620 (rtx_equal_for_cselib_p): Pass 0 to it.
31621 * cselib.c (cselib_hasher::equal): Likewise.
31622 (rtx_equal_for_cselib_1): Add depth argument. If depth
31623 is 128, don't look up VALUE locs and punt. Increment
31624 depth in recursive calls when walking VALUE locs.
31626 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31628 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31629 (make_gcov_file_name): Use the canonical path name for generating
31631 (read_line): Fix handling of files with ascii null bytes.
31633 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31635 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31636 to initialise a vector register instead
31637 of using a const_int.
31639 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31641 PR translation/80189
31642 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31643 diagnostic messages.
31645 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31648 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31649 (dfp_diex_<mode>): Update mode of operand 1.
31650 * doc/extend.texi (dxex, dxexq): Document change to return type.
31651 (diex, diexq): Document change to argument type.
31653 2017-03-30 Martin Jambor <mjambor@suse.cz>
31656 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31657 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31658 it reflects the signature changes performed at the callee side.
31659 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31660 to cgraph_build_function_type_skip_args.
31661 (build_function_decl_skip_args): Adjust call to the above function.
31663 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31666 * config/i386/sse.md
31667 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31668 register as dest whenever it is a MEM not rtx_equal_p to the
31669 corresponding dup operand, and when forcing into reg move the
31670 reg into the memory afterwards.
31671 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31672 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31673 for the force_reg mode.
31674 (avx512vl_vextractf128<mode>): Use register as dest either
31675 always when a MEM, or when it is a MEM not rtx_equal_p to the
31676 corresponding dup operand, or even not when it is a CONST_VECTOR
31677 depending on the mode and lo vs. hi.
31678 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31680 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31681 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31682 Likewise. Require that operands[2] is even.
31683 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31684 Remove extraneous parens. Require that operands[2] is a multiple
31686 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31687 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31688 disallow memory then.
31690 2017-03-30 Richard Biener <rguenther@suse.de>
31692 PR tree-optimization/77498
31693 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31694 to non-constants over backedges.
31696 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31698 PR rtl-optimization/80233
31699 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31700 as last_combined_insn. Do not test for BARRIER_P separately.
31702 2017-03-29 Andreas Schwab <schwab@suse.de>
31705 * calls.c (prepare_call_address): Convert funexp to Pmode before
31706 copying to temp reg.
31708 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31710 PR tree-optimization/80158
31711 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31712 Handle possible future case of more than one alternate
31714 (replace_rhs_if_not_dup): Likewise.
31715 (replace_one_candidate): Likewise.
31717 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31719 PR rtl-optimization/80193
31720 * ira.c (ira): Do not check allocation for LRA.
31722 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31724 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31725 (nvptx_output_simt_exit): Declare.
31726 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31727 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31728 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31729 (nvptx_declare_function_name): ...here. Emit declaration of local
31730 memory space buffer for omp_simt_enter insn.
31731 (nvptx_output_unisimt_switch): New.
31732 (nvptx_output_softstack_switch): New.
31733 (nvptx_output_simt_enter): New.
31734 (nvptx_output_simt_exit): New.
31735 * config/nvptx/nvptx.h (struct machine_function): New fields
31736 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31737 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31738 (UNSPECV_SIMT_EXIT): Ditto.
31739 (omp_simt_enter_insn): New insn.
31740 (omp_simt_enter): New expansion.
31741 (omp_simt_exit): New insn.
31742 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31744 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31745 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31746 (expand_GOMP_SIMT_EXIT): New.
31747 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31748 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31749 (GOMP_SIMT_EXIT): Ditto.
31750 * target-insns.def (omp_simt_enter): New insn.
31751 (omp_simt_exit): Ditto.
31752 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31754 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31755 (lower_rec_input_clauses): Likewise.
31756 (lower_lastprivate_clauses): Handle SIMT privatization.
31758 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31759 (ompdevlow_adjust_simt_enter): New.
31760 (find_simtpriv_var_op): New.
31761 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31762 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31764 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31765 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31766 (copy_decl_for_dup_finish): Ditto.
31768 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31770 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31773 * config/i386/i386.c (ix86_option_override_internal): Always
31774 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31776 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31778 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31780 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31782 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31783 mark new edge's irreducible flag accordign to it.
31784 (vect_do_peeling): Check loop preheader edge's irreducible flag
31785 and pass it to function slpeel_add_loop_guard.
31787 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31789 PR tree-optimization/80218
31790 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31791 Update block frequencies and counts.
31793 2017-03-28 Richard Biener <rguenther@suse.de>
31795 PR tree-optimization/78644
31796 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31797 of a simplification result we may not use it at all.
31799 2017-03-28 Richard Biener <rguenther@suse.de>
31802 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31803 without arguments, generate default definition of a SSA name.
31805 2017-03-28 Richard Biener <rguenther@suse.de>
31807 PR middle-end/80222
31808 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31809 TYPE_REF_CAN_ALIAS_ALL references.
31810 * fold-const.c (fold_indirect_ref_1): Likewise.
31812 2017-03-28 Martin Liska <mliska@suse.cz>
31815 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31816 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31818 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31819 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31821 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31822 (EXTRA_SPECS): Define.
31823 (SUBTARGET_EXTRA_SPECS): Likewise.
31824 (SUBTARGET_CPP_SPEC): Likewise.
31825 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31826 SUBTARGET_EXTRA_SPECS.
31827 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31829 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31831 * config/arc/simdext.md (vst64_insn): Update pattern.
31832 (vld32wh_insn): Likewise.
31833 (vld32wl_insn): Likewise.
31834 (vld64_insn): Likewise.
31835 (vld32_insn): Likewise.
31837 2017-03-28 Marek Polacek <polacek@redhat.com>
31840 * fold-const.c (fold_comparison): Use protected_set_expr_location
31841 instead of SET_EXPR_LOCATION.
31843 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31845 * tree.c (add_expr): Avoid name lookup warning.
31847 2017-03-27 Jeff Law <law@redhat.com>
31849 PR tree-optimization/80216
31850 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31851 function name. Limit recursion depth.
31852 (record_temporary_equivalences): Corresponding changes.
31854 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31856 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31859 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31862 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31864 * cfgcleanup.c (reg_note_cfa_p): New array.
31865 (insns_have_identical_cfa_notes): New function.
31866 (old_insns_match_p): Don't cross-jump in between /f
31867 and non-/f instructions. If both i1 and i2 are frame related,
31868 verify all CFA notes, their order and content.
31870 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31873 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31874 HImode and SImode with zero extend to DImode to one insn.
31875 (bswap<mode>2_extenddi): Likewise.
31876 (bswapsi2_extenddi): Likewise.
31877 (bswaphi2_extendsi): Likewise.
31878 (bswaphi2): Combine bswap HImode and SImode into one insn.
31879 Separate memory insns from swapping register.
31880 (bswapsi2): Likewise.
31881 (bswap<mode>2): Likewise.
31882 (bswaphi2_internal): Delete, no longer used.
31883 (bswapsi2_internal): Likewise.
31884 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31885 store, and gpr<-gpr swap insns.
31886 (bswap<mode>2_store): Likewise.
31887 (bswaphi2_reg): Register only splitter, combine with the splitter.
31888 (bswaphi2 splitter): Likewise.
31889 (bswapsi2_reg): Likewise.
31890 (bswapsi2 splitter): Likewise.
31891 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31892 the insns into load, store, and register/register insns.
31893 (bswapdi2_ldbrx): Likewise.
31894 (bswapdi2_load): Likewise.
31895 (bswapdi2_store): Likewise.
31896 (bswapdi2_reg): Likewise.
31898 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31900 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31901 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31903 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31906 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31908 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31909 special handling for target option conflicts between dform
31910 options (-mpower9-dform, -mpower9-dform-vector,
31911 -mpower9-dform-scalar) and -mno-direct-move.
31913 2017-03-27 Richard Biener <rguenther@suse.de>
31915 PR tree-optimization/80181
31916 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31918 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31920 * config/arc/predicates.md (move_double_src_operand): Replace the
31921 call to move_double_src_operand with a call to address_operand.
31923 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31925 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31926 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31927 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31929 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31931 * config/arc/predicates.md (long_immediate_loadstore_operand):
31932 Consider scaled addresses cases.
31934 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31936 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31937 restored when in interrupt.
31938 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31939 doesn't have delay slot.
31941 2017-03-27 Richard Biener <rguenther@suse.de>
31944 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31945 inlined thunk clones.
31947 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31950 * asan.c (instrument_derefs): Copy over last operand from
31951 original COMPONENT_REF to the new COMPONENT_REF with
31952 DECL_BIT_FIELD_REPRESENTATIVE.
31953 * ubsan.c (instrument_object_size): Likewise.
31955 2017-03-27 Richard Biener <rguenther@suse.de>
31957 PR tree-optimization/80170
31958 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31959 sure DR/SCEV didnt fold in constants we do not see when looking
31960 at the reference base alignment.
31962 2017-03-27 Richard Biener <rguenther@suse.de>
31964 PR middle-end/80171
31965 * gimple-fold.c (fold_ctor_reference): Properly guard against
31966 NULL return value from canonicalize_constructor_val.
31968 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31971 * config/i386/i386.c (ix86_expand_builtin)
31972 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31973 flags reg setting and flags reg using instructions.
31974 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31975 clobbering instructions to zero extend op2.
31977 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31979 * doc/install.texi (Configuration) <--with-aix-soname>:
31980 Update link to AIX ld.
31982 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31984 PR rtl-optimization/80160
31985 PR rtl-optimization/80159
31986 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31987 reg_alternate_class into account.
31989 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31992 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31993 to consider in curr_insn_transform.
31995 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31997 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31998 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31999 and emit_mode_inner.
32001 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32003 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32004 argument to the overloaded builtin variants. Use the new flag to
32005 deprecate certain builtin variants.
32006 * config/s390/s390-builtin-types.def: Add new builtin types.
32007 * config/s390/s390-builtins.h: Support new flags field for
32008 overloaded builtins.
32009 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32010 (s390_macro_to_expand): Enable vector float data type.
32011 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32012 builtins by incrementing the __VEC__ version number.
32013 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32015 (s390_resolve_overloaded_builtin): Emit error messages depending
32016 on the builtin flags.
32017 * config/s390/s390.c (s390_expand_builtin): Support additional
32018 flags argument. Change error message to match the messages
32019 emitted in s390-c.c.
32020 * config/s390/s390.md: New UNSPEC_* constants.
32021 (op_type): Add new instruction types.
32022 * config/s390/vecintrin.h: Add new builtins and test data class
32024 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32025 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32026 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32027 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32028 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32029 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32031 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32032 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32033 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32034 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32036 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32037 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32038 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32039 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32041 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32042 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32043 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32044 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32045 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32046 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32047 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32049 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32050 ("vec_scatter_element<V_HW_4:mode>_DI")
32051 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32052 ("vec_fpint<mode>", "vflls")
32053 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32054 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32055 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32056 ("*vec_cmphe<mode>_cc"): ... these.
32058 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32059 mode constant instead of magic value.
32061 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32063 * config/s390/s390.c (s390_expand_vec_compare): Support other
32064 vector floating point modes than just V2DF.
32065 (s390_expand_vcond): Likewise.
32066 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32067 (s390_cannot_change_mode_class): Prevent mode changes between TF
32068 and V1TF in vector registers.
32069 * config/s390/s390.md (DF, SF): New mode attributes.
32070 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32071 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32072 SFmode support for VRs.
32073 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32075 (VFT, VF_HW): New mode iterators.
32076 (vw, sdx): New mode attributes.
32077 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32078 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32079 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32080 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32081 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32082 also the new vector floating point modes. Renaming to ...
32084 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32085 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32086 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32087 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32088 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32089 ("vec_unordered<mode>"): ... these.
32091 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32092 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32093 ("*vec_extendv2df"): New insn definitions.
32095 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32097 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32098 ("mulditi3_2", "*muldi3_sign"): New patterns.
32099 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32100 rename the pattern definition.
32102 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32104 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32106 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32108 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32110 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32111 instruction if possible.
32112 * config/s390/vector.md (vec_halfnumelts): New mode
32114 ("*vec_vllezlf<mode>"): New pattern.
32116 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32118 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32119 ("popcountv4si2", "popcountv2di2"): Rename to ...
32120 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32121 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32123 ("popcount<mode>2_vxe"): New pattern.
32125 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32127 * common/config/s390/s390-common.c (processor_flags_table): Add
32129 * config.gcc: Add arch12.
32130 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32131 Default to arch12 for unknown CPU model numbers.
32132 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32133 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32134 PROCESSOR_max sanity check.
32135 * config/s390/s390-opts.h (enum processor_type): Add
32137 * config/s390/s390.c (processor_table): Add arch12.
32138 (s390_expand_builtin): Add check for B_VXE flag.
32139 (s390_issue_rate): Add PROCESSOR_ARCH12.
32140 (s390_get_sched_attrmask): Likewise.
32141 (s390_get_unit_mask): Likewise.
32142 (s390_sched_score): Enable z13 scheduling for arch12.
32143 (s390_sched_reorder): Likewise.
32144 (s390_sched_variable_issue): Likewise.
32145 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32147 (s390_tune_attr): Use z13 scheduling also for arch12.
32148 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32149 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32150 (TARGET_VXE_P): New macros.
32151 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32152 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32153 * config/s390/s390.opt: Add arch12 as processor_type.
32155 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32157 * config/s390/s390.md
32158 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32159 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32160 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32162 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32163 Rename expanders to ...
32165 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32166 ("fixuns_truncdddi2_emu"): ... these.
32168 ("fixuns_trunc<mode>si2_emu"): New expander.
32170 ("*fixuns_truncdfdi2_z13"): Rename to ...
32171 ("*fixuns_truncdfdi2_vx"): ... this.
32173 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32175 * config/s390/2964.md: Remove the single element vector compare
32176 instructions which are no longer used.
32177 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32179 (s390_canonicalize_comparison): Remove handling of DFmode
32181 (s390_expand_vec_compare_scalar): Remove function.
32182 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32183 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32185 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32187 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32189 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32191 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32192 will anyway by matched by mov<mode>_64dfp.
32194 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32196 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32197 vlef/vstef. Add missing operand to vleif.
32199 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32201 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32202 pair for all vector types with 64 bit elements.
32203 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32204 * config/s390/vector.md (V_HW_64): ... here.
32205 (V_128_NOSINGLE): New mode iterator.
32206 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32207 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32208 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32209 ("*vec_load_pairv2di"): Change to ...
32210 ("*vec_load_pair<mode>"): ... this one.
32212 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32214 * config/s390/constraints.md: Add comments.
32215 (jKK): Reject element sizes > 8 bytes.
32216 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32218 * config/s390/s390.md: Add the s_operand checks formerly in
32219 s390_split_ok_p to various splitters where they are still
32221 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32222 for 128 bit vectors. Plus two splitters.
32224 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32226 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32229 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32232 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32233 error if the boundary argument is not constant.
32235 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32237 PR rtl-optimization/80112
32238 * loop-doloop.c (doloop_condition_get): Don't check condition
32239 if cmp isn't SET with IF_THEN_ELSE src.
32241 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32243 PR tree-optimization/80158
32244 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32245 replacing a candidate statement, also replace it for the
32246 candidate's alternate interpretation.
32247 (replace_rhs_if_not_dup): Likewise.
32248 (replace_one_candidate): Likewise.
32250 2017-03-24 Richard Biener <rguenther@suse.de>
32252 PR tree-optimization/80167
32253 * graphite-isl-ast-to-gimple.c
32254 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32256 (translate_isl_ast_to_gimple::get_rename): Likewise.
32258 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32260 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32261 handling of certain combinations of target options, including the
32262 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32263 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32265 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32268 * config/arm/arm.md (*load_multiple): Add reload_completed to
32269 matching condition.
32271 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32272 Richard Biener <rguenth@suse.de>
32274 PR tree-optimization/79908
32275 PR tree-optimization/80136
32276 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32277 been cast away, gimplify_and_add suffices.
32279 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32281 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32283 2017-03-23 Richard Biener <rguenther@suse.de>
32285 PR tree-optimization/80032
32286 * gimplify.c (gimple_push_cleanup): Forced unconditional
32287 cleanups still have to go to the conditional_cleanups
32290 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32292 PR tree-optimization/80072
32293 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32295 (next_operand_entry_id): Change type to unsigned int.
32296 (sort_by_operand_rank): Make sure to return the right return value
32297 even if unsigned fields are bigger than INT_MAX.
32298 (struct oecount): Change cnt and id type to unsigned int.
32299 (oecount_hasher::equal): Formatting fix.
32300 (oecount_cmp): Make sure to return the right return value
32301 even if unsigned fields are bigger than INT_MAX.
32302 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32305 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32306 TREE_READONLY on result if writing it more than once.
32309 * doc/invoke.texi (-fsanitize=thread): Document that with
32310 -fnon-call-exceptions atomics are not able to throw
32314 * tsan.c: Include tree-eh.h.
32315 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32316 maybe_clean_or_replace_eh_stmt where needed.
32317 (instrument_memory_accesses): Add cfg_changed argument.
32318 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32319 if it returned true.
32320 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32322 PR rtl-optimization/63191
32323 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32324 wrapper function, moved the whole old content into ...
32325 (ix86_delegitimize_address_1): ... this. New inline function.
32326 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32327 true as last argument instead of ix86_delegitimize_address.
32329 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32331 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32332 cortexa57_branch_cost.
32334 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32336 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32338 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32341 * doc/md.texi (Constraints): Document wA constraint.
32342 * config/rs6000/constraints.md (wA): New.
32343 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32344 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32345 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32346 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32348 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32351 * gimplify.c (is_oacc_declared): New function.
32352 (oacc_default_clause): Use it to set default flags for acc declared
32353 variables inside parallel regions.
32354 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32355 declared variables.
32356 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32357 declare attribute to any decl as necessary.
32359 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32362 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32363 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32364 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32365 (arm_arch_lpae): This.
32366 * config/arm/arm.c (arm_arch7ve): Rename into ...
32367 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32368 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32371 2017-03-22 Martin Liska <mliska@suse.cz>
32374 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32375 error message instead of an ICE.
32377 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32379 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32381 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32383 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32385 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32388 2017-03-21 Martin Sebor <msebor@redhat.com>
32390 * doc/extend.texi: Use "cannot" instead of "can't."
32391 * doc/hostconfig.texi: Same.
32392 * doc/install.texi: Same.
32393 * doc/invoke.texi: Same.
32394 * doc/loop.texi: Same.
32395 * doc/md.texi: Same.
32396 * doc/objc.texi: Same.
32397 * doc/rtl.texi: Same.
32398 * doc/tm.texi: Same.
32399 * doc/tm.texi.in: Same.
32400 * doc/trouble.texi: Same.
32402 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32405 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32406 (collect_checksum_attributes): Set it.
32407 (die_checksum_ordered): Use it.
32409 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32411 PR tree-optimization/79908
32412 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32413 change: For a VA_ARG whose LHS has been cast away, use
32414 force_gimple_operand to construct the side effects.
32416 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32418 PR translation/80001
32419 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32420 more amenable to translation.
32421 (oacc_loop_auto_partitions): Likewise.
32423 2017-03-21 Marek Polacek <polacek@redhat.com>
32424 Martin Sebor <msebor@redhat.com>
32426 PR tree-optimization/80109
32427 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32428 on INTEGRAL_TYPE_P.
32430 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32431 Segher Boessenkool <segher@kernel.crashing.org>
32434 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32435 check reg_used_between_p between insn and one of succ or succ2
32436 depending on if succ is artificial insn not inserted into insn
32439 2017-03-21 Martin Liska <mliska@suse.cz>
32441 PR gcov-profile/80081
32442 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32443 * doc/gcc.texi: Include gcov-dump stuff.
32444 * doc/gcov-dump.texi: New file.
32446 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32448 PR rtl-optimization/79150
32449 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32450 conditional jump, if the jump is the last insn of the loop.
32452 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32453 Richard Biener <rguenth@suse.de>
32455 PR tree-optimization/79908
32456 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32457 been cast away, use force_gimple_operand to construct the side
32460 2017-03-21 Martin Liska <mliska@suse.cz>
32462 PR libfortran/79956
32463 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32466 2017-03-21 Brad Spengler <spender@grsecurity.net>
32469 * plugin.c (htab_hash_plugin): New function.
32470 (add_new_plugin): Use it and adjust.
32471 (parse_plugin_arg_opt): Adjust.
32472 (init_one_plugin): Likewise.
32474 2017-03-21 Richard Biener <rguenther@suse.de>
32476 PR tree-optimization/80032
32477 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32478 if set force the cleanup to happen unconditionally.
32479 (gimplify_target_expr): Push inserted clobbers with force_uncond
32480 to avoid them being removed by control-dependent DCE.
32482 2017-03-21 Richard Biener <rguenther@suse.de>
32484 PR tree-optimization/80122
32485 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32486 va_arg_pack_len when the inlined call stmt requires pack
32488 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32490 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32493 * tsan.c (instrument_builtin_call): If the memory model argument
32494 is not a constant, assume it is valid.
32497 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32500 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32502 PR rtl-optimization/79910
32503 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32504 if its dest is used by an insn before I2 (other than the combined
32505 insns themselves, which are properly handled already).
32507 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32510 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32512 * combine.c (record_used_regs): New static function.
32513 (try_combine): Handle situations where there is an additional
32514 instruction between I2 and I3 which needs to have a LOG_LINK
32518 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32520 * combine.c (try_combine): Delete redundant i1 test. Call
32521 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32523 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32526 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32527 alternatives 13/14.
32529 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32531 PR tree-optimization/80054
32532 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32533 the optimization if a PHI or any of its arguments is not dominated
32534 by the candidate's basis. Use gphi* rather than gimple* as
32536 (replace_profitable_candidates): Clean up a gimple* variable that
32537 should be a gphi* variable.
32539 2017-03-20 Martin Sebor <msebor@redhat.com>
32542 * doc/extend.texi (attribute constructor): Document present limitation.
32544 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32547 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32548 __POWER9_VECTOR__ #ifdef control, change template definition to
32549 use Power9-specific built-in function.
32550 (vec_any_eq): Likewise.
32551 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32552 to control outcomes from this test.
32553 (vector_ae_<mode>p): For VEC_F modes, likewise.
32555 2017-03-20 Ian Lance Taylor <iant@google.com>
32557 * config/i386/i386.c (ix86_function_regparm): Save an extra
32558 register for -fsplit-stack with DECL_STATIC_CHAIN.
32560 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32563 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32564 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32566 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32568 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32570 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32573 2017-03-20 Marek Polacek <polacek@redhat.com>
32576 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32578 2017-03-20 Richard Biener <rguenther@suse.de>
32580 PR tree-optimization/80113
32581 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32582 allocate extra SSA name for PHI def.
32583 (add_close_phis_to_outer_loops): Likewise.
32584 (add_close_phis_to_merge_points): Likewise.
32585 (copy_loop_close_phi_args): Likewise.
32586 (copy_cond_phi_nodes): Likewise.
32588 2017-03-20 Martin Liska <mliska@suse.cz>
32590 PR middle-end/79753
32591 * tree-chkp.c (chkp_build_returned_bound): Do not build
32592 returned bounds for a LHS that's not a BOUNDED_P type.
32594 2017-03-20 Martin Liska <mliska@suse.cz>
32598 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32599 COMPLEX_CST and VECTOR_CST.
32601 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32604 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32605 target operand. A new splitter adds the clobber statement in case
32606 the target operand is dead anyway.
32608 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32610 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32611 to age-old versions of binutils and glibc.
32613 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32615 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32617 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32619 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32621 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32623 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32624 requirement for binutils 2.13.
32626 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32628 * combine.c (try_combine): Delete redundant i1 test. Call
32629 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32631 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32633 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32634 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32636 <riscv64-*-elf>: Re-arrange section
32637 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32638 <riscv32-*-linux>: Likewise.
32639 <riscv64-*-elf>: Likewise
32640 <riscv64-*-linux>: Likewise.
32642 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32645 * aarch64.opt(verbose-cost-dump): Fix typo.
32647 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32650 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32651 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32653 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32655 * reload.c (find_reloads): When reloading a nonoffsettable address,
32656 use RELOAD_OTHER for it and its address reloads.
32658 PR rtl-optimization/79910
32659 * combine.c (record_used_regs): New static function.
32660 (try_combine): Handle situations where there is an additional
32661 instruction between I2 and I3 which needs to have a LOG_LINK
32664 2017-03-17 Jeff Law <law@redhat.com>
32666 PR tree-optimization/71437
32667 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32668 conditional in the hash table first.
32669 (vrp_dom_walker::before_dom_children): Extract condition from
32670 ASSERT_EXPR. Record condition, its inverion and any implied
32671 conditions as well.
32673 2017-03-17 Marek Polacek <polacek@redhat.com>
32674 Markus Trippelsdorf <markus@trippelsdorf.de>
32676 PR tree-optimization/80079
32677 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32680 2017-03-17 Richard Biener <rguenther@suse.de>
32682 PR middle-end/80075
32683 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32684 Properly verify the LHS before the RHS possibly claims to be
32686 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32689 2017-03-17 Martin Jambor <mjambor@suse.cz>
32691 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32692 (List of -O2 options): Likewise.
32693 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32696 2017-03-17 Tom de Vries <tom@codesourcery.com>
32698 * gcov-dump.c (print_usage): Print bug_report_url.
32700 2017-03-17 Richard Biener <rguenther@suse.de>
32702 PR middle-end/80050
32703 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32704 (parser::peek): Likewise.
32706 2017-03-17 Richard Biener <rguenther@suse.de>
32708 PR tree-optimization/80048
32709 * sese.c (free_sese_info): Properly release rename_map and
32710 copied_bb_map elements.
32712 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32714 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32715 Add linked-list forward and backlinks. Insert on
32716 construction, remove on destruction.
32717 (class pass_store_merging): Add m_stores_head field.
32718 (pass_store_merging::terminate_and_process_all_chains):
32719 Iterate over m_stores_head list.
32720 (pass_store_merging::terminate_all_aliasing_chains):
32722 (pass_store_merging::execute): Check for debug stmts first.
32723 Push new chains onto the m_stores_head stack.
32725 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32728 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32729 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32730 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32732 2017-03-16 Jeff Law <law@redhat.com>
32734 PR tree-optimization/71437
32735 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32736 member function. Implementation moved into after_dom_children
32737 member function and into the threader's thread_outgoing_edges
32739 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32740 some code into new thread_outgoing_edges.
32741 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32742 definition. Simplify marker handling (do it here). Assume we always
32743 have the available expression and the const/copies tables.
32744 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32746 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32747 * tree-vrp.c (equiv_stack): No longer file scoped.
32748 (vrp_dom_walker): New class.
32749 (vrp_dom_walker::before_dom_children): New member function.
32750 (vrp_dom_walker::after_dom_children): Likewise.
32751 (identify_jump_threads): Setup domwalker. Use it rather than
32752 walking edges in a random order by hand. Simplify setup/finalization.
32753 (finalize_jump_threads): Remove.
32754 (vrp_finalize): Do not call identify_jump_threads here.
32755 (execute_vrp): Do it here instead and call thread_through_all_blocks
32758 PR tree-optimization/71437
32759 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32761 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32763 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32764 (dom_opt_dom_walker::thread_across_edge): Remove
32765 handle_dominating_asserts argument. All callers changed.
32766 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32767 changes. Remove calls to lhs_of_dominating_assert. Other
32768 uses of handle_dominating_asserts turn into unconditional code
32769 (simplify_control_stmt_condition_1): Likewise.
32770 (simplify_control_stmt_condition): Likewise.
32771 (thread_through_normal_block, thread_across_edge): Likewise.
32772 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32773 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32774 object if it is not an SSA_NAME.
32775 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32776 before calling into the VRP specific simplifiers.
32777 (identify_jump_threads): Remove handle_dominating_asserts
32780 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32783 * tree-diagnostic.c (default_tree_printer): No longer static.
32784 * tree-diagnostic.h (default_tree_printer): New prototype.
32786 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32788 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32789 Change ins into fmov.
32791 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32793 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32794 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32795 Use h_con constraint for operand 1.
32796 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32797 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32799 2017-03-15 Jeff Law <law@redhat.com>
32801 PR tree-optimization/71437
32802 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32803 (record_temporary_equivalences): Use it.
32805 PR tree-optimization/71437
32806 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32807 tree-ssa-scopedtables.
32808 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32809 (record_conditions, record_cond, vuse_eq): Likewise.
32810 (record_edge_info): Adjust to API tweak of record_conditions.
32811 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32812 (record_temporary_equivalences, optimize_stmt): Likewise.
32813 (eliminate_redundant_computations): Likewise.
32814 (record_equivalences_from_stmt): Likewise.
32815 * tree-ssa-scopedtables.c: Include options.h and params.h.
32816 (vuse_eq): New function, moved from tree-ssa-dom.c
32817 (build_and_record_new_cond): Likewise.
32818 (record_conditions): Likewise. Accept vector of conditions rather
32819 than edge_equivalence structure for first argument.
32820 for the first argument.
32821 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32822 from tree-ssa-dom.c.
32823 (avail_exprs_stack::record_cond): Likewise.
32824 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32825 from tree-ssa-dom.c.
32826 (avail_exprs_stack): Add new member functions lookup_avail_expr
32828 (record_conditions): Declare.
32830 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32833 * lra-constraints.c (process_alt_operands): Increase reject for
32834 reloading an input/output operand.
32836 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32839 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32840 insns to convert from signed/unsigned char/short to IEEE 128-bit
32842 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32844 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32847 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32848 subreg of inner mode for values already in registers.
32850 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32852 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32853 iteration reg is used after the loop.
32855 2017-03-14 Martin Sebor <msebor@redhat.com>
32857 PR tree-optimization/79800
32858 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32859 precision in negative-positive range.
32860 (format_floating): Call non-const overload with adjusted precision.
32862 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32865 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32868 2017-03-14 Martin Sebor <msebor@redhat.com>
32870 PR middle-end/80020
32871 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32872 * builtins.def (aligned_alloc): Use it.
32875 * Makefile.in (GTFILES): Add calls.c.
32876 * calls.c: Include "gt-calls.h".
32878 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32880 PR rtl-optimization/79728
32881 * regs.h (struct target_regs): New field
32882 x_contains_allocatable_regs_of_mode.
32883 (contains_allocatable_regs_of_mode): New macro.
32884 * reginfo.c (init_reg_sets_1): Initialize it, and change
32885 contains_reg_of_mode so it includes global regs as well.
32886 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32887 rather than contains_regs_of_mode.
32889 2017-03-14 Martin Liska <mliska@suse.cz>
32891 * doc/invoke.texi: Document options that can't be combined with
32892 -fcheck-pointer-bounds.
32894 2017-03-14 Martin Liska <mliska@suse.cz>
32896 PR middle-end/79831
32897 * doc/invoke.texi (-Wchkp): Document the option.
32899 2017-03-14 Martin Liska <mliska@suse.cz>
32901 * Makefile.in: Install gcov-dump.
32903 2017-03-14 Martin Liska <mliska@suse.cz>
32905 * multiple_target.c (expand_target_clones): Bail out for
32906 an invalid attribute.
32908 2017-03-14 Richard Biener <rguenther@suse.de>
32910 * alias.c (struct alias_set_entry): Pack properly.
32911 * cfgloop.h (struct loop): Likewise.
32912 * cse.c (struct set): Likewise.
32913 * ipa-utils.c (struct searchc_env): Likewise.
32914 * loop-invariant.c (struct invariant): Likewise.
32915 * lra-remat.c (struct cand): Likewise.
32916 * recog.c (struct change_t): Likewise.
32917 * rtl.h (struct address_info): Likewise.
32918 * symbol-summary.h (function_summary): Likewise.
32919 * tree-loop-distribution.c (struct partition): Likewise.
32920 * tree-object-size.c (struct object_size_info): Likewise.
32921 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32922 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32923 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32924 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32925 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32926 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32927 (struct _stmt_vec_info): Likewise.
32929 2017-03-14 Martin Liska <mliska@suse.cz>
32932 * multiple_target.c (create_dispatcher_calls): Check that
32933 a target can create a function dispatcher.
32935 2017-03-14 Martin Liska <mliska@suse.cz>
32938 * multiple_target.c (expand_target_clones): Drop local.local
32939 flag for default implementation.
32941 2017-03-14 Richard Biener <rguenther@suse.de>
32943 PR tree-optimization/80030
32944 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32946 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32948 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32949 gcc_fallthrough() instead of __attribute__((fallthrough));
32951 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32953 * doc/gcc.texi: Remove "up" link to (DIR).
32954 * doc/gccint.texi: Ditto.
32956 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32958 * doc/install.texi (Specific) <avr>: Remove reference to
32961 2017-03-13 Jeff Law <law@redhat.com>
32963 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32964 attribute rather than comments.
32966 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32967 match_scratch operand is highest.
32969 2017-03-13 Martin Liska <mliska@suse.cz>
32971 PR middle-end/78339
32972 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32973 is a CHKP clone, use original declaration.
32975 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32977 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32978 (arc_conditional_register_usage): Use a different allocation order
32979 when optimizing for size.
32980 * common/config/arc/arc-common.c (arc_option_optimization_table):
32981 Section anchors default on when optimizing for size.
32983 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32985 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32987 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32989 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32990 * config/arc/arc.md (cpu_facility): Add cd variant.
32991 (*movqi_insn): Add code density variant.
32992 (*movhi_insn): Likewise.
32993 (*movqi_insn): Likewise.
32994 (*addsi3_mixed): Likewise.
32995 (subsi3_insn): Likewise.
32997 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32999 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33001 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33003 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33004 expressions with MINUS and UNARY ops.
33006 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33009 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33011 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33012 between vec_select and vector argument.
33013 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33014 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33015 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33016 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33017 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33018 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33020 2017-03-13 Richard Biener <rguenther@suse.de>
33023 * params.def (vect-max-peeling-for-alignment): Fix typo.
33025 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33027 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33028 issue that only occurred with binutils below 2.18.
33030 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33032 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33033 refer to binutils 2.11/2.12 minimum.
33035 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33037 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33038 ftp.kernel.org and simplify binutils requirement.
33040 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33042 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33044 (Optimize Options): Ditto. Also remove redundancy.
33046 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33048 PR translation/79848
33049 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33051 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33052 to G_ to avoid double translation.
33054 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33056 PR translation/79923
33057 * auto-profile.c (get_combined_location): Convert leading
33058 character of diagnostics to lower case and remove trailing period.
33059 (read_profile): Likewise for various diagnostics.
33060 * config/arm/arm.c (arm_option_override): Remove trailing period
33061 from various diagnostics.
33062 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33063 (msp430_expand_delay_cycles): Likewise.
33065 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33068 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33069 full command-line argument, rather than just "str".
33070 (aarch64_validate_march): Likewise.
33071 (aarch64_validate_mtune): Likewise.
33073 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33075 PR rtl-optimization/78911
33076 * lra-assigns.c (must_not_spill_p): New function.
33077 (spill_for): Use it.
33079 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33081 PR tree-optimization/79981
33082 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33083 ATOMIC_COMPARE_EXCHANGE ifn result.
33084 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33085 IFN_ATOMIC_COMPARE_EXCHANGE.
33087 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33090 * opts.c (parse_sanitizer_options): Add missing question mark to
33091 "did you mean" message.
33093 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33095 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33097 (VMULEUH_UNS): Likewise.
33098 (VMULOUB_UNS): Likewise.
33099 (VMULOUH_UNS): Likewise.
33100 * config/rs6000/rs6000.c (builtin_function_type): Remove
33101 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33103 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33106 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33107 x with result of extra_parsing_for_operand_code_0.
33108 (function_reader::extra_parsing_for_operand_code_0): Convert
33109 return type from void to rtx, returning x. When reading
33110 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33111 larger size containing struct block_symbol.
33113 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33115 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33116 -mfloat128-hardware without -m64.
33118 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33121 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33122 entries to the case statement that marks unsigned arguments to
33123 overloaded functions.
33125 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33127 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33128 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33130 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33133 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33134 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33136 2017-03-10 Martin Liska <mliska@suse.cz>
33140 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33141 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33144 2017-03-10 Olivier Hainque <hainque@adacore.com>
33146 * tree-switch-conversion (array_value_type): Start by resetting
33147 candidate type to it's main variant.
33149 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33151 PR rtl-optimization/79909
33152 * combine.c (try_combine): Use simplify_replace_rtx on individual
33153 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33154 of the whole CALL_INSN_FUNCTION_USAGE.
33156 PR tree-optimization/79972
33157 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33158 get_range_info on SSA_NAMEs. Formatting fixes.
33160 2017-03-10 Richard Biener <rguenther@suse.de>
33161 Jakub Jelinek <jakub@redhat.com>
33163 PR tree-optimization/77975
33164 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33165 edge to be constant.
33166 (get_val_for): For constant x return it. Formatting fix.
33167 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33168 would use the same bases as the current one.
33170 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33172 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33173 instead of vec_select for V1TImode.
33174 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33176 (VSX_LE_128): Add V1TI to this mode iterator.
33177 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33178 (*vsx_le_perm_store_<mode>): Likewise.
33179 (pre-reload splitter for VSX stores): Likewise.
33180 (post-reload splitter for VSX stores): Likewise.
33181 (*vsx_xxpermdi2_le_<mode>): Likewise.
33182 (*vsx_lxvd2x2_le_<mode>): Likewise.
33183 (*vsx_stxvd2x2_le_<mode>): Likewise.
33185 2017-03-09 Michael Eager <eager@eagercon.com>
33187 Correct failures with --enable-checking=yes,rtl.
33189 * config/microblaze/microblaze.c (microblaze_expand_shift):
33190 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33191 test for const0_rtx.
33192 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33193 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33195 2017-03-09 Richard Biener <rguenther@suse.de>
33197 PR tree-optimization/79977
33198 * graphite-scop-detection.c (scop_detection::merge_sese):
33199 Handle the case of extra exits to blocks dominating the entry.
33201 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33203 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33206 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33208 PR rtl-optimization/79949
33209 * lra-constraints.c (process_alt_operands): Check memory when
33210 trying to predict a cycle. Print about the overall increase.
33212 2017-03-09 Richard Biener <rguenther@suse.de>
33214 PR middle-end/79971
33215 * gimple-expr.c (useless_type_conversion_p): Preserve
33216 TYPE_SATURATING for fixed-point types.
33218 2017-03-09 Richard Biener <rguenther@suse.de>
33221 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33222 alignment of BLKmode params.
33224 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33227 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33228 (VALL_NO_V2Q): Likewise.
33230 * config/aarch64/aarch64-simd.md
33231 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33233 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33234 VALL_NO_V2Q mode iterator.
33235 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33237 2017-03-09 Martin Liska <mliska@suse.cz>
33239 PR tree-optimization/79631
33240 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33241 tree_int_cst_sign_bit just for INTEGER constants.
33243 2017-03-09 Martin Liska <mliska@suse.cz>
33247 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33250 2017-03-09 Marek Polacek <polacek@redhat.com>
33253 * tree.c (inchash::add_expr): Handle TREE_VEC.
33255 2017-03-09 Martin Liska <mliska@suse.cz>
33258 (chkp_narrow_size_and_offset): New function.
33259 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33260 (void chkp_parse_bit_field_ref): New function.
33261 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33262 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33264 2017-03-09 Martin Liska <mliska@suse.cz>
33267 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33268 (chkp_find_bounds_1): Remove gcc_unreachable.
33270 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33273 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33274 BUILT_IN_SYNC*, determine the access type from the size suffix and
33275 always build a MEM_REF with that type. Handle forgotten
33276 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33279 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33280 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33281 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33282 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33283 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33284 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33285 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33286 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33287 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33288 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33289 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33290 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33291 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33292 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33293 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33294 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33295 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33296 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33297 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33298 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33299 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33300 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33301 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33302 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33303 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33304 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33305 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33306 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33307 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33308 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33309 definitions outside of __OPTIMIZE__ guarded section.
33312 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33313 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33314 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33315 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33318 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33320 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33321 ("vfenez<mode>"): Add missing constraints.
33323 2017-03-08 Martin Sebor <msebor@redhat.com>
33326 * config/nds32/nds32.c (nds32_option_override):
33327 Fix misspelled diagnostic.
33329 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33332 * gimplify.c (gimplify_omp_for): Replace index var in outer
33333 taskloop statement with an artificial variable and add
33334 OMP_CLAUSE_PRIVATE clause for it.
33336 2017-03-08 Richard Biener <rguenther@suse.de>
33338 PR tree-optimization/79955
33339 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33340 for accesses that are completely outside of the variable.
33342 2017-03-08 Andrew Haley <aph@redhat.com>
33344 PR tree-optimization/79943
33345 * tree-ssa-loop-split.c (compute_new_first_bound): When
33346 calculating the new upper bound, (END-BEG) should be added, not
33349 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33351 * config/avr/avr.md (setmemhi): Make sure match_dup
33352 operand number comes before match_scratch.
33354 2017-03-08 Richard Biener <rguenther@suse.de>
33356 PR tree-optimization/79920
33357 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33358 with ncopies == 1 to ...
33359 (vect_transform_slp_perm_load): ... here. Properly compute
33360 all element loads by iterating VF times over the group. Do
33361 not handle ncopies (computed in a broken way) in
33362 vect_create_mask_and_perm.
33364 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33367 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33368 is a uniform vector, use uniform_vector_p return value instead of
33369 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33371 2017-03-07 Marek Polacek <polacek@redhat.com>
33373 PR middle-end/79809
33374 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33375 (alloca_call_type): Likewise.
33377 2017-03-07 Martin Liska <mliska@suse.cz>
33379 * gcov.c (process_args): Put comment to correct location.
33381 2017-03-07 Martin Liska <mliska@suse.cz>
33383 PR middle-end/68270
33384 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33385 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33386 (chkp_narrow_bounds_for_field): Likewise.
33387 (chkp_parse_array_and_component_ref): Pass one more argument to
33390 2017-03-07 Richard Biener <rguenther@suse.de>
33392 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33395 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33397 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33398 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33400 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33403 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33404 to end of description.
33405 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33407 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33409 PR rtl-optimization/79901
33410 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33412 (*avx512f_<code><mode>3<mask_name>): ... this.
33413 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33414 iterator instead of VI8_AVX2_AVX512BW.
33416 PR rtl-optimization/79901
33417 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33418 min/max expander, expand it using expand_vec_cond_expr.
33421 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33424 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33427 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33428 to void * for PCH reasons.
33429 * dwarf2out.c (output_loc_operands, output_die): Cast
33430 v.val_vec.array to unsigned char *.
33432 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33435 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33438 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33440 PR rtl-optimization/79571
33441 * lra-constraints.c (process_alt_operands): Calculate static
33442 reject and subtract it from overall when only addresses will be
33445 2017-03-06 Julia Koval <julia.koval@intel.com>
33448 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33449 incoming stack boundary to 128 for 64-bit targets.
33451 2017-03-06 Richard Biener <rguenther@suse.de>
33453 PR tree-optimization/79894
33454 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33455 to NULL after folding it.
33457 2017-03-06 Richard Biener <rguenther@suse.de>
33459 PR tree-optimization/79824
33460 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33461 check disabling peeling for gaps.
33463 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33465 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33466 attributes): Document gettimeofday.
33468 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33470 * config/s390/s390.c (s390_option_override_internal): Set
33471 PARAM_MIN_VECT_LOOP_BOUND
33473 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33475 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33476 * config/s390/s390.md: Likewise.
33478 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33481 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33482 (<avx2_avx512>_perm<mode>): Rename to ...
33483 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33485 (<avx512>_perm<mode>_mask): Rename to ...
33486 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33488 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33489 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33490 instead of VI8F_256_512.
33491 (avx512f_perm<mode>): New define_expand.
33492 (avx512f_perm<mode>_mask): Likewise.
33493 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33494 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33496 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33498 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33499 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33501 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33503 2017-03-06 Martin Liska <mliska@suse.cz>
33506 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33507 when having a SSA NAME w/o VAR_DECL assigned to it.
33509 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33511 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33512 msa_dpsub_<su>_d): Fix MODE for vec_select.
33514 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33516 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33518 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33520 2017-03-06 Richard Biener <rguenther@suse.de>
33522 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33523 * plugin.c (register_plugin_info): Likewise.
33524 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33526 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33528 * config/i386/sse.md (sse_storehps, sse_storelps,
33529 avx_<castmode><avxsizesuffix>_<castmode>,
33530 avx512f_<castmode><avxsizesuffix>_<castmode>,
33531 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33532 in condition that at least one operand is not a MEM.
33534 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33536 PR middle-end/79805
33537 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33538 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33540 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33541 gimple_call_nothrow_p flag based on whether original builtin can throw.
33542 If it can, emit following stmts on the fallthrough edge.
33543 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33544 don't create new bb if inserting just debug stmts on the edge, try to
33545 insert them on the fallthru bb or just reset debug stmts.
33547 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33550 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33551 restore recog_data (including the operand rtxes inside it) around
33552 the call to get_insn_template.
33554 2017-03-03 Martin Sebor <msebor@redhat.com>
33556 PR tree-optimization/79699
33557 * context.c (context::~context): Free MPFR caches to avoid
33558 a memory leak on program exit.
33560 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33562 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33563 Use wide_int::ulow () instead of .elt (0).
33565 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33567 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33568 (*pushxf): Limit oF constraint to 32bit targets and add oC
33569 constraint for 64bit targets.
33570 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33571 (*pushdf): Change rmF constraint to rmC.
33573 2017-03-03 Martin Liska <mliska@suse.cz>
33575 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33576 Remove unused variable.
33578 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33581 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33582 is a memory operand, increase num_memory.
33583 (ix86_expand_args_builtin): Likewise.
33585 2017-03-03 Jan Hubicka <jh@suse.cz>
33588 * ipa-devirt.c (maybe_record_node): Properly handle
33589 __cxa_pure_virtual visibility.
33591 2017-03-03 Martin Liska <mliska@suse.cz>
33593 PR tree-optimization/79803
33594 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33596 (pass_loop_prefetch::execute): Disabled optimization if an
33597 assumption about L1 cache size is not met.
33599 2017-03-03 Martin Liska <mliska@suse.cz>
33601 PR rtl-optimization/79574
33602 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33603 (hash_scan_set): Likewise.
33604 (dump_hash_table): Likewise.
33605 (hoist_code): Likewise.
33607 2017-03-03 Richard Biener <rguenther@suse.de>
33609 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33610 in place of uhigh ().
33611 (fixed_convert_from_real): Likewise.
33613 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33616 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33618 2017-03-03 Richard Biener <rguenther@suse.de>
33620 PR middle-end/79818
33621 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33622 TYPE_OVERFLOW_UNDEFINED check.
33624 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33626 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33628 (vector_ae_<mode>_p): Likewise.
33629 (vector_nez_<mode>_p): Likewise.
33630 (vector_ne_v2di_p): Likewise.
33631 (vector_ae_v2di_p): Likewise.
33632 (vector_ne_<mode>_p): Likewise.
33633 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33635 (vsx_tsqrt<mode>2_fe): Likewise.
33637 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33640 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33642 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33644 PR rtl-optimization/79780
33645 * cprop.c (one_cprop_pass): When second and further conditional trap
33646 in a single basic block is turned into an unconditional trap, turn it
33647 into a deleted note to avoid RTL verification failures.
33649 2017-03-02 Richard Biener <rguenther@suse.de>
33651 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33653 2017-03-02 Richard Biener <rguenther@suse.de>
33655 PR tree-optimization/79345
33657 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33658 param and abort the walk, returning -1 if it is hit.
33659 (walk_aliased_vdefs): Take a limit param and pass it on.
33660 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33661 defaulting to 0 and return a signed int.
33662 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33663 (check_defs): New helper.
33664 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33665 about uninitialized memory.
33666 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33667 bogus uninitialized warning.
33668 (fixed_convert_from_real): Likewise.
33670 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33672 PR tree-optimization/66768
33673 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33674 iv_use if base object can't be determined.
33676 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33678 PR tree-optimization/79345
33679 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33680 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33681 (get_pattern_stats): Initialize it.
33682 * genemit.c (gen_expand): Verify match_scratch numbers come after
33683 match_operand/match_dup numbers.
33684 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33685 match_scratch numbers.
33686 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33688 * config/s390/s390.md (trunctdsd2): Likewise.
33690 2017-03-02 Richard Biener <rguenther@suse.de>
33692 * wide-int.h (wide_int_storage::operator=): Implement in terms
33695 2017-03-02 Richard Biener <rguenther@suse.de>
33697 PR tree-optimization/79777
33698 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33699 the to insert expression to sth existing.
33701 2017-03-01 Martin Sebor <msebor@redhat.com>
33703 PR middle-end/79692
33704 * gimple-ssa-sprintf.c
33705 (directive::known_width_and_precision): New function.
33706 (format_integer): Use it.
33707 (get_mpfr_format_length): Consider the full range of precision
33708 when computing %g output with the # flag. Set the likely byte
33709 count to 3 rather than 1 when precision is indeterminate.
33710 (format_floating): Correct the lower bound of precision.
33712 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33714 * doc/invoke.texi: Document default code model for 64-bit Linux.
33716 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33719 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33720 udiv rather than div since input pattern is unsigned.
33722 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33724 * config/i386/i386.c (print_reg): Warn for values of
33725 unsupported size in integer register.
33727 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33730 * config/rs6000/predicates.md (current_file_function_operand): Do
33731 not allow self calls to be local if the function is replaceable.
33733 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33736 * config/rs6000/altivec.h (vec_ctz and others): Change the
33737 preprocessor macro that controls conditional compilation from
33738 _ARCH_PWR9 to __POWER9_VECTOR__.
33739 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33740 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33741 control (instead of _ARCH_PWR9 control) so that template
33742 definition uses power9-specific function.
33743 (vec_any_eq): Likewise.
33744 (vec_all_ne): Change macro definition to use a power9-specific
33745 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33746 _ARCH_PWR9 control).
33747 (vec_any_eq) Likewise.
33748 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33749 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33750 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33751 support for xvcmpnedp instruction.
33752 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33753 macro expansion so that Power9 implementation of vec_all_ne does
33754 not use the AltiVec predicate framework.
33755 (VCMPNEH_P): Likewise.
33756 (VCMPNEW_P): Likewise.
33757 (VCMPNED_P): Likewise.
33758 (VCMPNEFP_P): Likewise.
33759 (VCMPNEDP_P): Likewise.
33760 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33761 implementation of vec_any_eq to not use AltiVec predicate
33763 (VCMPAEH_P): Likewise.
33764 (VCMPAEW_P): Likewise.
33765 (VCMPAED_P): Likewise.
33766 (VCMPAEFP_P): Likewise.
33767 (VCMPAEDP_P): Likewise.
33768 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33769 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33770 not use the AltiVec predicate framework.
33771 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33772 of vec_any_eq to not use AltiVec predicate framework.
33773 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33774 support for predefined __POWER9_VECTOR__ macro to indicate that
33775 Power9 instruction selection is enabled.
33776 (altivec_overloaded_builtins): Remove extraneous
33777 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33778 function argument types RS6000_BTI_bool_V16QI and
33779 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33780 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33781 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33782 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33783 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33784 Power9 for implementations of vec_cmpne. Change the signature for
33785 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33786 (representing vec_all_ne) to remove the previously described first
33787 argument of type RS6000_BTI_INTSI, as this was an artifact of
33788 reliance on the AltiVec predicate framework, which is no longer
33789 used in the implementation of these functions. Add
33790 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33791 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33792 since, unlike the AltiVec predicate framework implementation, we
33793 do not share function descriptors between vec_alle and vec_anyeq.
33794 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33795 set of modes that receive special treatment even when
33796 TARGET_P9_VECTOR is true. The special treatment emits code that
33797 does not depend on Power9 instructions.
33798 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33799 define_expand to not rely on AltiVec predicate framework.
33800 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33802 (vector_ne_v2di_p): Change this define_expand to not rely on
33803 AltiVec predicate framework.
33804 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33806 (vector_ne_<mode>_p): Change this define_expand to not rely on
33807 AltiVec predicate framework.
33808 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33810 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33811 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33812 define_insn pattern.
33813 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33814 define_insn pattern because the xvcmpne<VSs>. instruction is not
33816 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33817 instruction is not supported.
33819 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33821 * config/nvptx/nvptx.c: Include intl.h.
33823 2017-03-01 Martin Jambor <mjambor@suse.cz>
33826 * ipa-prop.h (ipa_bits): Removed field known.
33827 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33828 to pointers. Adjusted their comments to warn about their sharing.
33829 (ipcp_transformation_summary): Change bits to a vector of pointers.
33830 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33831 (ipa_get_ipa_bits_for_value): Declare.
33832 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33833 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33834 (ipa_bits_hash_table): Likewise.
33835 (ipa_vr_ggc_hash_traits): Likewise.
33836 (ipa_vr_hash_table): Likewise.
33837 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33838 being pointers and vr_known being removed.
33839 (ipa_set_jf_unknown): Likewise.
33840 (ipa_get_ipa_bits_for_value): New function.
33841 (ipa_set_jfunc_bits): Likewise.
33842 (ipa_get_value_range): New overloaded functions.
33843 (ipa_set_jfunc_vr): Likewise.
33844 (ipa_compute_jump_functions_for_edge): Use the above functions to
33845 construct bits and vr parts of jump functions.
33846 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33847 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33849 (ipcp_grow_transformations_if_necessary): Also allocate
33850 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33852 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33853 them. Fix too long lines.
33854 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33855 vr_known being removed.
33856 (ipa_read_jump_function): Use new setter functions to construct bits
33857 and vr parts of jump functions or set them to NULL.
33858 (write_ipcp_transformation_info): Adjust for bits being pointers.
33859 (read_ipcp_transformation_info): Likewise.
33860 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33862 Include gt-ipa-prop.h.
33863 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33865 (ipcp_store_bits_results): Likewise.
33866 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33867 Do not write to existing jump functions but use a temporary instead.
33869 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33872 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33873 attempt to use its first operand as BIT_FIELD_REF base.
33875 2017-03-01 Richard Biener <rguenther@suse.de>
33877 PR middle-end/79721
33878 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33879 interpolating formula in wrapping arithmetic.
33880 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33882 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33884 PR tree-optimization/79734
33885 * tree-vect-generic.c (expand_vector_condition): Optimize
33886 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33887 Handle VEC_COND_EXPR where comparison has different inner width from
33888 type's inner width.
33890 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33892 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33893 markup, and similar issues. Remove @opindex entries for things
33894 that aren't options. Add missing -mmpy-option entries.
33896 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33898 PR tree-optimization/79737
33899 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33900 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33901 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33902 instead of byte_size. Formatting fix.
33903 (shift_bytes_in_array_right): Formatting fix.
33905 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33908 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33909 condition on optimize for the leaf function test.
33911 2017-02-28 Martin Liska <mliska@suse.cz>
33914 * read-rtl-function.c (function_reader::handle_unknown_directive):
33915 Bail out when one uses -flto.
33917 2017-02-28 Martin Liska <mliska@suse.cz>
33919 * common.opt: Replace space with tabular for options of <number>
33921 * config/i386/i386.opt: Show <number> value for
33922 -mlarge-data-threshold.
33923 * opts.c (print_filtered_help): Do not display number in hexadecimal
33926 2017-02-28 Martin Liska <mliska@suse.cz>
33928 * common.opt: Fix --help=option -Q for options which are of
33931 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33933 * config/i386/i386.c (print_reg): Error out for values
33934 of 8-bit size in invalid integer register.
33936 2017-02-28 Martin Sebor <msebor@redhat.com>
33938 PR tree-optimization/79691
33939 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33941 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33944 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33945 gcc_unreachable with output_operand_lossage.
33947 2017-02-28 Richard Biener <rguenther@suse.de>
33949 PR tree-optimization/79740
33950 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33952 (visit_nary_op): Insert the nary into the hashtable if we
33953 pattern-matched sth.
33954 * tree-ssa-pre.c (eliminate_insert): Robustify.
33956 2017-02-28 Richard Biener <rguenther@suse.de>
33958 PR middle-end/79731
33959 * fold-const.c (decode_field_reference): Reject out-of-bound
33962 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33964 * config/i386/i386.c: Include intl.h.
33965 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33966 instead of just cond ? "..." : "...".
33967 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33968 * coverage.c (read_counts_file): Likewise.
33969 * omp-offload.c: Include intl.h.
33970 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33971 of just cond ? "..." : "...".
33972 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33973 of just cond ? "..." : "...".
33975 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33978 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33980 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33981 'tune for' CPU name.
33982 * config/arm/arm-cpu-data.h: Regenerated.
33984 2017-02-28 Richard Biener <rguenther@suse.de>
33986 PR tree-optimization/79732
33987 * tree-inline.c (expand_call_inline): Do not shadow var.
33989 2017-02-28 Richard Biener <rguenther@suse.de>
33991 PR tree-optimization/79723
33992 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33993 address-space properly.
33995 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33997 * doc/optinfo.texi (Optimization groups): Fix option used for
33999 * doc/invoke.texi (-fopt-info): Document "omp".
34000 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34001 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34002 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34003 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34004 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34006 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34008 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34009 OPTGROUP_OMP with "omp".
34011 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34014 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34015 for arithmetic shift of unsigned V2DI.
34017 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34019 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34020 arc/linux.h headers.
34021 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34022 (LINK_SPEC): Likewise.
34023 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34024 (EXTRA_SPECS): Likewise.
34025 (STARTFILE_SPEC): Likewise.
34026 (ENDFILE_SPEC): Likewise.
34027 (LIB_SPEC): Likewise.
34028 (TARGET_SDATA_DEFAULT): Likewise.
34029 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34030 (MULTILIB_DEFAULTS): Likewise.
34031 (DWARF2_UNWIND_INFO): Likewise.
34032 * config/arc/big.h: New file.
34033 * config/arc/elf.h: Likewise.
34034 * config/arc/linux.h: Likewise.
34035 * config/arc/t-uClibc: Remove.
34037 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34039 PR tree-optimization/77536
34040 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34041 (tree_transform_and_unroll_loop): Use above function to compute the
34042 estimated niter of unrolled loop and use it when scaling profile.
34043 Also use count info rather than frequency if it's non-zero.
34044 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34045 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34046 (vect_transform_loop): Call above function.
34048 2017-02-27 Richard Biener <rguenther@suse.de>
34050 PR tree-optimization/45397
34051 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34052 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34053 (visit_nary_op): Add pattern matching for CSEing sign-changed
34054 or truncated operations with wider ones.
34056 2017-02-27 Richard Biener <rguenther@suse.de>
34058 PR tree-optimization/79690
34059 * tree-vect-stmts.c (vectorizable_store): Use vector type
34060 built from the DR with address-space.
34062 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34064 * doc/invoke.texi (Optimize Options): Refine the description
34065 of asan-use-after-return.
34067 2017-02-25 Alan Modra <amodra@gmail.com>
34069 PR rtl-optimization/79584
34070 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34071 base, not ad->base_term, the reg within base. Remove assertion
34072 that ad->base == ad->base_term. Replace gen_int_mode using
34073 bogus mode with const0_rtx.
34075 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34077 PR middle-end/79396
34078 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34079 FMA_EXPR like tcc_binary or tcc_unary.
34081 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34084 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34086 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34087 (output_loc_operands): Handle DW_OP_call_ref and
34088 DW_OP_GNU_variable_value.
34089 (struct variable_value_struct): New type.
34090 (struct variable_value_hasher): Likewise.
34091 (variable_value_hash): New variable.
34092 (string_types): Remove.
34093 (copy_loc_descr): New function.
34094 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34095 (prepend_loc_descr_to_each): New function.
34096 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34097 instead of add_loc_descr_to_each if the first argument is single
34098 location list and the second has multiple.
34099 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34100 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34101 when looking for variable value which doesn't have other location info.
34102 (loc_list_from_tree): Formatting fix.
34103 (gen_array_type_die): Simplify DW_AT_string_length handling.
34104 (adjust_string_types): Remove.
34105 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34106 string_types. Call resolve_variable_values.
34107 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34108 (resolve_addr_in_expr): Likewise. Add A argument.
34109 (copy_deref_exprloc): Remove deref argument. Adjust for the
34110 original expression being DW_OP_GNU_variable_value with optionally
34111 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34112 optionally after it.
34113 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34114 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34115 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34116 (variable_value_hasher::hash, variable_value_hasher::equal): New
34118 (resolve_variable_value_in_expr, resolve_variable_value,
34119 resolve_variable_values, note_variable_value_in_expr,
34120 note_variable_value): New functions.
34121 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34124 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34127 * opts.h (handle_generated_option): Add GENERATED_P argument.
34128 * opts-common.c (handle_option): Adjust function comment.
34129 (handle_generated_option): Add GENERATED_P argument, pass it to
34131 (control_warning_option): Pass false to handle_generated_option
34133 * opts.c (maybe_default_option): Pass true to handle_generated_option
34135 * optc-gen.awk: Likewise.
34137 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34139 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34140 a REG, look at the REG it is a SUBREG of.
34141 (splitter for cmpeqsi_t): Ditto.
34143 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34145 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34146 the special USEs with the pattern of the insn, not the insn itself.
34148 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34151 * doc/invoke.texi: Document -mload-store-pairs.
34153 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34154 Sandra Loosemore <sandra@codesourcery.com>
34156 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34157 argument isn't a CONST_INT.
34158 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34159 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34160 (nios2_validate_compare): Bypass alternate compare logic if *op2
34161 is not a CONST_INT.
34162 (ldstwm_operation_p): Return false if first_base is not a REG or
34163 if first_offset is not a CONST_INT.
34165 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34167 * config/cris/cris.md: Use correct operand in a define_peephole2.
34169 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34171 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34173 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34175 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34176 this_insn if it is an INSN or JUMP_INSN.
34177 (force_offsettable): Look at base, not at addr.
34178 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34179 on things that aren't necessarily CONST_INTs.
34181 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34183 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34184 -mfpmath=sse is the default also for x86-32 targets with SSE2
34185 instruction set when @option{-ffast-math} is enabled
34187 2017-02-24 Jeff Law <law@redhat.com>
34189 PR rtl-optimizatoin/79286
34190 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34193 2017-02-24 Richard Biener <rguenther@suse.de>
34195 PR tree-optimization/79389
34196 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34199 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34201 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34202 function comment to reflect reality.
34203 (loop_exits_before_overflow): Fix typo in function description.
34205 2017-02-24 Richard Biener <rguenther@suse.de>
34207 PR tree-optimization/79389
34208 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34209 properly that a threading opportunity exists. Detect conditional
34210 copy/constant propagation opportunities.
34212 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34214 * config/visium/visium.md (type): Add trap.
34215 (b): New mode attribute.
34216 (*btst): Rename into...
34217 (*btst<mode>): ...this and adjust.
34218 (*cbranchsi4_btst_insn): Rename into...
34219 (*cbranch<mode>4_btst_insn): ...this and adjust.
34220 (trap): New define_insn.
34222 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34224 PR tree-optimization/79389
34225 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34226 (noce_reversed_cond_code): New function.
34227 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34228 reversed_comparison_code. Formatting fix.
34229 (noce_try_store_flag): Test rev_cond != NULL in addition to
34230 reversed_comparison_code.
34231 (noce_try_store_flag_constants): Likewise.
34232 (noce_try_store_flag_mask): Likewise.
34233 (noce_try_addcc): Use rev_cond if non-NULL instead of
34234 reversed_comparison_code.
34235 (noce_try_cmove_arith): Likewise. Formatting fixes.
34236 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34237 (noce_find_if_block): Initialize rev_cond.
34238 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34239 instead of false as last argument never attempt to reverse it
34242 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34244 PR tree-optimization/79663
34245 * tree-predcom.c (combine_chains): Process refs in reverse order
34246 only for ZERO length chains, and add explaining comment.
34248 2017-02-23 Jeff Law <law@redhat.com>
34250 PR tree-optimization/79578
34251 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34252 in call to operand_equal_p.
34254 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34257 * config/i386/cpuid.h: Fix another undefined behavior.
34259 2017-02-23 Richard Biener <rguenther@suse.de>
34261 PR tree-optimization/79683
34262 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34263 vector types for data-refs.
34265 2017-02-23 Martin Liska <mliska@suse.cz>
34267 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34269 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34271 PR middle-end/79665
34272 * internal-fn.c (get_range_pos_neg): Moved to ...
34273 * tree.c (get_range_pos_neg): ... here. No longer static.
34274 * tree.h (get_range_pos_neg): New prototype.
34275 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34276 are known to be in between 0 and signed maximum inclusive, try to
34277 expand both unsigned and signed divmod and use the cheaper one from
34280 2017-02-22 Jeff Law <law@redhat.com>
34282 PR tree-optimization/79578
34283 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34284 to compare base operands.
34286 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34289 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34290 gpc_reg_operand instead of fpr_reg_operand.
34292 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34294 * config/mips/mips.c (mips_return_in_memory): Force FP
34295 vector types to be returned in memory for o32 ABI.
34297 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34299 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34300 instead of DW_TAG_member for static data member declarations and don't
34301 set no_linkage_name for static inline data members.
34302 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34305 2017-02-22 Martin Liska <mliska@suse.cz>
34307 * doc/invoke.texi: Replace inequality signs with square brackets
34310 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34313 * lra-constraints.c (simplify_operand_subreg): Handle
34314 WORD_REGISTER_OPERATIONS targets.
34316 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34319 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34320 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34321 elimination by swapping fld*.
34323 2017-02-22 Richard Biener <rguenther@suse.de>
34325 PR tree-optimization/79673
34326 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34327 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34328 irrelevant address-space qualifiers and avoiding a
34329 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34331 2017-02-22 Richard Biener <rguenther@suse.de>
34333 PR tree-optimization/79666
34334 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34335 to not symbolically negate if that may introduce undefined
34338 2017-02-22 Martin Liska <mliska@suse.cz>
34341 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34342 * data-streamer-out.c (streamer_write_gcov_count_stream):
34344 * value-prof.c (stream_out_histogram_value): Make assert more
34345 precise based on type of counter.
34347 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34350 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34351 Use nonimmediate_operand instead of memory_operand for operand 1.
34352 (float-extend standard_x87sse_constant_load splitter): Ditto.
34354 2017-02-21 Jeff Law <law@redhat.com>
34356 PR tree-optimization/79621
34357 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34358 blocks with edges to themselves.
34360 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34363 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34364 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34365 Use gimple_call_builtin_p.
34368 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34369 on temporarily removed DEBUG_INSNs.
34371 PR tree-optimization/79649
34372 * tree-loop-distribution.c (classify_partition): Give up on
34373 non-generic address space loads/stores.
34375 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34377 * doc/loop.texi (Loop manipulation): Remove nonexistent
34378 tree_ssa_loop_version from the documentation.
34379 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34381 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34384 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34385 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34386 * config/rs6000/rs6000.c: Include except.h.
34387 (rs6000_expand_split_stack_prologue): Call
34388 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34390 2017-02-21 Martin Jambor <mjambor@suse.cz>
34393 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34394 have been analyzed.
34396 2017-02-21 Martin Jambor <mjambor@suse.cz>
34398 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34399 for backward compatibility only.
34400 * doc/invoke.texi (Option Summary): Remove all references to
34401 -fipa-cp-alignment.
34403 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34407 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34409 * lra-constraints.c (curr_insn_transform): Handle
34410 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34412 2017-02-21 Martin Liska <mliska@suse.cz>
34414 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34416 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34418 PR translation/79638
34419 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34421 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34424 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34425 (arm_function_ok_for_sibcall): Return false for an indirect call by
34426 descriptor if all the argument registers are used.
34427 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34428 alignment of the function.
34430 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34432 PR tree-optimization/61441
34433 * simplify-rtx.c (simplify_const_unary_operation): For
34434 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34435 the sNaN unmodified.
34437 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34439 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34440 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34441 instead of SYSTEM_HEADER_DIR.
34443 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34444 Martin Liška <mliska@suse.cz>
34446 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34447 Fix typos and grammar, use active voice, and clarify.
34449 2017-02-20 Marek Polacek <polacek@redhat.com>
34451 PR middle-end/79537
34452 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34455 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34457 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34460 * config/i386/i386.c (ix86_expand_builtin): Handle
34461 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34462 ix86_builtins_isa[fcode].isa as a requirement of those
34463 flags and any other flag in the bitmask.
34464 (ix86_init_mmx_sse_builtins): Use 0 instead of
34465 ~OPTION_MASK_ISA_64BIT as mask.
34466 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34467 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34468 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34469 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34471 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34474 * lra-constraints.c (split_reg): Check requested split mode
34475 is supported by the register.
34477 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34479 * lra-constraints.c (simplify_operand_subreg): Remove early
34482 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34485 * lra-constraints.c (curr_insn_transform): Tighten condition
34486 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34488 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34491 * lra-constraints.c (curr_insn_transform): Handle
34492 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34494 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34497 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34499 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34501 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34504 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34507 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34509 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34512 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34514 * final.c (last_columnnum, override_columnnum): New variables.
34515 (final_start_function): Set last_columnnum, pass it to begin_prologue
34516 hook and pass 0 to dwarf2out_begin_prologue.
34517 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34518 to source_line debug hook.
34519 (notice_source_line): Compute last_columnnum and for debug_column_info
34520 return true on column changes.
34521 * debug.h (struct gcc_debug_hooks): Add column argument to
34522 source_line and begin_prologue hooks.
34523 (debug_nothing_int_charstar_int_bool): Remove prototype.
34524 (debug_nothing_int_int_charstar,
34525 debug_nothing_int_int_charstar_int_bool): New prototypes.
34526 (dwarf2out_begin_prologue): Add column argument.
34527 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34528 begin_prologue hooks.
34529 (debug_nothing_int_charstar_int_bool): Remove.
34530 (debug_nothing_int_int_charstar,
34531 debug_nothing_int_int_charstar_int_bool): New functions.
34532 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34533 through to dwarf2out_source_line.
34534 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34535 (dwarf2out_source_line): Add column argument, emit it if requested.
34536 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34538 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34539 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34540 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34541 through to dwarf2out_begin_prologue.
34542 (vmsdbgout_source_line): Add column argument, pass it through to
34543 dwarf2out_source_line.
34544 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34545 dbxout_source_line caller.
34546 (dbxout_source_line): Add column argument.
34548 * common.opt (gno-column-info, gcolumn-info): New options.
34549 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34550 (check_die): Also test for multiple DW_AT_decl_column attributes.
34551 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34552 DW_AT_decl_column if requested.
34553 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34555 (gen_variable_die): Likewise.
34556 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34557 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34560 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34561 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34562 (ix86_handle_option): Handle OPT_m3dnowa.
34563 * doc/invoke.texi (-m3dnowa): Document.
34564 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34565 -m3dnowa instead of -m3dnow -march=athlon.
34568 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34569 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34571 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34574 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34575 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34576 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34577 generator for vsx_xxpermdi_<mode>_be.
34578 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34579 force big-endian semantics.
34580 (vsx_xxpermdi_<mode>_be): New define_expand with same
34581 implementation as previous version of vsx_xxpermdi_<mode>.
34583 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34585 PR tree-optimization/79327
34586 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34587 variable, its initialization and use.
34589 2017-02-17 Julia Koval <julia.koval@intel.com>
34591 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34592 (OPTION_MASK_ISA_PKU_UNSET): New.
34593 (ix86_handle_option): Handle -mrdpid.
34594 * config/i386/cpuid.h (bit_RDPID): New.
34595 * config/i386/driver-i386.c (host_detect_local_cpu):
34596 Detect RDPID feature.
34597 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34598 * config/i386/i386-c.c (ix86_target_macros_internal):
34600 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34601 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34602 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34603 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34604 * config/i386/i386.md (define_insn "rdpid"): New.
34605 * config/i386/i386.opt Add -mrdpid.
34606 * config/i386/immintrin.h (_rdpid_u32): New.
34608 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34610 PR rtl-optimization/79541
34611 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34612 instead of transforming it into USE.
34614 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34616 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34617 If HONOR_SNANS (SFmode) force the input to a register.
34618 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34619 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34620 an frsp or similar insn.
34622 2017-02-17 Martin Liska <mliska@suse.cz>
34624 PR rtl-optimization/79577
34625 * params.def (selsched-max-sched-times): Increase minimum to 1.
34627 2017-02-17 Martin Liska <mliska@suse.cz>
34629 PR rtl-optimization/79574
34630 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34632 2017-02-17 Martin Liska <mliska@suse.cz>
34634 PR tree-optimization/79529
34635 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34636 ssa_defined_default_def_p to handle cases which are implicitly
34638 * tree-ssa.c (ssa_defined_default_def_p): New function.
34639 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34640 which are implicitly defined.
34641 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34643 2017-02-17 Richard Biener <rguenther@suse.de>
34645 PR middle-end/79576
34646 * params.def (max-ssa-name-query-depth): Limit to 10.
34648 2017-02-17 Richard Biener <rguenther@suse.de>
34650 PR tree-optimization/79552
34651 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34654 2017-02-17 Richard Biener <rguenther@suse.de>
34657 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34659 2017-02-17 Marek Polacek <polacek@redhat.com>
34661 PR middle-end/79536
34662 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34663 (fold_negate_expr): New wrapper.
34665 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34667 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34668 Correct terminology and de-emphasize pre-standard behavior.
34670 2017-02-16 Alan Modra <amodra@gmail.com>
34672 PR rtl-optimization/79286
34673 * ira.c (def_dominates_uses): New function.
34674 (update_equiv_regs): Don't create an equivalence for insns that
34675 may trap where the register def does not dominate the use.
34677 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34679 PR rtl-optimization/78127
34680 * lra.c (lra): Call lra_eliminate before finish the loop after
34683 2017-02-16 Richard Biener <rguenther@suse.de>
34685 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34687 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34688 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34689 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34690 (isl_val_int_from_wi): New function.
34691 (extract_affine_gmp): Rename to ...
34692 (extract_affine_wi): ... this, take a widest_int.
34693 (extract_affine_int): Just wrap extract_affine_wi.
34694 (add_param_constraints): Use isl_val_int_from_wi.
34695 (add_loop_constraints): Likewise, and extract_affine_wi.
34697 2017-02-15 Jeff Law <law@redhat.com>
34699 PR middle-end/79521
34700 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34701 ira_init_register_move_cost_if_necessary.
34703 2017-02-15 Martin Sebor <msebor@redhat.com>
34705 PR middle-end/32003
34706 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34707 removed in a prior commit.
34709 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34711 PR tree-optimization/79347
34712 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34713 counters during peeling.
34715 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34717 * Makefile.in (site.exp): Remove "set ISLVER".
34719 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34722 * real.c (real_from_integer): Call real_convert even for decimal.
34724 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34727 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34729 2017-02-14 Andrew Pinski <apinski@cavium.com>
34731 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34732 cores and change the partno/implementer to be correct.
34733 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34734 the 'B" as the implementer.
34735 * config/aarch64/aarch64-tune.md: Regenerate.
34737 2017-02-14 Carl Love <cel@us.ibm.com>
34739 * config/rs6000/rs6000.c: Add case statement entry to make the
34740 xvcvuxdsp built-in argument unsigned.
34741 * config/rs6000/vsx.md: Fix the source and return operand types so they
34742 match the instruction definitions from the ISA document. Fix typo
34743 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34746 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34749 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34750 member early_clobber_alts.
34751 * lra-lives.c (reg_early_clobber_p): New.
34752 (process_bb_lives): Use it.
34753 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34754 (debug_operand_data): Initialize early_clobber_alts.
34755 (setup_operand_alternative): Set up early_clobber_alts.
34756 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34757 alternatives to new_insn_reg.
34758 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34760 (lra_update_insn_regno_info): Pass the new arg.
34762 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34764 PR middle-end/79505
34765 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34766 (new_oacc_loop_raw): Don't clear already cleared fields.
34769 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34770 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34771 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34773 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34776 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34778 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34781 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34782 the extra instruction to the right place to store 128-bit constant
34785 2017-02-14 Martin Sebor <msebor@redhat.com>
34787 PR middle-end/79448
34788 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34789 warning for strings of unknown length.
34791 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34793 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34795 2017-02-14 Jeff Law <law@redhat.com>
34798 * ira-costs.c (scan_one_insn): Initialize register move costs
34799 for pseudos seen in USE/CLOBBER insns.
34801 PR tree-optimization/79095
34802 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34803 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34804 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34805 if the operands are known to be not equal, then the resulting range
34807 (intersect_ranges): If the new range is ~[0,0] and the old range is
34808 wide, then prefer ~[0,0].
34809 * tree-vrp.c (overflow_comparison_p_1): New function.
34810 (overflow_comparison_p): New function.
34811 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34812 if NAME is used in an overflow test.
34813 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34814 overflow check that can be expressed as an equality test, then adjust
34815 ops to be that equality test.
34817 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34819 * config/s390/s390-builtin-types.def: Remove flags argument.
34820 * config/s390/s390.c (s390_init_builtins): Likewise.
34822 2017-02-14 Martin Liska <mliska@suse.cz>
34824 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34825 vector. Fix trailing white spaces.
34827 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34829 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34832 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34834 PR rtl-optimization/68664
34835 * config/arm/arm.c (arm_sched_can_speculate_insn):
34836 New function. Declare prototype.
34837 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34839 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34841 PR rtl-optimization/68664
34842 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34844 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34846 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34848 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34849 max skip bytes for function, loop and jump.
34851 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34853 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34854 ABS_EXPR for gimple dump.
34856 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34859 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34861 PR tree-optimization/79408
34862 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34863 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34864 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34865 also if rhs1 is INTEGER_CST.
34867 2017-02-14 Richard Biener <rguenther@suse.de>
34869 PR middle-end/79432
34870 * tree-into-ssa.c (insert_phi_nodes): When the function can
34871 have abnormal edges rewrite SSA names with broken use-def
34872 dominance out of SSA and register them for PHI insertion.
34874 2017-02-13 Martin Sebor <msebor@redhat.com>
34876 PR middle-end/79496
34877 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34878 clearing info.nowrite flag when snprintf size argument is a range.
34880 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34882 * cprop.c (cprop_jump): Add missing space in string literal.
34883 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34884 (get_constraint_for_component_ref): Likewise.
34885 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34886 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34887 * lra-constraints.c (process_alt_operands): Likewise.
34888 * ipa-inline.c (inline_small_functions): Likewise.
34889 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34890 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34891 * trans-mem.c (diagnose_tm_1_op): Likewise.
34892 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34893 (grid_parallel_clauses_gridifiable): Likewise.
34895 * config/nvptx/mkoffload.c (process): Add space in between
34898 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34899 "MOD4_SSE_REGS" and "ALL_REGS".
34901 * spellcheck.c (test_data): Add , in between "foo" and "food".
34903 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34906 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34907 boundary crossing check and subsequent code generation agree.
34909 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34911 * config/aarch64/aarch64.c (has_memory_op): Delete.
34912 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34915 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34917 PR rtl-optimization/79388
34918 PR rtl-optimization/79450
34919 * combine.c (distribute_notes): When removing TEM_INSN for which
34920 corresponding dest has last value recorded, invalidate that last
34923 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34925 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34926 of explicit '@'. Add missing assembly comment marker on branch costs
34929 2017-02-13 Nathan Sidwell <nathan@acm.org>
34931 * gengtype-lex.l (<in_struct>): Add '/'.
34933 2017-02-13 Martin Liska <mliska@suse.cz>
34936 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34938 2017-02-13 Richard Biener <rguenther@suse.de>
34940 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34942 * configure: Re-generate.
34943 * config.in: Likewise.
34944 * graphite-dependences.c: Simplify as if
34945 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34946 * graphite-isl-ast-to-gimple.c: Likewise.
34947 * graphite-optimize-isl.c: Likewise.
34948 * graphite-poly.c: Likewise.
34949 * graphite-sese-to-poly.c: Likewise.
34950 * graphite.h: Likewise.
34951 * toplev.c: Include isl/version.h and use isl_version () for
34952 printing the ISL version.
34953 * doc/install.texi: Update ISL requirement.
34955 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34957 * doc/standards.texi (Standards): Update reference to
34960 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34962 * doc/extend.texi (Named Address Spaces): sourceware.org now
34964 * doc/install.texi (Binaries): Ditto.
34967 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34969 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34970 standard terminology "stringize"/"stringizing" throughout.
34971 * doc/cppinternals.texi: Likewise.
34973 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34975 * doc/extend.texi: Fix some spelling mistakes and typos.
34976 * doc/invoke.texi: Likewise.
34978 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34981 * params.def (inline-min-speedup) Change from 10 to 8.
34983 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34985 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34988 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34991 * ipa-inline-analysis.c (get_minimal_bb): New function.
34992 (record_modified): Use it.
34993 (remap_edge_change_prob): Handle also ancestor functions.
34995 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34997 * doc/contrib.texi (Contributors): Remove broken link into
34998 the Mauve CVS repository.
35000 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35002 PR middle-end/79454
35003 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35004 result computation whenever lhs doesn't have vector mode, not
35005 just when it has BLKmode.
35007 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35009 * doc/makefile.texi (profiledbootstrap): Refer to the
35010 installation instructions only in textual form.
35012 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35015 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35017 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35019 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35020 (Specific): Update mingw-w64 reference.
35022 (Specific): Remove broken link to Renesas RX processor.
35024 2017-02-10 Richard Biener <rguenther@suse.de>
35026 * toplev.c (process_options): Do not mention obsolete graphite
35027 options when printing sorry message about missing graphite support.
35028 Mention -floop-nest-optimize.
35030 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35032 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35033 (vtst_p16): Likewise.
35034 (vtstq_p8): Likewise.
35035 (vtstq_p16): Likewise.
35037 (vtstq_p64): Likewise.
35038 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35039 (vset_lane_p64): New.
35040 (vsetq_lane_p64): New.
35042 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35044 PR tree-optimization/79411
35045 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35046 stmt operands are SSA_NAMEs used in abnormal phis.
35047 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35050 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35053 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35056 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35058 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35060 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35062 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35066 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35067 not arbitrary TREE_CONSTANT.
35070 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35071 "omp declare target link" attribute unless is_global_var.
35072 * omp-offload.c (find_link_var_op): Likewise.
35074 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35075 Chung-Lin Tang <cltang@codesourcery.com>
35077 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35079 (gimplify_adjust_omp_clauses): Don't delete TILE.
35080 (gimplify_omp_for): Deal with TILE.
35081 * internal-fn.c (expand_GOACC_TILE): New function.
35082 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35084 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35085 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35087 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35088 avoid DIV for outermost collapse var.
35089 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35090 Remove out of date comments, fix whitespace.
35091 * omp-general.c (omp_extract_for_data): Deal with tiling.
35092 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35093 adjust OLF_DIM_BASE value.
35094 (struct omp_for_data): Add tiling field.
35095 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35096 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35097 for auto loops. Remove default auto determining, moved to
35098 oacc_loop_fixed_partitions.
35099 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35100 stmts, add e_mask field.
35101 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35102 (oacc_thread_numbers): Use oacc_dim_call.
35103 (oacc_xform_tile): New.
35104 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35105 (finish_oacc_loop): Adjust for ifns vector.
35106 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35107 add case for GOACC_TILE fns.
35108 (oacc_loop_xform_loop): Delete.
35109 (oacc_loop_process): Iterate over call list directly, and add
35110 handling for GOACC_TILE fns.
35111 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35113 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35114 vector partitioning to outer loops. Assign 2 partitions to loops
35115 when available. Add TILE handling.
35116 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35117 (execite_oacc_device_lower): Process GOACC_TILE fns,
35118 ignore unknown specs.
35119 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35120 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35121 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35123 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35125 * configure.ac (ACX_BUGURL): Update.
35126 * configure: Regenerate.
35128 2017-02-09 Richard Biener <rguenther@suse.de>
35130 PR tree-optimization/69823
35131 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35132 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35134 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35136 * config/arc/arc-c.def: Add __NPS400__ definition.
35137 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35138 (TARGET_NPS400): Define.
35140 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35142 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35144 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35145 pointer, arch_info.
35146 (arc_cpu_types): Fill the arch_info field with a pointer into the
35147 arc_arch_types table.
35148 (arc_selected_cpu): Declare.
35149 * config/arc/arc.c (arc_selected_cpu): Make global.
35150 (arc_selected_arch): Delete.
35151 (arc_base_cpu): Delete.
35152 (arc_override_options): Remove references to deleted variables,
35153 update access to arch information.
35154 (ARC_OPT): Update access to arch information.
35155 (ARC_OPTX): Likewise.
35156 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35157 (TARGET_ARC600): Update access to arch information.
35158 (TARGET_ARC601): Likewise.
35159 (TARGET_ARC700): Likewise.
35160 (TARGET_EM): Likewise.
35161 (TARGET_HS): Likewise.
35162 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35165 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35168 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35169 condition/operands for integer GE/LE/GEU/LEU operations.
35171 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35173 PR translation/79397
35174 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35177 2017-02-08 Martin Jambor <mjambor@suse.cz>
35180 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35181 whether allocation happened.
35182 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35183 nothing was allocated.
35185 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35187 PR tree-optimization/79408
35188 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35189 constant, but SSA_NAME with a known integer range, use the minimum
35190 of that range instead of op1 to determine if modulo can be replaced
35191 with its first operand.
35193 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35195 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35197 2017-02-08 Richard Biener <rguenther@suse.de>
35199 PR tree-optimization/71824
35200 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35201 Check all loops contained in the merged region.
35203 2017-02-07 Andrew Pinski <apinski@cavium.com>
35205 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35207 2017-02-07 Andrew Pinski <apinski@cavium.com>
35209 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35210 (thunderxt88): Likewise.
35211 (thunderxt81): Disable LSE and change v8.1 to v8.
35212 (thunderxt83): Likewise.
35214 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35215 Richard Biener <rguenther@suse.de>
35217 PR middle-end/79399
35218 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35219 type from int to size_t.
35220 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35222 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35224 PR rtl-optimization/79386
35225 * cprop.c (bypass_conditional_jumps): Initialize
35226 bypass_last_basic_block already before splitting bbs after
35227 unconditional traps...
35228 (bypass_conditional_jumps): ... rather than here.
35231 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35232 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35233 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35234 fix -masm=intel patterns.
35236 2017-02-07 Richard Biener <rguenther@suse.de>
35238 PR tree-optimization/79256
35239 PR middle-end/79278
35240 * builtins.c (get_object_alignment_2): Use min_align_of_type
35241 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35242 and ADJUST_FIELD_ALIGN.
35244 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35246 * doc/tm.texi: Regenerate.
35247 * stor-layout.c (layout_decl): Adjust.
35248 (update_alignment_for_field): Likewise.
35249 (place_field): Likewise.
35250 (min_align_of_type): Likewise.
35251 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35252 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35253 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35254 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35255 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35256 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35257 * config/i386/i386.c (x86_field_alignment): Likewise.
35258 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35259 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35260 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35261 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35262 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35263 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35267 2017-01-30 Richard Biener <rguenther@suse.de>
35269 PR tree-optimization/79256
35270 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35271 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35274 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35276 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35277 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35278 builtins to SImode and emit a zero-extend, if necessary.
35280 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35282 * docs/invoke.texi (RISC-V Options): Alphabetize.
35284 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35286 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35289 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35291 * config/riscv/riscv.c: New file.
35292 * common/config/riscv/riscv-common.c: Likewise.
35293 * config.gcc: Likewise.
35294 * config/riscv/constraints.md: Likewise.
35295 * config/riscv/elf.h: Likewise.
35296 * config/riscv/generic.md: Likewise.
35297 * config/riscv/linux.h: Likewise.
35298 * config/riscv/multilib-generator: Likewise.
35299 * config/riscv/peephole.md: Likewise.
35300 * config/riscv/pic.md: Likewise.
35301 * config/riscv/predicates.md: Likewise.
35302 * config/riscv/riscv-builtins.c: Likewise.
35303 * config/riscv/riscv-c.c: Likewise.
35304 * config/riscv/riscv-ftypes.def: Likewise.
35305 * config/riscv/riscv-modes.def: Likewise.
35306 * config/riscv/riscv-opts.h: Likewise.
35307 * config/riscv/riscv-protos.h: Likewise.
35308 * config/riscv/riscv.h: Likewise.
35309 * config/riscv/riscv.md: Likewise.
35310 * config/riscv/riscv.opt: Likewise.
35311 * config/riscv/sync.md: Likewise.
35312 * config/riscv/t-elf-multilib: Likewise.
35313 * config/riscv/t-linux: Likewise.
35314 * config/riscv/t-linux-multilib: Likewise.
35315 * config/riscv/t-riscv: Likewise.
35316 * configure.ac: Likewise.
35317 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35318 Waterman as RISC-V maintainers.
35319 * doc/install.texi: Add RISC-V entries.
35320 * doc/invoke.texi: Add RISC-V options section.
35321 * doc/md.texi: Add RISC-V constraints section.
35322 * configure: Regenerated.
35324 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35327 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35328 false values to be constant vectors with all 0 or all 1 bits set.
35329 (vcondu<mode><mode>): Likewise.
35330 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35332 (fpmask_comparison_operator): Update comment.
35333 (vecint_comparison_operator): New predicate.
35334 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35335 vector conditionals when the true and false values are constant
35336 vectors with all 0 bits or all 1 bits set.
35338 2017-02-06 Martin Sebor <msebor@redhat.com>
35340 PR tree-optimization/79376
35341 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35343 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35345 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35346 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35347 to simplify split condition.
35349 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35351 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35352 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35355 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35357 PR rtl-optimization/68664
35358 * target.def (can_speculate_insn): New hook.
35359 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35360 * doc/tm.texi: Regenerate.
35361 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35362 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35363 (rs6000_sched_can_speculate_insn): New function.
35365 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35367 PR tree-optimization/79284
35368 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35369 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35370 vectorizable_mask_load_store, vectorizable_operation,
35371 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35372 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35373 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35374 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35375 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35376 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35377 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35378 is_gimple_assign (stmt). Replace another such test with
35379 is_gimple_assign (stmt).
35381 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35384 * config/avr/avr.c (rtl-iter.h): Include it.
35385 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35386 (avr_legitimate_combined_insn): ...and implementation.
35388 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35390 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35391 * config/s390/s390.c (s390_const_operand_ok)
35392 (s390_canonicalize_comparison, s390_extract_part)
35393 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35394 (s390_contiguous_bitmask_p, s390_rtx_costs)
35395 (legitimize_pic_address): Likewise.
35396 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35397 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35398 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35399 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35400 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35402 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35404 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35405 REGNO($0) == REGNO($1).
35407 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35409 * config/s390/linux.h(SIZE_TYPE): Add comment.
35411 2017-02-06 Julian Brown <julian@codesourcery.com>
35412 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35413 Virendra Pathak <virendra.pathak@broadcom.com>
35415 * config/aarch64/aarch64-cores.def: Change the scheduler
35417 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35418 * config/aarch64/thunderx2t99.md: New file.
35420 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35422 * doc/standards.texi (Go Language): Update link to language
35425 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35427 * tree-eh.c (lower_resx): Sanitize profile.
35428 (cleanup_empty_eh_move_lp): Likewise.
35430 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35433 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35435 * cfgloopmanip.h (loop_version): Update prototype.
35436 * modulo-sched.c (sms_schedule): Update call of loop_version.
35437 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35438 * tree-parloops.c (gen_parallel_loop): Likewise.
35439 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35440 * tree-ssa-loop-split.c (split_loop): Likewise.
35441 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35442 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35444 2017-02-05 Martin Liska <mliska@suse.cz>
35447 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35449 (print_operand_address): Initialize a struct to zero.
35451 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35453 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35454 garbage collector only in textual form.
35456 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35458 * doc/extend.texi (x86 specific memory model extensions for
35459 transactional memory): Simplify a phrase.
35461 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35464 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35465 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35466 (atomic_storedi_1): Likewise.
35468 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35470 PR tree-optimization/79338
35471 * tree-parloops.c (gather_scalar_reductions): Don't call
35472 vect_analyze_loop_form for loop->inner before destroying loop's
35475 2017-02-03 Martin Sebor <msebor@redhat.com>
35477 PR tree-optimization/79327
35478 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35479 when precision has resulted in leading zeros.
35480 (format_integer): Adjust the likely counter to assume an unknown
35481 argument that may be zero is non-zero.
35483 2017-02-03 Jason Merrill <jason@redhat.com>
35486 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35487 avoid copying non-taken branch.
35489 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35491 PR tree-optimization/79340
35492 * tree-vect-loop.c (vectorizable_reduction): Release
35493 vec_defs elements after safe_splicing them into other vectors.
35496 PR tree-optimization/79327
35497 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35498 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35500 (format_integer): Use wide_int_to_tree instead of build_int_cst
35501 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35502 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35503 of shortest and longest sequence.
35505 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35507 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35508 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35510 2017-02-03 Walter Lee <walt@tilera.com>
35513 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35514 after initial stackframe link reg save.
35515 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35517 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35520 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35521 wu for stxssp alternative.
35523 2017-02-03 Martin Sebor <msebor@redhat.com>
35525 PR tree-optimization/79352
35526 * gimple-fold.c (get_range_strlen): Add argument.
35527 (get_range_strlen): Change return type to bool.
35528 (get_maxval_strlen): Pass in a dummy argument.
35529 * gimple-fold.h (get_range_strlen): Change return type to bool.
35530 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35531 * tree.h (array_at_struct_end_p): Add argument.
35532 * tree.c (array_at_struct_end_p): Handle it.
35534 2017-02-03 Martin Liska <mliska@suse.cz>
35537 * multiple_target.c (create_dispatcher_calls): Redirect edge
35538 from a caller of a dispatcher.
35539 (expand_target_clones): Make the clones local.
35540 (ipa_target_clone): Do both target clones and resolvers.
35541 (ipa_dispatcher_calls): Remove the pass.
35542 (pass_dispatcher_calls::gate): Likewise.
35543 (make_pass_dispatcher_calls): Likewise.
35544 * passes.def (pass_target_clone): Put as very first IPA early
35547 2017-02-03 Martin Liska <mliska@suse.cz>
35549 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35550 in case of a function with ifunc attribute.
35552 2017-02-03 Martin Liska <mliska@suse.cz>
35554 * cgraph.c (cgraph_node::dump): Dump function version info.
35555 * symtab.c (symtab_node::dump_base): Add missing new line.
35557 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35559 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35560 (ifcombine_ifandif): Use it.
35562 2017-02-03 Martin Liska <mliska@suse.cz>
35564 * doc/invoke.texi: Document default value for
35565 use-after-scope-direct-emission-threshold.
35567 2017-02-03 Martin Liska <mliska@suse.cz>
35569 PR tree-optimization/79339
35570 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35571 (format_floating): Likewise.
35573 2017-02-03 Martin Liska <mliska@suse.cz>
35576 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35578 (ipa_node_params_t::remove): Likewise.
35579 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35580 initialization from removed ipa_node_params_t::insert.
35581 (ipa_node_params::~ipa_node_params): Move from removed
35582 ipa_node_params_t::release.
35583 * symbol-summary.h (symbol_summary::m_released): New member.
35584 Do not release a summary twice. Do not allow to call finalizer
35585 for types of a summary that live in GGC memory.
35587 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35589 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35592 2017-02-02 Martin Sebor <msebor@redhat.com>
35594 PR middle-end/79275
35595 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35596 (format_string): Tighten up the range of output for non-constant
35597 strings and correct the expected range for wide non-constant strings.
35599 2017-02-02 Martin Sebor <msebor@redhat.com>
35601 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35603 PR middle-end/32003
35604 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35606 (-fdump-tree-@var): Add to index and document how to come up
35607 with pass-specific option and dump file names.
35608 (-fdump-passes): Clarify where to look for output.
35610 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35612 PR middle-end/77445
35613 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35614 statistics of the analyzed path; allow threading for speed when
35615 any of BBs along the path are optimized for speed.
35617 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35619 PR middle-end/78468
35620 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35621 settings of the virtual registers.
35624 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35626 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35627 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35630 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35632 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35633 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35635 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35637 * config/s390/s390.md: Add missing comments with the expanded
35639 * config/s390/vector.md: Likewise.
35640 * config/s390/vx-builtins.md: Likewise.
35642 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35645 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35646 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35647 conditions on a single line.
35649 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35651 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35652 __S390_VX__ to __VX__.
35654 2017-02-01 Andrew Pinski <apinski@cavium.com>
35656 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35657 stmt_info to record_stmt_cost.
35658 (vect_get_known_peeling_cost): Pass stmt_info if known to
35660 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35661 cpu_vector_cost field into
35662 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35663 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35664 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35665 splitting of scalar_stmt_cost and vec_stmt_cost.
35666 (thunderx_vector_cost): Likewise.
35667 (cortexa57_vector_cost): LIkewise.
35668 (exynosm1_vector_cost): Likewise.
35669 (xgene1_vector_cost): Likewise.
35670 (thunderx2t99_vector_cost): Improve after the splitting of the two
35672 (aarch64_builtin_vectorization_cost): Update for the splitting of
35673 scalar_stmt_cost and vec_stmt_cost.
35675 2017-02-01 Torvald Riegel <triegel@redhat.com>
35676 Richard Henderson <rth@redhat.com>
35678 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35679 conditional on existance of a fast atomic load.
35680 * optabs-query.c (can_atomic_load_p): New function.
35681 * optabs-query.h (can_atomic_load_p): Declare it.
35682 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35683 no fast atomic load is available for the particular size of access.
35684 (expand_atomic_compare_and_swap): Likewise.
35685 (expand_atomic_load): Likewise.
35686 (expand_atomic_store): Likewise.
35687 (expand_atomic_fetch_op): Likewise.
35688 * testsuite/lib/target-supports.exp
35689 (check_effective_target_sync_int_128): Remove x86 because it provides
35690 no fast atomic load.
35691 (check_effective_target_sync_int_128_runtime): Likewise.
35693 2017-02-01 Richard Biener <rguenther@suse.de>
35695 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35696 (graphite_transform_loops): Provide opt-info for optimized nests.
35697 * tree-parloop.c (parallelize_loops): Provide opt-info for
35698 parallelized loops.
35700 2017-02-01 Richard Biener <rguenther@suse.de>
35702 PR middle-end/79315
35703 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35704 was not set before.
35706 2017-02-01 Richard Biener <rguenther@suse.de>
35708 PR tree-optimization/71824
35709 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35710 Verify the loops are valid in the merged SESE region.
35711 (scop_detection::can_represent_loop_1): Check analyzing the
35712 evolution of the number of iterations in the region succeeds.
35714 2017-01-31 Ian Lance Taylor <iant@golang.org>
35716 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35717 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35719 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35721 PR preprocessor/79210
35722 * input.c (get_substring_ranges_for_loc): Replace line_width
35723 assertion with error-handling.
35725 2017-01-31 Richard Biener <rguenther@suse.de>
35727 PR tree-optimization/77318
35728 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35729 (create_pw_aff_from_tree): Take loop parameter.
35730 (add_condition_to_pbb): Pass loop of the condition to
35731 create_pw_aff_from_tree.
35733 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35735 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35736 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35738 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35742 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35744 (convert_int_to_float128): Likewise.
35745 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35746 (convert_int_to_float128): Likewise.
35747 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35748 (UNSPEC_IEEE128_CONVERT): Likewise.
35749 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35750 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35751 Use local variables for IBM extended format.
35752 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35753 (fix_trunc<mode>si2_fprs): Likewise.
35754 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35755 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35756 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35757 to know that we can now have integers of all sizes in vector
35759 (fix<uns>_<mode>di2_hw): Likewise.
35760 (float<uns>_<mode>si2_hw): Likewise.
35761 (fix_<mode>si2_hw): Likewise.
35762 (fixuns_<mode>si2_hw): Likewise.
35763 (float<uns>_<mode>di2_hw): Likewise.
35764 (float_<mode>di2_hw): Likewise.
35765 (float_<mode>si2_hw): Likewise.
35766 (floatuns_<mode>di2_hw): Likewise.
35767 (floatuns_<mode>si2_hw): Likewise.
35768 (xscvqp<su>wz_<mode>): Delete, no longer used.
35769 (xscvqp<su>dz_<mode>): Likewise.
35770 (xscv<su>dqp_<mode>): Likewise.
35771 (ieee128_mfvsrd_64bit): Likewise.
35772 (ieee128_mfvsrd_32bit): Likewise.
35773 (ieee128_mfvsrwz): Likewise.
35774 (ieee128_mtvsrw): Likewise.
35775 (ieee128_mtvsrd_64bit): Likewise.
35776 (ieee128_mtvsrd_32bit): Likewise.
35778 2017-01-31 Martin Liska <mliska@suse.cz>
35781 * ipa-prop.c (ipa_free_all_node_params): Call release method
35782 instead of ~sumbol_summary to not to trigger double times
35785 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35787 PR tree-optimization/71691
35788 * bitmap.h (class auto_bitmap): New.
35789 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35790 is_maybe_undefined instead of ssa_undefined_value_p.
35792 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35794 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35795 __S390_ARCH_LEVEL__ to __ARCH__.
35797 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35799 PR tree-optimization/79267
35800 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35801 if should_remove_lhs_p is true.
35803 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35806 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35807 (add_alignment_attribute): New.
35808 (base_type_die): Add alignment attribute.
35809 (subrange_type_die): Likewise.
35810 (modified_type_die): Likewise.
35811 (gen_array_type_die): Likewise.
35812 (gen_descr_array_type_die: Likewise.
35813 (gen_enumeration_type_die): Likewise.
35814 (gen_subprogram_die): Likewise.
35815 (gen_variable_die): Likewise.
35816 (gen_field_die): Likewise.
35817 (gen_ptr_to_mbr_type_die): Likewise.
35818 (gen_struct_or_union_type_die): Likewise.
35819 (gen_subroutine_type_die): Likewise.
35820 (gen_typedef_die): Likewise.
35821 (base_type_cmp): Compare alignment attribute.
35823 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35826 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35827 (setb_unsigned) New pattern for setb with CCUNS.
35828 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35829 subfc./subfe sequence to avoid overflow problems. Generate a
35830 shorter sequence with cmpld/setb for power9.
35831 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35832 for generating subfc. instruction.
35833 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35834 now uses this instruction.
35836 2017-01-30 Ian Lance Taylor <iant@google.com>
35839 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35840 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35842 2017-01-30 Martin Sebor <msebor@redhat.com>
35844 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35845 Move constant to the right of a relational operator.
35846 (get_mpfr_format_length, format_character, format_string): Ditto.
35847 (should_warn_p, maybe_warn): Same.
35849 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35851 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35854 * asan.c (get_translation_unit_decl): Remove function.
35855 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35857 2017-01-30 Martin Liska <mliska@suse.cz>
35859 PR gcov-profile/79259
35860 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35861 -fprofile-generate.
35863 2017-01-30 Martin Liska <mliska@suse.cz>
35866 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35867 Initialize variables with NULL value.
35869 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35872 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35874 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35876 2017-01-30 Richard Biener <rguenther@suse.de>
35878 PR tree-optimization/79276
35879 * tree-vrp.c (process_assert_insertions): Properly adjust common
35880 when removing a duplicate.
35882 2017-01-30 Richard Biener <rguenther@suse.de>
35884 PR tree-optimization/79256
35885 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35886 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35888 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35890 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35893 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35894 ("*r<noxa>sbg_<mode>_sll_bitmask")
35895 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35896 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35897 Use contiguous_bitmask_nowrap_operand.
35899 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35902 * config/rs6000/altivec.h (vec_xl): Revise #define.
35903 (vec_xst): Likewise.
35905 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35907 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35909 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35911 PR rtl-optimization/79194
35912 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35913 traps before call to bypass_conditional_jumps.
35915 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35917 PR tree-optimization/71374
35918 * lra-constraints.c (check_conflict_input_operands): New.
35919 (match_reload): Use it.
35921 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35924 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35925 account to calculate conflict_set.
35927 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35929 PR rtl-optimization/78559
35930 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35931 other_insn in combine.
35933 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35935 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35936 uint16_type_node for BT_UINT16.
35938 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35940 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35941 "RTL Tests" to menu.
35942 (GIMPLE Tests): New node.
35943 (RTL Tests): New node.
35945 2017-01-27 Richard Biener <rguenther@suse.de>
35947 PR tree-optimization/79245
35948 * tree-loop-distribution.c (distribute_loop): Apply cost
35949 modeling also to detected patterns.
35951 2017-01-27 Richard Biener <rguenther@suse.de>
35953 PR tree-optimization/71433
35954 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35955 (compare_assert_loc): New function.
35956 (process_assert_insertions): Sort and optimize assert locations
35957 to remove duplicates and push down identical assertions on
35958 edges to their destination block.
35960 2017-01-27 Richard Biener <rguenther@suse.de>
35962 PR tree-optimization/79244
35963 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35964 out SSA names even if abnormal.
35966 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35968 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35969 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35970 instead of MPFR_RNDN.
35972 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35975 * arm.c (arm_option_override): Don't call build_target_option_node
35976 until after doing all option overrides.
35977 (arm_valid_target_attribute_tree): Likewise.
35979 2017-01-27 Martin Liska <mliska@suse.cz>
35981 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35982 for {cd}tors and C++ {cd}tors.
35984 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35986 * config/s390/s390.md ("*setmem_long_and")
35987 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35989 2017-01-26 Martin Sebor <msebor@redhat.com>
35991 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35994 2017-01-26 Martin Sebor <msebor@redhat.com>
35996 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35997 HAVE_DFmode before using XFmode or DFmode.
35998 (parse_directive): Avoid using the z length modifier to avoid
35999 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36001 PR middle-end/78703
36002 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36003 to accept adjustment as an array.
36004 (get_int_range): New function.
36005 (struct directive): Make width and prec arrays.
36006 (directive::set_width, directive::set_precision): Call get_int_range.
36007 (format_integer, format_floating): Handle width and precision ranges.
36008 (format_string, parse_directive): Same.
36010 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36013 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36014 comdat_type_p set, just clone them, but keep the children in the
36018 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36019 which have direct callers with -fvar-tracking-assignments enabled
36021 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36022 inside of type units.
36024 2017-01-26 Martin Sebor <msebor@redhat.com>
36026 PR middle-end/78703
36027 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36029 (struct format_result): Replace number_chars, number_chars_min,
36030 and number_chars_max with a single member of struct result_range.
36032 (format_result::operator+=): Adjust.
36033 (struct fmtresult): Remove bounded. Handle likely and unlikely
36035 (fmtresult::adjust_for_width_or_precision): New function.
36036 (fmtresult:type_max_digits): New function.
36037 (bytes_remaining): Handle likely and unlikely counters.
36038 (min_bytes_remaining): Remove.
36039 (format_percent): Simplify.
36040 (format_integer, format_floating): Set likely and unlikely counters.
36041 (get_string_length, format_character, format_string): Same.
36042 (format_plain, should_warn_p): New function.
36043 (maybe_warn): Call should_warn_p. Update diagnostic messages
36044 and handle those for all directives, including plain strings.
36045 (format_directive): Handle likely and unlikely counters.
36046 Remove unnecessary quoting from diagnostics. Add an informational
36048 (add_bytes): Remove.
36049 (pass_sprintf_length::compute_format_length): Simplify.
36050 (try_substitute_return_value): Handle likely and unlikely counters.
36052 2017-01-26 Carl Love <cel@us.ibm.com>
36054 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36055 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36057 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36060 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36061 endianess for subregs into account.
36062 * lra-constraints.c (lra_constraints): Do risky transformations
36063 always on the first iteration.
36064 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36065 last_call_used_reg_set.
36066 (process_bb_lives): Define and use last_call_used_reg_set.
36067 * lra.c (lra): Always continue after lra_constraints on the first
36070 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36072 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36074 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36076 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36078 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36079 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36080 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36081 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36082 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36083 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36084 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36085 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36086 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36088 2017-01-26 Marek Polacek <polacek@redhat.com>
36091 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36092 for the third operand.
36094 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36096 PR middle-end/79236
36097 * omp-low.c (struct omp_context): Add simt_stmt field.
36098 (scan_omp_for): Return omp_context *.
36099 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36100 context to the _simt_ SIMD stmt.
36101 (lower_omp_for): For combined SIMD with sibling _simt_
36102 SIMD, make sure to use the same decls in _looptemp_
36103 clauses as in the sibling.
36105 2017-01-26 David Sherwood <david.sherwood@arm.com>
36107 PR middle-end/79212
36108 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36111 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36114 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36115 emit fld b; fld a; if possible.
36117 * brig-builtins.def: Update copyright years.
36118 * config/arm/arm_acle_builtins.def: Update copyright years.
36120 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36123 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36124 constraint instead of o for the stxsd instruction.
36126 2017-01-25 Carl Love <cel@us.ibm.com>
36128 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36129 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36131 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36133 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36135 2017-01-25 Richard Biener <rguenther@suse.de>
36137 PR tree-optimization/69264
36138 * target.def (vector_alignment_reachable): Improve documentation.
36139 * doc/tm.texi: Regenerate.
36140 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36142 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36143 earlier changes with respect to TYPE_USER_ALIGN.
36144 (vector_alignment_reachable_p): Likewise. Improve dumping.
36146 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36149 * config/arm/arm.md (xordi3): Force constant operand into a register
36152 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36154 * doc/invoke.texi (-fstore-merging): Correct default optimization
36155 levels at which it is enabled.
36156 (-O): Move -fstore-merging from list to...
36159 2017-01-25 Richard Biener <rguenther@suse.de>
36162 * omp-expand.c: Include debug.h.
36163 (expand_omp_taskreg): Make sure to generate early debug before
36164 outlining anything from a function.
36165 (expand_omp_target): Likewise.
36166 (grid_expand_target_grid_body): Likewise.
36168 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36171 * asan.c (get_translation_unit_decl): New function.
36172 (asan_add_global): Extract modules file name from globals
36173 TRANSLATION_UNIT_DECL name.
36175 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36178 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36179 for long calls with APCS frame and VFP.
36181 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36183 * cfg.c (original_copy_tables_initialized_p): New function.
36184 * cfg.h (original_copy_tables_initialized_p): New decl.
36185 * cfgrtl.c (relink_block_chain): Guard the call to
36186 free_original_copy_tables with a call to
36187 original_copy_tables_initialized_p.
36188 * cgraph.h (symtab_node::native_rtl_p): New decl.
36189 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36190 (symtab_node::needed_p): Don't assert for early assembly output
36191 for __RTL functions.
36192 (cgraph_node::finalize_function): Set "force_output" for __RTL
36194 (cgraph_node::analyze): Bail out early for __RTL functions.
36195 (analyze_functions): Update assertion to support __RTL functions.
36196 (cgraph_node::expand): Bail out early for __RTL functions.
36197 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36199 * function.h (struct function): Update comment for field
36201 * gimple-expr.c: Include "tree-pass.h".
36202 (gimple_has_body_p): Return false for __RTL functions.
36203 * Makefile.in (OBJS): Add run-rtl-passes.o.
36204 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36206 (gcc::pass_manager::get_clean_slate): New accessor.
36207 * passes.c: Include "insn-addr.h".
36208 (should_skip_pass_p): Add logging. Update logic for running
36209 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36210 property-provider override so it is only done for gimple passes.
36212 (skip_pass): New function.
36213 (execute_one_pass): Call skip_pass when skipping passes.
36214 * read-md.c (md_reader::read_char): Support filtering
36215 the input to a subset of line numbers.
36216 (md_reader::md_reader): Initialize fields
36217 m_first_line and m_last_line.
36218 (md_reader::read_file_fragment): New function.
36219 * read-md.h (md_reader::read_file_fragment): New decl.
36220 (md_reader::m_first_line): New field.
36221 (md_reader::m_last_line): New field.
36222 * read-rtl-function.c (function_reader::create_function): Only
36223 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36224 curr_properties. Set DECL_INITIAL to a dummy block.
36225 (read_rtl_function_body_from_file_range): New function.
36226 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36228 * run-rtl-passes.c: New file.
36229 * run-rtl-passes.h: New file.
36231 2017-01-24 Jeff Law <law@redhat.com>
36233 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36236 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36238 PR tree-optimization/79159
36239 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36240 (record_nonwrapping_iv): Improve boundary using above function if no
36241 value range information.
36243 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36244 Martin Jambor <mjambor@suse.cz>
36246 * brig-builtins.def: New file.
36247 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36248 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36249 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36250 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36251 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36252 * builtin-types.def (BT_INT8): New.
36253 (BT_INT16): Likewise.
36254 (BT_UINT8): Likewise.
36255 (BT_UINT16): Likewise.
36256 (BT_FN_ULONG): Likewise.
36257 (BT_FN_UINT_INT): Likewise.
36258 (BT_FN_UINT_ULONG): Likewise.
36259 (BT_FN_UINT_LONG): Likewise.
36260 (BT_FN_UINT_PTR): Likewise.
36261 (BT_FN_ULONG_PTR): Likewise.
36262 (BT_FN_INT8_FLOAT): Likewise.
36263 (BT_FN_INT16_FLOAT): Likewise.
36264 (BT_FN_UINT32_FLOAT): Likewise.
36265 (BT_FN_UINT16_FLOAT): Likewise.
36266 (BT_FN_UINT8_FLOAT): Likewise.
36267 (BT_FN_UINT64_FLOAT): Likewise.
36268 (BT_FN_UINT16_UINT32): Likewise.
36269 (BT_FN_UINT32_UINT16): Likewise.
36270 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36271 (BT_FN_INT_PTR_INT): Likewise.
36272 (BT_FN_UINT_PTR_UINT): Likewise.
36273 (BT_FN_LONG_PTR_LONG): Likewise.
36274 (BT_FN_ULONG_PTR_ULONG): Likewise.
36275 (BT_FN_VOID_UINT64_UINT64): Likewise.
36276 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36277 (BT_FN_INT8_INT8_INT8): Likewise.
36278 (BT_FN_INT16_INT16_INT16): Likewise.
36279 (BT_FN_INT_INT_INT): Likewise.
36280 (BT_FN_UINT_FLOAT_UINT): Likewise.
36281 (BT_FN_FLOAT_UINT_UINT): Likewise.
36282 (BT_FN_ULONG_UINT_UINT): Likewise.
36283 (BT_FN_ULONG_UINT_PTR): Likewise.
36284 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36285 (BT_FN_UINT_UINT_UINT): Likewise.
36286 (BT_FN_VOID_UINT_PTR): Likewise.
36287 (BT_FN_UINT_UINT_PTR: Likewise.
36288 (BT_FN_UINT32_UINT64_PTR): Likewise.
36289 (BT_FN_INT_INT_UINT_UINT): Likewise.
36290 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36291 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36292 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36293 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36294 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36295 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36296 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36297 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36298 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36299 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36300 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36301 * doc/frontends.texi: List BRIG FE.
36302 * doc/install.texi (Testing): Add BRIG tesring requirements.
36303 * doc/invoke.texi (Overall Options): Mention BRIG.
36304 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36306 2017-01-24 Richard Biener <rguenther@suse.de>
36308 PR translation/79208
36309 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36311 2017-01-24 Martin Jambor <mjambor@suse.cz>
36314 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36315 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36316 and known_contexts.
36318 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36320 PR middle-end/79123
36321 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36322 casts from signed to unsigned really don't have a range.
36324 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36326 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36327 GMP_RNDx for compatiblity.
36329 2017-01-24 Martin Liska <mliska@suse.cz>
36332 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36333 that would prevent us to call alloca with -1 as argument.
36335 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36337 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36338 Avoid -Wformat-security warning.
36340 2017-01-23 Andrew Pinski <apinski@cavium.com>
36342 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36345 2017-01-23 Martin Sebor <msebor@redhat.com>
36347 PR middle-end/78703
36348 * gimple-ssa-sprintf.c (warn_level): New global.
36349 (format_integer): Use it here and throughout the rest of the file.
36350 Use the same switch to compute sign as base.
36351 (maybe_warn): New function.
36352 (format_directive): Factor out warnings into maybe_warn.
36353 Add debugging output. Use warn_level.
36354 (add_bytes): Use warn_level.
36355 (pass_sprintf_length::compute_format_length): Add debugging output.
36356 (try_substitute_return_value): Same.
36357 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36359 PR middle-end/78703
36360 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36361 (struct fmtresult, format_integer, format_floating): Adjust.
36362 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36363 (get_string_length, format_string,format_directive): Same.
36364 (pass_sprintf_length::compute_format_length): Same.
36365 (try_substitute_return_value): Simplify slightly.
36367 PR middle-end/78703
36368 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36369 (fmtresult::operator+=): Outlined.
36370 (struct fmtresult): Add ctors.
36371 (struct conversion_spec): Rename...
36372 (struct directive): ...to this. Add and remove data members.
36373 (directive::set_width, directive::set_precision): New functions.
36374 (format_percent): Use fmtresult ctor.
36375 (get_width_and_precision): Remove.
36376 (format_integer): Make naming changes. Avoid computing width and
36378 (format_floating): Same. Adjust indentation.
36379 (format_character, format_none): New functions.
36380 (format_string): Moved character handling to format_character.
36381 (format_directive): Remove arguments, change return type.
36382 (parse_directive): New function.
36383 (pass_sprintf_length::compute_format_length): Move directive
36384 parsing to parse_directive.
36386 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36388 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36389 (assign_assembler_name_if_needed): ... this.
36390 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36391 (assign_assembler_name_if_needed): ... this.
36392 (free_lang_data_in_cgraph): Adjust callers.
36393 * cgraphunit.c (cgraph_node::analyze): Likewise.
36394 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36397 2017-01-23 Richard Biener <rguenther@suse.de>
36399 PR tree-optimization/79088
36400 PR tree-optimization/79188
36401 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36402 resetting loop bounds after last path deletion. Reset loop
36403 bounds of the target loop, make code match the comments.
36404 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36405 Make sure loops need no fixups.
36407 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36409 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36410 exponent support with double type for first argument.
36411 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36412 type returned by __builtin_vec_extract_sig,
36413 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36414 functions from "vector int" to "vector unsigned int" or from
36415 "vector long long int" to "vector unsigned long long int".
36416 Changed type returned by __builtin_vec_extract_exp,
36417 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36418 functions from "vector int" to "vector unsigned int" or from
36419 "vector long long int" to "vector unsigned long long int".
36420 Changed return type of __builtin_vec_test_data_class,
36421 __builtin_vec_test_data_class_sp, and
36422 __builtin_vec_test_data_class_dp from "vector int" to
36423 "vector bool int" or from "vector long long int" to "vector bool
36424 long long int" and changed second argument type from "unsigned
36425 int" to "int". Added new overloaded function forms "vector float
36426 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36427 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36428 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36429 double, vector unsigned long long int)" and "vector double
36430 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36431 long int)". Changed return type of
36432 __builtin_scalar_test_data_class and
36433 __builtin_scalar_test_data_class_sp and
36434 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36435 int" and changed second argument from "unsigned int" to "int".
36436 Changed type returned by __builtin_scalar_test_neg,
36437 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36438 from "int" to "bool int". Added new overloaded function form
36439 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36440 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36441 exponent double-precision with floating point first argument.
36442 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36443 documentation of scalar_test_data_class, scalar_test_neg,
36444 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36445 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36446 vec_test_data_class built-in functions to reflect refinements in
36447 their type signatures.
36449 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36451 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36453 (aarch64_elf_asm_destructor): Likewise.
36455 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36457 PR rtl-optimization/78634
36458 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36459 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36460 * ifcvt.c (noce_try_cmove): Add missing cost check.
36462 PR rtl-optimization/71724
36463 * combine.c (if_then_else_cond): Look for situations where it is
36464 beneficial to undo the work of one of the recursive calls.
36466 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36468 PR tree-optimization/70754
36469 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36470 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36471 combined stmt before it if not NULL.
36472 (combine_chains): Process refs reversely and compute dominance point
36475 2017-01-23 Martin Liska <mliska@suse.cz>
36477 PR tree-optimization/79196
36478 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36479 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36481 (strlen_optimize_stmt): Call the renamed function.
36483 2017-01-23 Michael Matz <matz@suse.de>
36485 PR tree-optimization/78384
36486 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36488 2017-01-23 Richard Biener <rguenther@suse.de>
36490 PR tree-optimization/79186
36491 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36492 both incoming edges before moving an assert.
36494 2017-01-23 Martin Jambor <mjambor@suse.cz>
36496 * ipa-prop.c (load_from_param_1): Removed.
36497 (load_from_unmodified_param): Bits from load_from_param_1 put back
36499 (load_from_param): Removed.
36500 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36501 with stmt. Reverted back to use of load_from_unmodified_param.
36503 2017-01-23 Martin Jambor <mjambor@suse.cz>
36506 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36507 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36508 field a pointer to garbage collected vector, mark lattices and
36509 ipcp_orig_node with GTY((skip)).
36510 (ipa_get_param_count): Adjust to descriptors being a pointer.
36511 (ipa_get_param): Likewise.
36512 (ipa_get_type): Likewise.
36513 (ipa_get_param_move_cost): Likewise.
36514 (ipa_set_param_used): Likewise.
36515 (ipa_get_controlled_uses): Likewise.
36516 (ipa_set_controlled_uses): Likewise.
36517 (ipa_is_param_used): Likewise.
36518 (ipa_node_params_t): Move into garbage collector. New methods insert
36520 (ipa_node_params_sum): Annotate wth GTY(()).
36521 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36523 (ipa_load_from_parm_agg): Adjust declaration.
36524 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36525 * ipa-profile.c (ipa_profile): Likewise.
36526 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36527 (ipa_populate_param_decls): Make descriptors parameter garbage
36529 (ipa_dump_param): Adjust to descriptors being a pointer.
36530 (ipa_alloc_node_params): Likewise.
36531 (ipa_initialize_node_params): Likewise.
36532 (load_from_param_1): Make descriptors parameter garbage collected.
36533 (load_from_unmodified_param): Likewise.
36534 (load_from_param): Likewise.
36535 (ipa_load_from_parm_agg): Likewise.
36536 (ipa_node_params::~ipa_node_params): Removed.
36537 (ipa_free_all_node_params): Remove call to delete operator.
36538 (ipa_node_params_t::insert): New.
36539 (ipa_node_params_t::remove): Likewise.
36540 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36541 copy known_csts and known_contexts vectors.
36542 (ipa_read_node_info): Adjust to descriptors being a pointer.
36543 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36545 (ipcp_transform_function): Make descriptors variable garbage
36548 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36550 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36551 * config/i386/avx512dqintrin.h: Ditto.
36552 * config/i386/avx512fintrin.h: Ditto.
36553 * config/i386/i386.c: Handle new builtins.
36554 * config/i386/i386-builtin.def: Add new builtins.
36555 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36556 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36558 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36559 Martin Liska <mliska@suse.cz>
36561 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36562 * asan.c (asan_expand_poison_ifn): Support stores and use
36563 appropriate ASAN report function.
36564 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36565 * internal-fn.def (ASAN_POISON_USE): Declare.
36566 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36567 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36568 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36569 ASAN_POISON calls w/o LHS.
36570 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36571 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36572 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36573 * gimplify.c (asan_poison_variables): Add attribute
36574 use_after_scope_memory to variables that really needs to live
36576 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36577 having the attribute.
36579 2017-01-23 Martin Liska <mliska@suse.cz>
36581 * asan.c (create_asan_shadow_var): New function.
36582 (asan_expand_poison_ifn): Likewise.
36583 * asan.h (asan_expand_poison_ifn): New declaration.
36584 * internal-fn.c (expand_ASAN_POISON): Likewise.
36585 * internal-fn.def (ASAN_POISON): New builtin.
36586 * sanopt.c (pass_sanopt::execute): Expand
36587 asan_expand_poison_ifn.
36588 * tree-inline.c (copy_decl_for_dup_finish): Make function
36590 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36591 * tree-ssa.c (is_asan_mark_p): New function.
36592 (execute_update_addresses_taken): Rewrite local variables
36593 (identified just by use-after-scope as addressable) into SSA.
36595 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36597 * doc/install.texi (Specific): opensource.apple.com uses https
36598 now. Remove trailing slash.
36600 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36602 * README.Portability: Remove note on an Irix compatibility issue.
36604 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36606 * gcov.c (INCLUDE_ALGORITHM): Define.
36607 (INCLUDE_VECTOR): Define.
36608 No longer include <vector> and <algorithm> directly.
36610 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36612 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36614 * doc/invoke.texi (Code Gen Options): Ditto.
36616 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36619 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36621 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36623 rtl-optimization/79125
36624 * cprop.c (local_cprop_pass): Handle cases where we make an
36625 unconditional trap.
36627 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36631 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36632 read from, for big endian.
36634 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36636 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36637 register pauth builtins for LP64 only.
36639 2017-01-20 Marek Polacek <polacek@redhat.com>
36642 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36645 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36647 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36649 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36650 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36651 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36653 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36656 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36657 in big-endian mode when they are not a single duplicated value.
36659 2017-01-20 Richard Biener <rguenther@suse.de>
36661 * BASE-VER: Bump to 7.0.1.
36663 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36665 * omp-low.c (omplow_simd_context): New struct. Use it...
36666 (lower_rec_simd_input_clauses): ...here and...
36667 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36668 references to idx, lane, max_vf, is_simt.
36670 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36672 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36675 2017-01-20 Martin Jambor <mjambor@suse.cz>
36677 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36678 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36680 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36681 (GTFILES): Rename hsa.c to hsa-common.c.
36682 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36683 * hsa-dump.c: Likewise.
36684 * hsa-gen.c: Likewise.
36685 * hsa-regalloc.c: Likewise.
36686 * ipa-hsa.c: Likewise.
36687 * omp-expand.c: Likewise.
36688 * omp-low.c: Likewise.
36689 * toplev.c: Likewise.
36691 2017-01-20 Marek Polacek <polacek@redhat.com>
36694 * doc/invoke.texi: Document -Wduplicated-branches.
36695 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36696 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36697 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36698 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36699 return 0 only when not OEP_LEXICOGRAPHIC.
36700 (fold_build_cleanup_point_expr): Use the expression
36701 location when building CLEANUP_POINT_EXPR.
36702 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36703 * tree.c (add_expr): Handle error_mark_node.
36705 2017-01-20 Martin Liska <mliska@suse.cz>
36708 * tree-profile.c (init_ic_make_global_vars): Do not call
36710 (gimple_init_gcov_profiler): Likewise.
36712 2017-01-20 Martin Liska <mliska@suse.cz>
36715 * cgraph.h (maybe_create_reference): Remove argument and
36717 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36719 * ipa-cp.c (create_specialized_node): Likewise.
36720 * symtab.c (symtab_node::maybe_create_reference): Handle
36721 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36723 2017-01-20 Martin Liska <mliska@suse.cz>
36725 * read-rtl-function.c (function_reader::create_function): Use
36726 build_decl instread of build_decl_stat.
36728 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36730 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36731 * config/i386/avx512dqintrin.h: Ditto.
36732 * config/i386/avx512fintrin.h: Ditto.
36733 * config/i386/i386-builtin-types.def: Add new types.
36734 * config/i386/i386.c: Handle new types.
36735 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36736 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36737 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36738 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36739 (__builtin_ia32_kshiftridi): New.
36740 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36742 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36746 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36747 define to rs6000_init_stack_protect_guard.
36748 (rs6000_init_stack_protect_guard): New function.
36750 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36751 Yunqiang Su <yunqiang.su@imgtec.com>
36753 * config.gcc (supported_defaults): Add madd4.
36754 (with_madd4): Add validation.
36755 (all_defaults): Add madd4.
36756 * config/mips/mips.opt (mmadd4): New option.
36757 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36759 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36761 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36762 (ISA_HAS_FUSED_MADD4): Likewise.
36763 * doc/invoke.texi (-mmadd4): Document the new option.
36764 * doc/install.texi (--with-madd4): Document the new option.
36766 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36768 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36769 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36770 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36771 (aarch64_init_pauth_hint_builtins): New.
36772 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36773 (aarch64_expand_builtin): Expand new builtins.
36775 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36777 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36778 * combine-stack-adj.c (no_unhandled_cfa): Handle
36779 REG_CFA_TOGGLE_RA_MANGLE.
36780 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36781 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36782 info for return address signing.
36783 (aarch64_expand_epilogue): Likewise.
36785 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36787 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36788 * config/aarch64/aarch64-protos.h
36789 (aarch64_return_address_signing_enabled): New declaration.
36790 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36792 (aarch64_expand_prologue): Sign return address before it's pushed onto
36794 (aarch64_expand_epilogue): Authenticate return address fetched from
36796 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36797 (aarch64_attributes): New function attributes for "sign-return-address".
36798 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36799 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36800 ("*do_return"): Generate combined instructions according to key index.
36801 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36802 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36804 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36805 * config/aarch64/aarch64.opt (msign-return-address=): New.
36806 * doc/extend.texi (AArch64 Function Attributes): Documents
36807 "sign-return-address=".
36808 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36810 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36812 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36813 overall option summary.
36815 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36817 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36818 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36819 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36820 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36822 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36824 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36825 -mpower9-minmax by default for -mcpu=power9.
36826 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36827 128-bit floating point.
36829 2017-01-20 Alan Modra <amodra@gmail.com>
36831 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36832 optimizing for size.
36834 2017-01-20 Alan Modra <amodra@gmail.com>
36837 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36838 for strcmp and strncmp from corresponding builtin decl.
36840 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36842 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36843 instead of i386/rtems-64.h.
36844 * config/i386/rtems-64.h: Remove.
36846 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36850 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36852 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36854 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36856 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36857 Change int to HOST_WIDE_INT.
36858 * config/aarch64/aarch64-protos.h
36859 (aarch64_simd_gen_const_vector_dup): Likewise.
36860 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36862 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36864 * langhooks-def.h (lhd_type_for_size): New decl.
36865 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36866 * langhooks.c (lhd_type_for_size): New function, taken from
36869 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36871 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36872 define_bypass for CR latency.
36873 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36874 (power9-alu2): Add define_bypass for CR latency.
36876 (power9-mul): Update insn latency.
36877 (power9-mul-compare): Update insn latency, bypass latency and remove
36880 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36882 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36884 * config/aarch64/aarch64.md
36885 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36886 aarch64_nopcrelative_literal_loads.
36887 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36889 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36891 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36892 TARGET_LOONGSON_3A.
36893 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36895 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36898 * config.gcc (supported_defaults): Add lxc1-sxc1.
36899 (with_lxc1_sxc1): Add validation.
36900 (all_defaults): Add lxc1-sxc1.
36901 * config/mips/mips.opt (mlxc1-sxc1): New option.
36902 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36904 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36905 __mips_no_lxc1_sxc1.
36906 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36907 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36908 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36910 2017-01-19 Richard Biener <rguenther@suse.de>
36912 PR tree-optimization/72488
36913 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36914 sure to restore SSA info.
36915 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36917 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36919 PR rtl-optimization/79121
36920 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36921 of the inner type when shifting an extended value.
36923 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36926 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36927 interposable aliases.
36929 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36932 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36933 Use the evmergelohi instruction.
36934 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36935 (mov_sitf_e500_subreg8_2_be): Likewise.
36936 (mov_sitf_e500_subreg12_2_le): Likewise.
36937 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36938 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36939 (mov_sitf_e500_subreg8_2_le): Likewise.
36940 (mov_sitf_e500_subreg12_2_be): Likewise.
36942 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36944 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36945 attribute from vecsimple to vecperm.
36946 (altivec_vbpermq2): Likewise.
36948 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36951 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36953 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36954 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36955 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36956 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36957 case where N arg is SIZE_MAX.
36958 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36959 (cmpstrsi): Add pattern.
36961 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36963 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36964 __builtin_vec_revb builtins.
36965 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36966 built-in functions to support generation of the ISA 3.0 XXBR<x>
36967 vector byte reverse instructions.
36968 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36969 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36970 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36971 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36972 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36973 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36974 (P9V_BUILTIN_VEC_REVB): Likewise.
36975 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36976 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36977 (p9_xxbrq_v16qi): Likewise.
36978 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36979 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36980 (p9_xxbrh_v8hi): Likewise.
36981 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36982 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36983 vec_revb built-in functions.
36985 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36987 PR rtl-optimization/78952
36988 * config/i386/i386.md (any_extract): New code iterator.
36989 (*insvqi_2): Use any_extract for source operand.
36990 (*insvqi_3): Use any_shiftrt for source operand.
36992 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36994 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36996 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36998 2017-01-18 Matthias Klose <doko@ubuntu.com>
37000 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37002 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37004 * config/rs6000/altivec.h (vec_bperm): Change #define.
37005 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37006 (altivec_vbpermq2): New define_insn.
37007 (altivec_vbpermd): Likewise.
37008 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37009 function interface.
37010 (VBPERMD): Likewise.
37011 (VBPERM): New polymorphic function interface.
37012 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37013 Add entries for P9V_BUILTIN_VEC_VBPERM.
37014 * doc/extend.texi: Add interfaces for vec_bperm.
37016 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37018 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37019 first letter of error messages.
37020 (s390_resolve_overloaded_builtin): Likewise.
37021 * config/s390/s390.c (s390_expand_builtin): Likewise.
37022 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37023 (s390_valid_target_attribute_inner_p): Likewise.
37024 * config/s390/s390.md ("tabort"): Likewise.
37026 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37028 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37029 (ISA_AVOID_DIV_HILO): New macro.
37030 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37031 (ISA_HAS_DDIV): Likewise.
37033 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37035 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37037 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37039 * doc/invoke.texi (fabi-version): Spelling fix.
37041 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37044 * doc/invoke.texi (fabi-version): Mention mangling fix for
37047 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37050 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37052 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37055 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37058 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37060 (stack_protect_test): Ditto.
37061 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37062 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37064 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37065 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37066 -mstack-protector-guard-offset=.
37067 (RS/6000 and PowerPC Options): Ditto.
37069 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37071 * config/i386/i386.h (MASK_CLASS_P): New define.
37072 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37073 there are no registers from different register sets also when
37074 mask registers are used. Update function comment.
37075 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37076 to (*k/*r) and (*k/*km) alternatives.
37078 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37080 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37081 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37082 (EH_RETURN_HANDLER_RTX): New define.
37083 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37084 Force frame pointer in EH return functions.
37085 (aarch64_expand_epilogue): Add barrier for eh_return.
37086 (aarch64_final_eh_return_addr): Remove.
37087 (aarch64_eh_return_handler_rtx): New function.
37088 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37090 (aarch64_eh_return_handler_rtx): New prototype.
37092 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37094 * config/rs6000/altivec.h (vec_rlmi): New #define.
37095 (vec_vrlnm): Likewise.
37096 (vec_rlnm): Likewise.
37097 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37098 (UNSPEC_VRLNM): Likewise.
37099 (VIlong): New mode iterator.
37100 (altivec_vrl<VI_char>mi): New define_insn.
37101 (altivec_vrl<VI_char>nm): Likewise.
37102 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37104 (VRLDNM): Likewise.
37105 (RLNM): New polymorphic function entry.
37106 (VRLWMI): New monomorphic function entry.
37107 (VRLDMI): Likewise.
37108 (RLMI): New polymorphic function entry.
37109 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37110 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37111 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37114 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37117 * dwarf2out.c (field_byte_offset): Restore the
37118 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37119 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37120 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37123 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37126 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37128 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37131 * dwarf2out.c (add_data_member_location_attribute): For constant
37132 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37133 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37134 DW_AT_byte_size attributes.
37136 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37138 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37139 after forcing to constant memory when the code model is medium.
37141 2017-01-17 Julia Koval <julia.koval@intel.com>
37144 * config/i386/avx512fintrin.h
37145 (_mm512_i32gather_ps): Change __addr type to void const*.
37146 (_mm512_mask_i32gather_ps): Ditto.
37147 (_mm512_i32gather_pd): Ditto.
37148 (_mm512_mask_i32gather_pd): Ditto.
37149 (_mm512_i64gather_ps): Ditto.
37150 (_mm512_mask_i64gather_ps): Ditto.
37151 (_mm512_i64gather_pd): Ditto.
37152 (_mm512_mask_i64gather_pd): Ditto.
37153 (_mm512_i32gather_epi32): Ditto.
37154 (_mm512_mask_i32gather_epi32): Ditto.
37155 (_mm512_i32gather_epi64): Ditto.
37156 (_mm512_mask_i32gather_epi64): Ditto.
37157 (_mm512_i64gather_epi32): Ditto.
37158 (_mm512_mask_i64gather_epi32): Ditto.
37159 (_mm512_i64gather_epi64): Ditto.
37160 (_mm512_mask_i64gather_epi64): Ditto.
37161 (_mm512_i32scatter_ps): Change __addr type to void*.
37162 (_mm512_mask_i32scatter_ps): Ditto.
37163 (_mm512_i32scatter_pd): Ditto.
37164 (_mm512_mask_i32scatter_pd): Ditto.
37165 (_mm512_i64scatter_ps): Ditto.
37166 (_mm512_mask_i64scatter_ps): Ditto.
37167 (_mm512_i64scatter_pd): Ditto.
37168 (_mm512_mask_i64scatter_pd): Ditto.
37169 (_mm512_i32scatter_epi32): Ditto.
37170 (_mm512_mask_i32scatter_epi32): Ditto.
37171 (_mm512_i32scatter_epi64): Ditto.
37172 (_mm512_mask_i32scatter_epi64): Ditto.
37173 (_mm512_i64scatter_epi32): Ditto.
37174 (_mm512_mask_i64scatter_epi32): Ditto.
37175 (_mm512_i64scatter_epi64): Ditto.
37176 (_mm512_mask_i64scatter_epi64): Ditto.
37177 * config/i386/avx512pfintrin.h
37178 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37179 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37180 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37181 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37182 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37183 (_mm512_prefetch_i32scatter_ps): Ditto.
37184 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37185 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37186 (_mm512_prefetch_i64scatter_pd): Ditto.
37187 (_mm512_prefetch_i64scatter_ps): Ditto.
37188 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37189 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37190 * config/i386/avx512vlintrin.h
37191 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37192 (_mm_mmask_i32gather_ps): Ditto.
37193 (_mm256_mmask_i32gather_pd): Ditto.
37194 (_mm_mmask_i32gather_pd): Ditto.
37195 (_mm256_mmask_i64gather_ps): Ditto.
37196 (_mm_mmask_i64gather_ps): Ditto.
37197 (_mm256_mmask_i64gather_pd): Ditto.
37198 (_mm_mmask_i64gather_pd): Ditto.
37199 (_mm256_mmask_i32gather_epi32): Ditto.
37200 (_mm_mmask_i32gather_epi32): Ditto.
37201 (_mm256_mmask_i32gather_epi64): Ditto.
37202 (_mm_mmask_i32gather_epi64): Ditto.
37203 (_mm256_mmask_i64gather_epi32): Ditto.
37204 (_mm_mmask_i64gather_epi32): Ditto.
37205 (_mm256_mmask_i64gather_epi64): Ditto.
37206 (_mm_mmask_i64gather_epi64): Ditto.
37207 (_mm256_i32scatter_ps): Change __addr type to void*.
37208 (_mm256_mask_i32scatter_ps): Ditto.
37209 (_mm_i32scatter_ps): Ditto.
37210 (_mm_mask_i32scatter_ps): Ditto.
37211 (_mm256_i32scatter_pd): Ditto.
37212 (_mm256_mask_i32scatter_pd): Ditto.
37213 (_mm_i32scatter_pd): Ditto.
37214 (_mm_mask_i32scatter_pd): Ditto.
37215 (_mm256_i64scatter_ps): Ditto.
37216 (_mm256_mask_i64scatter_ps): Ditto.
37217 (_mm_i64scatter_ps): Ditto.
37218 (_mm_mask_i64scatter_ps): Ditto.
37219 (_mm256_i64scatter_pd): Ditto.
37220 (_mm256_mask_i64scatter_pd): Ditto.
37221 (_mm_i64scatter_pd): Ditto.
37222 (_mm_mask_i64scatter_pd): Ditto.
37223 (_mm256_i32scatter_epi32): Ditto.
37224 (_mm256_mask_i32scatter_epi32): Ditto.
37225 (_mm_i32scatter_epi32): Ditto.
37226 (_mm_mask_i32scatter_epi32): Ditto.
37227 (_mm256_i32scatter_epi64): Ditto.
37228 (_mm256_mask_i32scatter_epi64): Ditto.
37229 (_mm_i32scatter_epi64): Ditto.
37230 (_mm_mask_i32scatter_epi64): Ditto.
37231 (_mm256_i64scatter_epi32): Ditto.
37232 (_mm256_mask_i64scatter_epi32): Ditto.
37233 (_mm_i64scatter_epi32): Ditto.
37234 (_mm_mask_i64scatter_epi32): Ditto.
37235 (_mm256_i64scatter_epi64): Ditto.
37236 (_mm256_mask_i64scatter_epi64): Ditto.
37237 (_mm_i64scatter_epi64): Ditto.
37238 (_mm_mask_i64scatter_epi64): Ditto.
37239 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37240 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37241 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37242 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37243 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37244 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37245 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37246 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37247 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37248 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37249 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37250 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37251 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37252 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37253 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37254 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37255 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37256 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37257 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37258 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37259 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37260 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37261 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37262 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37263 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37264 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37265 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37266 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37267 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37268 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37269 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37270 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37271 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37272 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37273 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37274 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37275 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37276 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37277 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37278 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37279 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37280 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37281 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37282 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37283 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37284 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37285 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37286 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37287 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37288 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37289 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37290 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37291 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37292 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37293 definitions accordingly.
37295 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37296 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37299 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37302 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37305 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37306 hard regs for inner regno.
37308 2017-01-17 Martin Liska <mliska@suse.cz>
37311 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37312 assumption and add comment.
37314 2017-01-17 Nathan Sidwell <nathan@acm.org>
37316 * ipa-visibility.c (localize_node): New function, broken out of ...
37317 (function_and_variable_visibility): ... here. Call it.
37319 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37321 PR middle-end/77445
37322 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37323 correctly set frequency of oudgoing edge.
37324 (duplicate_thread_path): Fix profile updating.
37326 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37329 * configure.ac: Add GCC_BASE_VER.
37330 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37331 version from BASE-VER file.
37332 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37333 (gcc.o): Depend on $(BASEVER).
37334 * common.opt (dumpfullversion): New option.
37335 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37336 * doc/invoke.texi: Document -dumpfullversion.
37337 * doc/install.texi: Document --with-gcc-major-version-only.
37338 * configure: Regenerated.
37340 2017-01-17 Richard Biener <rguenther@suse.de>
37342 PR tree-optimization/71433
37343 * tree-vrp.c (register_new_assert_for): Merge same asserts
37344 on all incoming edges.
37345 (process_assert_insertions_for): Handle insertions at the
37348 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37350 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37351 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37353 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37356 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37359 2017-01-17 Alan Modra <amodra@gmail.com>
37362 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37363 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37364 symbolic stack limit when pic.
37366 2017-01-16 Martin Sebor <msebor@redhat.com>
37368 PR tree-optimization/78608
37369 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37371 2017-01-16 Jeff Law <law@redhat.com>
37374 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37375 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37376 for several include directories that may be relative to sysroot.
37377 * config/i386/x-mingw32 (gplus_includedir): Define.
37378 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37379 (native_system_includedir): Likewise.
37380 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37381 override if TARGET_SYSTEM_ROOT is defined.
37382 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37384 PR tree-optimization/79090
37385 PR tree-optimization/33562
37386 PR tree-optimization/61912
37387 PR tree-optimization/77485
37388 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37389 and computed trims into the dump file.
37391 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37393 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37395 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37398 * gimplify.c (gimplify_init_constructor): If want_value and
37399 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37403 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37404 sequence. Formatting fixes.
37405 (doloop_optimize): Formatting fixes.
37408 * gcc.c (debug_level_greater_than_spec_func): New function.
37409 (static_spec_functions): Add debug-level-gt spec function.
37410 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37412 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37413 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37414 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37415 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37416 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37417 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37419 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37421 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37422 QImode fixups to general and mask registers only.
37424 2017-01-16 Carl Love <cel@us.ibm.com>
37426 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37427 for built-in functions
37428 vector signed char vec_nabs (vector signed char)
37429 vector signed short vec_nabs (vector signed short)
37430 vector signed int vec_nabs (vector signed int)
37431 vector signed long long vec_nabs (vector signed long long)
37432 vector float vec_nabs (vector float)
37433 vector double vec_nabs (vector double)
37434 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37436 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37437 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37438 * doc/extend.texi: Update the documentation file for the new built-in
37441 2017-01-16 Martin Sebor <msebor@redhat.com>
37443 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37446 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37448 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37449 UNSPEC_VSX__XXSPLTD to require special splat handling.
37451 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37454 * system.h: Poison strndup.
37456 2017-01-16 Alan Modra <amodra@gmail.com>
37459 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37462 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37464 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37466 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37468 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37469 call recog here. Assert that INSN_CODE (insn) is non-negative.
37471 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37474 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37476 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37477 in the currently scheduled RTL fragment.
37479 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37481 PR rtl-optimization/78751
37482 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37485 2017-01-14 Jeff Law <law@redhat.com>
37487 PR tree-optimization/79090
37488 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37489 variable length stores.
37490 (compute_trims): Delete dead assignment to *trim_tail.
37491 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37494 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37496 PR rtl-optimization/78626
37497 PR rtl-optimization/78727
37498 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37499 of a block, and split such blocks after everything else is finished.
37501 2017-01-14 Alan Modra <amodra@gmail.com>
37504 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37505 target legitimate_combined_insn.
37506 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37507 (rs6000_legitimate_combined_insn): New function.
37508 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37510 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37511 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37512 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37514 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37516 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37518 2017-01-13 Jeff Law <law@redhat.com>
37520 PR tree-optimization/33562
37521 PR tree-optimization/61912
37522 PR tree-optimization/77485
37523 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37525 (delete_dead_assignment): Likewise.
37526 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37527 statement to delete_dead_call and delete_dead_assignment.
37529 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37532 * substring-locations.c (format_warning_va): Strengthen case 1 so
37533 that both endpoints of the substring must be within the format
37534 range for just the substring to be printed.
37536 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37538 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37539 * config/i386/i386.c (ix86_target_string): Add missing options
37540 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37541 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37542 flags_other and ix86_target_other to flags2_other. Display unknown
37544 (ix86_valid_target_attribute_inner_p): Add missing options and
37545 reorder options by implied ISAs, as in ix86_target_string.
37547 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37549 * hash-table.h (hash_table::too_empty_p): New function.
37550 (hash_table::expand): Use it.
37551 (hash_table::traverse): Likewise.
37552 (hash_table::empty_slot): Use sizeof (value_type) instead of
37553 sizeof (PTR) to convert bytes to elements. Shrink the table
37554 if the current size is excessive for the current number of
37557 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37559 * ira-costs.c (record_reg_classes): Break from the inner loop
37560 early once alt_fail is known to be true. Update outer loop
37561 handling accordingly.
37563 2017-01-13 Jeff Law <law@redhat.com>
37565 * tree-ssa-dse.c (decrement_count): New function.
37566 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37567 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37568 when we know the partially dead statement is a mem* function.
37570 PR tree-optimization/61912
37571 PR tree-optimization/77485
37572 * tree-ssa-dse.c: Include expr.h.
37573 (maybe_trim_constructor_store): New function.
37574 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37576 PR tree-optimization/33562
37577 PR tree-optimization/61912
37578 PR tree-optimization/77485
37579 * doc/invoke.texi: Document new dse-max-object-size param.
37580 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37581 * tree-ssa-dse.c: Include params.h.
37582 (dse_store_status): New enum.
37583 (initialize_ao_ref_for_dse): New, partially extracted from
37585 (valid_ao_ref_for_dse, normalize_ref): New.
37586 (setup_live_bytes_from_ref, compute_trims): Likewise.
37587 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37588 (maybe_trim_partially_dead_store): Likewise.
37589 (maybe_trim_complex_store): Likewise.
37590 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37591 Track what bytes live from the original store. Return tri-state
37592 for dead, partially dead or live.
37593 (dse_dom_walker): Add constructor, destructor and new private members.
37594 (delete_dead_call, delete_dead_assignment): New extracted from
37596 (dse_optimize_stmt): Make a member of dse_dom_walker.
37597 Use initialize_ao_ref_for_dse.
37599 PR tree-optimization/33562
37600 PR tree-optimization/61912
37601 PR tree-optimization/77485
37602 * sbitmap.h (bitmap_count_bits): Prototype.
37603 (bitmap_clear_range, bitmap_set_range): Likewise.
37604 * sbitmap.c (bitmap_clear_range): New function.
37605 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37607 2017-01-13 Martin Liska <mliska@suse.cz>
37610 * function.c (set_cfun): Add new argument force.
37611 * function.h (set_cfun): Likewise.
37612 * ipa-inline-transform.c (inline_call): Use the function when
37613 strict alising from is dropped for function we inline to.
37615 2017-01-13 Richard Biener <rguenther@suse.de>
37617 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37618 for dumping GIMPLE INTEGER_CSTs.
37620 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37622 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37623 to 201112L since C++17.
37625 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37628 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37629 if -fsanitize=kernel-address is present.
37631 2017-01-13 Richard Biener <rguenther@suse.de>
37633 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37634 as _Literal ( type ) number in case usual suffixes do not
37635 preserve all information.
37637 2017-01-13 Richard Biener <rguenther@suse.de>
37639 PR tree-optimization/77283
37640 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37641 and ssa-iterators.h.
37642 (is_feasible_trace): Implement a cost model based on joiner
37645 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37648 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37649 char or short to __float128/_Float128 directly.
37651 2017-01-12 Martin Sebor <msebor@redhat.com>
37653 to -Wformat-overflow.
37654 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37655 (min_bytes_remaining): Same.
37656 (get_string_length): Same.
37657 (format_string): Same.
37658 (format_directive): Same.
37660 (pass_sprintf_length::handle_gimple_call): Same.
37662 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37664 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37665 info.nowrite calls with no lhs that can't throw. Return bool
37666 whether gsi_remove has been called or not.
37667 (pass_sprintf_length::handle_gimple_call): Return bool whether
37668 try_substitute_return_value called gsi_remove. Formatting fix.
37669 (pass_sprintf_length::execute): Don't use gsi_remove if
37670 handle_gimple_call returned true.
37673 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37674 be removed due to side-effects, don't remove following barrier nor
37675 turn the successor edge into fallthru edge.
37677 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37680 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37681 element-reversing loads and stores as not swappable.
37683 2017-01-12 Nathan Sidwell <nathan@acm.org>
37684 Nicolai Stange <nicstange@gmail.com>
37686 * combine.c (try_combine): Don't ignore result of overlap checking
37687 loop. Combine overlap & asm check into single loop.
37689 2017-01-12 Richard Biener <rguenther@suse.de>
37691 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37692 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37694 2017-01-12 Richard Biener <rguenther@suse.de>
37696 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37697 and TS_TARGET_OPTION directly derive from TS_BASE.
37698 * tree-core.h (tree_optimization_option): Derive from tree_base.
37699 (tree_target_option): Likewise.
37701 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37703 * config/i386/i386.c (memory_address_length): Increase len
37704 only when rip_relative_addr_p returns false.
37706 2017-01-11 Julia Koval <julia.koval@intel.com>
37708 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37709 (OPTION_MASK_ISA_SGX_SET): New.
37710 (ix86_handle_option): Handle OPT_msgx.
37711 * config.gcc: Added sgxintrin.h.
37712 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37713 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37714 * config/i386/i386.c (ix86_target_string): Add -msgx.
37716 (ix86_option_override_internal): Handle new options.
37717 (ix86_valid_target_attribute_inner_p): Add sgx.
37718 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37719 * config/i386/i386.opt: Add msgx.
37720 * config/i386/sgxintrin.h: New file.
37721 * config/i386/x86intrin.h: Add sgxintrin.h.
37723 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37726 * fold-const.c (maybe_nonzero_address): Return 1 for function
37728 (tree_single_nonzero_warnv_p): Don't handle function local objects
37732 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37735 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37738 * opts.c: Include "spellcheck.h"
37739 (struct string_fragment): New struct.
37740 (struct edit_distance_traits<const string_fragment &>): New
37742 (get_closest_sanitizer_option): New function.
37743 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37745 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37747 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37749 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37750 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37751 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37752 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37753 for initial die_offset if dwarf_split_debug_info.
37754 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37755 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37757 (output_skeleton_debug_sections): Formatting fix. Use
37758 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37759 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37761 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37763 * config/arm/cortex-a53.md: Add bypasses for
37764 cortex_a53_r2f_cvt.
37765 (cortex_a53_r2f): Only use for transfers.
37766 (cortex_a53_f2r): Likewise.
37767 (cortex_a53_r2f_cvt): Add reservation for conversions.
37768 (cortex_a53_f2r_cvt): Likewise.
37770 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37772 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37773 to all inlined functions, change static to extern.
37775 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37778 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37780 (arm_assemble_integer): Likewise.
37782 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37784 * config.gcc: Use new awk script to check CPU, FPU and architecture
37785 parameters for --with-... options.
37786 * config/arm/parsecpu.awk: New file
37787 * config/arm/arm-cpus.in: New file.
37788 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37790 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37792 * config/arm/t-arm: Update dependency rules.
37793 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37794 of processing .def files.
37795 * config/arm/genopt.sh: Deleted.
37796 * config/arm/gentune.sh: Deleted.
37797 * config/arm/arm-cores.def: Deleted.
37798 * config/arm/arm-arches.def: Deleted.
37799 * config/arm/arm-fpus.def: Deleted.
37800 * config/arm/arm-tune.md: Regenerated.
37801 * config/arm/arm-tables.opt: Regenerated.
37802 * config/arm/arm-cpu.h: New generated file.
37803 * config/arm/arm-cpu-data.h: New generated file.
37804 * config/arm/arm-cpu-cdata.h: New generated file.
37806 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37809 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37811 (input_varpool_node): Unpack dynamically_initialized bit.
37813 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37815 PR rtl-optimization/79032
37816 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37817 the alignment of the adjusted memory reference against that of MODE,
37818 instead of the alignment of the original memory reference.
37820 2017-01-11 Martin Jambor <mjambor@suse.cz>
37822 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37824 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37825 decorated functions.
37827 2017-01-11 Richard Biener <rguenther@suse.de>
37829 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37830 set range/nonnull info for PHI results. Do not set it on
37831 stmts marked for removal.
37833 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37835 * expr.c (store_field): In the bitfield case, fetch the return value
37836 from the registers before applying a single big-endian adjustment.
37837 Always do a final load for a BLKmode value not larger than a word.
37839 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37842 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37843 that we correctly handle column numbers greater than
37844 LINE_MAP_MAX_COLUMN_NUMBER.
37846 2017-01-10 Martin Sebor <msebor@redhat.com>
37848 PR middle-end/78245
37849 * gimple-ssa-sprintf.c (get_destination_size): Call
37850 {init,fini}object_sizes.
37851 * tree-object-size.c (addr_object_size): Adjust.
37852 (pass_through_call): Adjust.
37853 (pass_object_sizes::execute): Adjust.
37854 * tree-object-size.h (fini_object_sizes): Declare.
37856 2017-01-10 Martin Sebor <msebor@redhat.com>
37858 PR tree-optimization/78775
37859 * builtins.c (get_size_range): Move...
37860 * calls.c: ...to here.
37861 (alloc_max_size): Accept zero argument.
37862 (operand_signed_p): Remove.
37863 (maybe_warn_alloc_args_overflow): Call get_size_range.
37864 * calls.h (get_size_range): Declare.
37866 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37868 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37869 from TI's devices.csv file as of September 2016.
37870 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37872 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37874 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37875 * doc/invoke.texi: Likewise.
37876 * doc/md.texi: Likewise.
37877 * doc/objc.texi: Likewise.
37879 2017-01-10 Joshua Conner <joshconner@google.com>
37881 * config/arm/fuchsia-elf.h: New file.
37882 * config/fuchsia.h: New file.
37883 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37884 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37886 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37888 2016-01-10 Richard Biener <rguenther@suse.de>
37890 PR tree-optimization/79034
37891 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37892 Propagate out degenerate PHIs in the joiner.
37894 2017-01-10 Martin Liska <mliska@suse.cz>
37896 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37897 (sort_congruence_classes_by_decl_uid): Likewise.
37898 (sort_congruence_class_groups_by_decl_uid): Likewise.
37899 (sem_item_optimizer::merge_classes): Sort class, groups in these
37900 classes and members in the groups by DECL_UID of declarations.
37901 This would make merge operations stable.
37903 2017-01-10 Martin Liska <mliska@suse.cz>
37905 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37906 usage of m_classes_vec.
37907 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37908 (sem_item_optimizer::get_group_by_hash): Likewise.
37909 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37910 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37911 (sem_item_optimizer::verify_classes): Likewise.
37912 (sem_item_optimizer::process_cong_reduction): Likewise.
37913 (sem_item_optimizer::dump_cong_classes): Likewise.
37914 (sem_item_optimizer::merge_classes): Likewise.
37915 * ipa-icf.h (congruence_class_hash): Rename from
37916 congruence_class_group_hash. Remove declaration of m_classes_vec.
37918 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37920 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37921 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37922 * config.gcc: Add avx512vpopcntdqintrin.h.
37923 * config/i386/avx512vpopcntdqintrin.h: New.
37924 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37925 * config/i386/i386-builtin-types.def: Add new types.
37926 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37927 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37928 __builtin_ia32_vpopcountq_v8di_mask): New.
37929 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37930 __AVX512VPOPCNTDQ__.
37931 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37932 (PTA_AVX512VPOPCNTDQ): Define.
37933 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37934 TARGET_AVX512VPOPCNTDQ_P): Define.
37935 * config/i386/i386.opt: Add mavx512vpopcntdq.
37936 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37937 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37939 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37941 PR middle-end/77484
37942 * predict.def (PRED_CALL): Set to 67.
37944 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37946 * expr.c (store_field): In the bitfield case, if the value comes from
37947 a function call and is of an aggregate type returned in registers, do
37948 not modify the field mode; extract the value in all cases if the mode
37949 is BLKmode and the size is not larger than a word.
37951 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37954 * config/i386/cpuid.h: Fix undefined behavior.
37956 2017-01-04 Jeff Law <law@redhat.com>
37958 PR tree-optimization/79007
37959 PR tree-optimization/67955
37960 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37961 conservative for pt.null when flag_non_call_exceptions is on.
37963 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37965 PR translation/79019
37966 PR translation/79020
37967 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37968 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37969 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37971 * config/avr/avr.opt (maccumulate-args): Likewise.
37972 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37973 * common.opt (freport-bug): Likewise.
37974 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37975 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37976 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37977 translatable string.
37978 * config/i386/i386.c (function_value_32): Likewise.
37979 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37980 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37982 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37983 * common/config/msp430/msp430-common.c (msp430_handle_option):
37985 * symtab.c (symtab_node::verify_base): Likewise.
37986 * opts.c (set_debug_level): Likewise.
37987 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37988 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37989 missing whitespace to translatable strings.
37990 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37991 * config/sh/superh.h: Likewise.
37992 * config/i386/xopintrin.h: Likewise.
37993 * config/i386/znver1.md: Likewise.
37994 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37995 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37996 * double-int.h (struct double_int): Likewise.
37997 * double-int.c (div_and_round_double): Likewise.
37998 * wide-int.cc: Likewise.
37999 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38000 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38001 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38002 (crossjumps_occurred): ... this.
38003 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38006 PR tree-optimization/78899
38007 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38008 returning bool return struct loop *, NULL for failure and the new
38010 (versionable_outer_loop_p): Don't version outer loop if it has
38011 dont_vectorized bit set.
38012 (tree_if_conversion): When versioning outer loop, ensure
38013 tree_if_conversion is performed also on the inner loop of the
38014 non-vectorizable outer loop copy.
38015 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38016 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38017 prevent vectorization of it.
38018 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38019 the outer loop vectorization of the non-scalar version is attempted
38020 before vectorization of the inner loop in scalar version. If
38021 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38022 vectorization of its inner loop.
38023 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38024 has 2 inner loops, rename also on edges from bb whose single pred
38025 is outer_loop->header. Fix typo in function comment.
38027 2017-01-09 Martin Sebor <msebor@redhat.com>
38030 * asan.c (asan_emit_stack_protection): Increase local buffer size
38031 to avoid snprintf truncation warning.
38033 2017-01-09 Andrew Pinski <apinski@cavium.com>
38035 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38036 to reference thunderx2t99 for the tuning structure
38037 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38039 (thunderx2t99_extra_costs): This.
38040 * config/aarch64/aarch64-tune.md: Regenerate.
38041 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38042 (vulcan_addrcost_table): This.
38043 (vulcan_regmove_cost): Rename to ...
38044 (thunderx2t99_regmove_cost): This.
38045 (vulcan_vector_cost): Rename to ...
38046 (thunderx2t99_vector_cost): this.
38047 (vulcan_branch_cost): Rename to ...
38048 (thunderx2t99_branch_cost): This.
38049 (vulcan_tunings): Rename to ...
38050 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38051 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38053 2017-01-09 Martin Jambor <mjambor@suse.cz>
38057 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38058 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38059 (propagate_vr_accross_jump_function): Use the above function for all
38060 value range computations for pass-through jump functions and type
38061 converasion from explicit value range values.
38062 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38063 parameters from TYPE_ARG_TYPES.
38064 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38065 (ipa_write_node_info): Stream type of the actual argument.
38066 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38068 2017-01-09 Martin Liska <mliska@suse.cz>
38071 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38072 (lookup_compiler): Do not show error message with have_E.
38074 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38076 PR tree-optimization/78938
38077 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38078 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38079 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38080 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38083 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38085 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38088 2017-01-09 Richard Biener <rguenther@suse.de>
38090 PR tree-optimization/78997
38091 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38092 name condition properly.
38094 2017-01-09 Richard Biener <rguenther@suse.de>
38097 * dwarf2out.c (is_cxx): New overload with context.
38098 (is_naming_typedef_decl): Use it.
38100 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38102 * invoke.texi (Option Summary): Correct spacing in option lists
38103 and add line breaks to fix over-long lines.
38105 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38107 PR middle-end/17660
38109 * extend.texi (Common Variable Attributes): Add xref to GCC
38110 Internals manual to explain mode attribute keywords.
38112 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38115 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38116 and Preprocessor Options.
38117 (Options for Linking): Document -pthread here....
38118 (RS/6000 and PowerPC Options): ...not here.
38119 (Solaris 2 Options): ...or here.
38120 * doc/cppopts.texi: Document -pthread.
38122 2017-01-08 Martin Sebor <msebor@redhat.com>
38124 PR middle-end/77708
38125 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38126 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38127 New member functions.
38128 (format_directive): Used them.
38130 (pass_sprintf_length::handle_gimple_call): Same.
38131 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38132 to avoid truncation for any argument.
38133 (extract_affine_mul): Same.
38134 * tree.c (get_file_function_name): Same.
38136 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38138 PR middle-end/77484
38139 * predict.def (PRED_INDIR_CALL): Set to 86.
38141 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38143 PR preprocessor/54124
38144 * doc/cppopts.texi: Reformat -d subtable to list the full name
38145 of the options. Add cross-reference to the docs for the general
38146 compiler -d options.
38147 * doc/invoke.texi (Developer Options): Add cross-reference to the
38148 preprocessor-specific -d option documentation.
38150 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38152 PR preprocessor/13498
38153 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38154 redudant material, and reflect new command-line options.
38155 (System Headers): Likewise.
38157 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38159 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38160 -isystem, and -idirafter. Copy-edit.
38161 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38162 default for -ftrack-macro-expansion. Delete obsolete and
38163 badly-formatted implementation details about -fdebug-cpp output.
38164 * doc/cppwarnopts.texi: Copy-edit.
38166 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38169 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38170 that the transition from a max line width >= 1<<10 to narrower
38171 lines works correctly.
38173 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38175 * doc/options.texi (PerFunction): New.
38176 * opt-functions.awk (switch_flags): Map both Optimization and
38177 PerFunction to CL_OPTIMIZATION.
38178 * opth-gen.awk: Test for PerFunction flag along with
38180 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38181 it only when the latter is present. Skip those that don't in
38182 the hash function generator.
38183 * common.opt (fvar-tracking): Mark as PerFunction instead of
38185 (fvar-tracking-assignments): Likewise.
38186 (fvar-tracking-assignments-toggle): Likewise.
38187 (fvar-tracking-uninit): Likewise.
38189 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38191 PR translation/79018
38192 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38195 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38198 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38199 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38200 TARGET_LONG_JUMP_TABLE_OFFSETS.
38201 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38202 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38203 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38204 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38205 * config/m68k/m68k.md (tablejump expander): Likewise.
38206 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38207 TARGET_LONG_JUMP_TABLE_OFFSETS.
38208 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38209 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38211 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38212 David Holsgrove <david.holsgrove@xilinx.com>
38214 * common/config/microblaze/microblaze-common.c
38215 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38216 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38218 * config/microblaze/microblaze.c (microblaze_must_save_register)
38219 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38221 (microblaze_eh_return): New function.
38222 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38223 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38224 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38225 * config/microblaze/microblaze.md (eh_return): New pattern.
38227 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38229 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38230 GCC_DIAGNOSTIC_STRINGIFY): Define.
38232 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38234 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38236 * config/arm/arm.md (<mcrr>): New.
38238 * config/arm/arm.c (arm_arch5te): New.
38239 (arm_option_override): Set arm_arch5te.
38240 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38242 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38243 (arm_mcrr_qualifiers): ... this. New.
38244 (MRRC_QUALIFIERS): Define to...
38245 (arm_mrrc_qualifiers): ... this. New.
38246 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38248 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38249 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38250 (MRRCI, mrrc, MRRC): New.
38251 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38252 VUNSPEC_MRRC2): New.
38254 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38256 * config/arm/arm.md (<mcr>): New.
38258 * config/arm/arm.c (arm_coproc_builtin_available): Add
38259 support for mcr, mrc, mcr2 and mrc2.
38260 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38261 (arm_mcr_qualifiers): ... this. New.
38262 (MRC_QUALIFIERS): Define to ...
38263 (arm_mrc_qualifiers): ... this. New.
38264 (MCR_QUALIFIERS): Define to ...
38265 (arm_mcr_qualifiers): ... this. New.
38266 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38268 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38269 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38270 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38271 VUNSPEC_MRC2): New.
38273 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38275 * config/arm/arm.md (*ldc): New.
38279 * config/arm/arm.c (arm_coproc_builtin_available): Add
38280 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38281 (arm_coproc_ldc_stc_legitimate_address): New.
38282 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38283 'qualifier_const_pointer'.
38284 (LDC_QUALIFIERS): Define to...
38285 (arm_ldc_qualifiers): ... this. New.
38286 (STC_QUALIFIERS): Define to...
38287 (arm_stc_qualifiers): ... this. New.
38288 * config/arm/arm-protos.h
38289 (arm_coproc_ldc_stc_legitimate_address): New.
38290 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38291 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38292 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38293 stc2, stcl, stc2l): New.
38294 * config/arm/constraints.md (Uz): New.
38295 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38296 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38297 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38298 VUNSPEC_STC2L): New.
38300 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38302 * config/arm/arm.md (<cdp>): New.
38303 * config/arm/arm.c (neon_const_bounds): Rename this ...
38304 (arm_const_bounds): ... this.
38305 (arm_coproc_builtin_available): New.
38306 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38307 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38308 (CDP_QUALIFIERS): Define to...
38309 (arm_cdp_qualifiers): ... this. New.
38311 (arm_expand_builtin_args): Add case for 6 arguments.
38312 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38313 (arm_const_bounds): ... this.
38314 (arm_coproc_builtin_available): New.
38315 * config/arm/arm_acle.h (__arm_cdp): New.
38317 * config/arm/arm_acle_builtins.def (cdp): New.
38319 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38320 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38321 'arm_const_bounds'.
38322 * config/arm/types.md (coproc): New.
38323 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38324 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38325 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38326 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38328 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38330 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38331 (UBINOP_QUALIFIERS): New.
38333 (acle_builtin_data): New. Change comment.
38334 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38335 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38336 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38337 arm_acle_builtins.def.
38338 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38339 (arm_init_acle_builtins): New.
38340 (CRC32_BUILTIN): Remove.
38341 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38342 crc32cb, crc32ch and crc32cw.
38343 (arm_init_crc32_builtins): Remove.
38344 (arm_init_builtins): Use arm_init_acle_builtins rather
38345 than arm_init_crc32_builtins.
38346 (arm_expand_acle_builtin): New.
38347 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38348 * config/arm/arm_acle_builtins.def: New.
38350 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38352 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38353 (arm_builtin_datum): ... this.
38354 (arm_init_neon_builtin): Rename to ...
38355 (arm_init_builtin): ... this. Add a new parameters PREFIX
38356 and USE_SIG_IN_NAME.
38357 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38358 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38359 'arm_builtin_datum'.
38360 (arm_init_vfp_builtins): Likewise.
38361 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38362 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38363 (arm_expand_neon_args): Rename to ...
38364 (arm_expand_builtin_args): ... this. Rename builtin_arg
38365 enum values and differentiate between ARG_BUILTIN_MEMORY
38366 and ARG_BUILTIN_NEON_MEMORY.
38367 (arm_expand_neon_builtin_1): Rename to ...
38368 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38369 values, arm_expand_builtin_args and add bool parameter NEON.
38370 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38371 (arm_expand_vfp_builtin): Likewise.
38372 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38374 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38376 PR middle-end/77484
38377 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38378 * predict.c (tree_estimate_probability_bb): Reverse direction of
38379 polymorphic call predictor.
38381 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38383 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38384 (determine_pass_name_match): ...this new function and...
38385 (should_skip_pass_p): ...this new function.
38387 2017-01-06 Nathan Sidwell <nathan@acm.org>
38389 * ipa-visibility.c (function_and_variable_visibility): Reformat
38390 comments and long lines. Remove extrneous if.
38391 * symtab.c (symtab_node::make_decl_local): Fix code format.
38392 (symtab_node::set_section_for_node): Fix comment typo.
38394 2017-01-06 Martin Liska <mliska@suse.cz>
38397 * lra-constraints.c: Rename invariant to lra_invariant.
38398 * predict.c (set_even_probabilities): Initialize e to NULL.
38400 2017-01-05 Martin Sebor <msebor@redhat.com>
38402 PR tree-optimization/78910
38403 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38404 (format_integer): Correct off-by-one error in the handling
38405 of precision with negative numbers in signed conversions..
38407 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38409 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38411 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38413 PR tree-optimization/71016
38414 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38415 factor_out_conditional_conversion. Formatting fix.
38416 (factor_out_conditional_conversion): Add cond_stmt argument.
38417 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38418 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38421 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38423 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38424 read-rtl-function.o, and selftest-rtl.o.
38425 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38426 (selftest::aarch64_test_loading_full_dump): New function.
38427 (selftest::aarch64_run_selftests): New function.
38428 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38429 selftest::aarch64_run_selftests.
38430 * config/i386/i386.c
38431 (selftest::ix86_test_loading_dump_fragment_1): New function.
38432 (selftest::ix86_test_loading_call_insn): New function.
38433 (selftest::ix86_test_loading_full_dump): New function.
38434 (selftest::ix86_test_loading_unspec): New function.
38435 (selftest::ix86_run_selftests): Call the new functions.
38436 * emit-rtl.c (maybe_set_max_label_num): New function.
38437 * emit-rtl.h (maybe_set_max_label_num): New decl.
38438 * function.c (instantiate_decls): Guard call to
38439 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38440 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38442 * gensupport.c (gen_reader::gen_reader): Pass "false"
38443 for new "compact" param of rtx_reader.
38444 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38445 rather than an empty string for NULL strings.
38446 * read-md.c: Potentially include config.h rather than bconfig.h.
38447 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38448 (have_error): New global, copied from errors.c.
38449 (md_reader::read_name): Rename to...
38450 (md_reader::read_name_1): ...this, adding "out_loc" param,
38451 and converting "missing name or number" to returning false, rather
38453 (md_reader::read_name): Reimplement in terms of read_name_1.
38454 (md_reader::read_name_or_nil): New function.
38455 (md_reader::read_string): Handle "(nil)" by returning NULL.
38456 (md_reader::md_reader): Add new param "compact".
38457 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38458 (md_reader::read_file): New method.
38459 * read-md.h (md_reader::md_reader): Add new param "compact".
38460 (md_reader::read_file): New method.
38461 (md_reader::is_compact): New accessor.
38462 (md_reader::read_name): Convert return type from void to file_location.
38463 (md_reader::read_name_or_nil): New decl.
38464 (md_reader::read_name_1): New decl.
38465 (md_reader::m_compact): New field.
38466 (noop_reader::noop_reader): Pass "false" for new "compact" param
38468 (rtx_reader::rtx_reader): Add new "compact" param.
38469 (rtx_reader::read_rtx_operand): Make virtual and convert return
38470 type from void to rtx.
38471 (rtx_reader::read_until): New decl.
38472 (rtx_reader::handle_any_trailing_information): New virtual function.
38473 (rtx_reader::postprocess): New virtual function.
38474 (rtx_reader::finalize_string): New virtual function.
38475 (rtx_reader::m_in_call_function_usage): New field.
38476 (rtx_reader::m_reuse_rtx_by_id): New field.
38477 * read-rtl-function.c: New file.
38478 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38479 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38480 (selftest::verify_three_block_rtl_cfg): New decl.
38481 * read-rtl-function.h: New file.
38482 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38483 For host, include function.h, memmodel.h, and emit-rtl.h.
38484 (one_time_initialization): New function.
38485 (struct compact_insn_name): New struct.
38486 (compact_insn_names): New array.
38487 (find_code): Handle insn codes in compact dumps.
38488 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38489 (bind_subst_iter_and_attr): Likewise.
38490 (add_condition_to_string): Likewise.
38491 (add_condition_to_rtx): Likewise.
38492 (apply_attribute_uses): Likewise.
38493 (add_current_iterators): Likewise.
38494 (apply_iterators): Likewise.
38495 (initialize_iterators): Guard usage of apply_subst_iterator with
38496 #ifdef GENERATOR_FILE.
38497 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38498 (md_reader::read_mapping): Likewise.
38499 (add_define_attr_for_define_subst): Likewise.
38500 (add_define_subst_attr): Likewise.
38501 (read_subst_mapping): Likewise.
38502 (check_code_iterator): Likewise.
38503 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38505 (one_time_initialization): New function.
38506 (rtx_reader::read_until): New method.
38507 (read_flags): New function.
38508 (parse_reg_note_name): New function.
38509 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38510 Handle reuse_rtx ids.
38511 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38512 Add parsing support for RTL dumps, mirroring the special-cases in
38513 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38514 values, and calling handle_any_trailing_information.
38515 (rtx_reader::read_rtx_operand): Convert return type from void
38516 to rtx, returning return_rtx. Handle case 'e'. Call
38517 finalize_string on XSTR and XTMPL fields.
38518 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38519 "(nil)" values were omitted. Call the postprocess vfunc on the
38521 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38522 class ctor. Initialize m_in_call_function_usage. Call
38523 one_time_initialization.
38524 * rtl-tests.c (selftest::test_uncond_jump): Call
38525 set_new_first_and_last_insn.
38526 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38527 * selftest-rtl.c: New file.
38528 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38529 (selftest::get_insn_by_uid): New decl.
38530 * selftest-run-tests.c (selftest::run_tests): Call
38531 read_rtl_function_c_tests.
38532 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38533 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38536 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38538 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38539 operands in a special way. Assert that pos+len <= mode precision.
38541 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38543 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38544 3 argument Alias with unlimited for the negative form.
38545 (fno-vect-cost-model): Removed.
38547 2017-01-05 Martin Liska <mliska@suse.cz>
38549 * hsa-gen.c (gen_hsa_divmod): New function.
38550 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38552 2017-01-05 Martin Liska <mliska@suse.cz>
38555 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38558 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38560 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38561 small constant length operands.
38563 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38565 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38566 between loop iterations.
38568 2017-01-05 Martin Liska <mliska@suse.cz>
38571 * gimplify.c (gimplify_decl_expr): Compare to
38572 asan_poisoned_variables instread of checking flags.
38573 (gimplify_target_expr): Likewise.
38574 (gimplify_expr): Likewise.
38575 (gimplify_function_tree): Conditionally initialize
38576 asan_poisoned_variables.
38578 2017-01-04 Jeff Law <law@redhat.com>
38580 PR tree-optimizatin/78812
38581 * rtl.h (contains_mem_rtx_p): Prototype.
38582 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38583 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38584 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38585 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38586 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38588 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38590 * input.c (assert_char_at_range): Default-initialize actual_range.
38592 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38594 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38595 to match the caller.
38597 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38599 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38600 insns after final jump in test to emit dummy move.
38602 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38604 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38605 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38607 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38609 * multiple_target.c (create_dispatcher_calls): Init e_next.
38610 * tree-ssa-loop-split.c (split_loop): Init border.
38611 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38614 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38619 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38620 (altivec_register_operand): Do not return true if the operand
38621 contains a SUBREG mixing SImode and SFmode.
38622 (vsx_register_operand): Likewise.
38623 (vsx_reg_sfsubreg_ok): New predicate.
38624 (vfloat_operand): Do not return true if the operand contains a
38625 SUBREG mixing SImode and SFmode.
38626 (vint_operand): Likewise.
38627 (vlogical_operand): Likewise.
38628 (gpc_reg_operand): Likewise.
38629 (int_reg_operand): Likewise.
38630 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38631 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38632 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38634 (rs6000_emit_move_si_sf_subreg): New helper function.
38635 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38636 fixup SUBREGs involving SImode and SFmode.
38637 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38638 numbers for the new peephole2 optimization.
38639 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38640 the GLIBC math library that do AND/IOR/XOR operations on single
38641 precision floating point.
38642 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38643 target macros to say whether we need to avoid SUBREGs mixing
38645 (TARGET_ALLOW_SF_SUBREG): Likewise.
38646 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38647 (UNSPEC_SI_FROM_SF): Likewise.
38648 (iorxor): Change spacing.
38649 (and_ior_xor): New iterator for AND, IOR, and XOR.
38650 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38651 (movdi_from_sf_zero_ext): Likewise.
38652 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38653 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38654 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38655 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38656 (fms<mode>4): Likewise.
38657 (fnma<mode>4): Likewise.
38658 (fnms<mode>4): Likewise.
38659 (nfma<mode>4): Likewise.
38660 (nfms<mode>4): Likewise.
38662 2017-01-04 Marek Polacek <polacek@redhat.com>
38665 * doc/invoke.texi: Document -Wpointer-compare.
38667 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38669 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38672 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38673 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38676 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38679 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38680 documentation of the powerpc_popcntb_ok attribute.
38681 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38682 code to issue warning messages if a requested CPU configuration is
38683 not supported by the binary (assembler and loader) toolchain.
38684 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38685 made to define a built-in function that has been disabled.
38686 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38687 made to define a built-in function that has been disabled.
38688 (altivec_init_builtins): Add comment explaining why definition
38689 of the DST built-in functions is not preceded by an assertion
38690 check. Add assertions to prevent ICE if attempts are made to
38691 define an altivec predicate or an abs* built-in function that has
38693 (htm_init_builtins): Add comment explaining why definition of the
38694 htm built-in functions is not preceded by an assertion check.
38696 2017-01-04 Jeff Law <law@redhat.com>
38698 PR tree-optimizatin/67955
38699 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38700 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38701 the points-to solution does not include pt_null. Use DECL_PT_UID
38704 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38706 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38707 Use gen_int_mode instead of gen_lopwart for const_int operands.
38709 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38711 PR tree-optimization/71563
38712 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38713 out of range value - has low bits known to be zero.
38715 2017-01-04 Alan Modra <amodra@gmail.com>
38717 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38718 * configure: Regenerate.
38719 * config.in: Regenerate.
38721 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38724 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38725 a substring of the message, but strcmp with the whole message. Ifdef
38726 ENABLE_NLS, translate the message first using dgettext.
38728 2017-01-03 Jeff Law <law@redhat.com>
38730 PR tree-optimizatin/78856
38731 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38732 (mark_threaded_blocks): Remove code to truncate thread paths that
38733 cross multiple loop headers. Instead invalidate the cached loop
38734 iteration information and handle case of a thread path walking
38735 into an irreducible region.
38737 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38740 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38741 assertions. Add support for doing the signbit if the IEEE 128-bit
38742 floating point value is in a GPR.
38743 * config/rs6000/rs6000.md (Fsignbit): Delete.
38744 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38745 Update the length attribute if the value is in a GPR.
38746 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38747 the sign or zero extension instruction, since the value is always 0/1.
38748 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38751 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38752 extracting SImode to a GPR register so that we can generate a
38753 store, limit the vector to be in a traditional Altivec register
38754 for the vextuwrx instruction.
38756 2017-01-03 Ian Lance Taylor <iant@google.com>
38758 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38760 2017-01-03 Martin Sebor <msebor@redhat.com>
38762 PR tree-optimization/78696
38763 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38764 precision. Use MPFR for %f for greater fidelity. Correct handling
38766 (pass_sprintf_length::compute_format_length): Set width and precision
38767 specified by asrerisk to void_node for vararg functions.
38768 (try_substitute_return_value): Adjust dump output.
38770 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38772 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38774 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38776 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38777 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38778 -mlra/-mno-lra was passed to the compiler.
38780 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38782 PR rtl-optimization/65618
38783 * emit-rtl.c (try_split): Move initialization of "before" and
38784 "after" to just before the call to emit_insn_after_setloc.
38786 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38788 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38790 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38792 * dwarf2out.c (gen_enumeration_type_die): When
38793 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38795 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38797 PR tree-optimization/78965
38798 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38799 Change first argument from const call_info & to call_info &. For %n
38800 set info.nowrite to false.
38802 PR middle-end/78901
38803 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38804 possibly throwing calls.
38806 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38807 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38808 and fns handling, rather than in a separate case SSA_NAME.
38810 2017-01-02 Jeff Law <law@redhat.com>
38812 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38813 fixes for first_period and second_period variables.
38815 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38818 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38819 (*insvqi_1): New insn pattern.
38820 (*insvqi_1_mem_rex64): Ditto.
38821 (*insvqi_2): Ditto.
38822 (*insvqi_3): Rename from *insvqi.
38824 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38826 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38828 * doc/cfg.texi (Edges): Remove reference to Java.
38829 (Maintaining the CFG): Ditto.
38831 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38833 PR middle-end/77674
38834 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38835 transparent aliases.
38837 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38839 PR middle-end/77484
38840 * predict.def (PRED_CALL): Update hitrate.
38841 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38842 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38843 into direct/indirect/polymorphic variants.
38845 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38847 Update copyright years.
38849 * gcc.c (process_command): Update copyright notice dates.
38850 * gcov-dump.c (print_version): Ditto.
38851 * gcov.c (print_version): Ditto.
38852 * gcov-tool.c (print_version): Ditto.
38853 * gengtype.c (create_file): Ditto.
38854 * doc/cpp.texi: Bump @copying's copyright year.
38855 * doc/cppinternals.texi: Ditto.
38856 * doc/gcc.texi: Ditto.
38857 * doc/gccint.texi: Ditto.
38858 * doc/gcov.texi: Ditto.
38859 * doc/install.texi: Ditto.
38860 * doc/invoke.texi: Ditto.
38862 Copyright (C) 2017 Free Software Foundation, Inc.
38864 Copying and distribution of this file, with or without modification,
38865 are permitted in any medium without royalty provided the copyright
38866 notice and this notice are preserved.