1 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
4 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
6 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
7 Alan Hayward <alan.hayward@arm.com>
8 David Sherwood <david.sherwood@arm.com>
10 * poly-int.h: New file.
11 * poly-int-types.h: Likewise.
12 * coretypes.h: Include them.
13 (POLY_INT_CONVERSION): Define.
14 * target.def (estimated_poly_value): New hook.
15 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
16 * doc/tm.texi: Regenerate.
17 * doc/poly-int.texi: New file.
18 * doc/gccint.texi: Include it.
19 * doc/rtl.texi: Describe restrictions on subreg modes.
20 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
21 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
22 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
23 * targhooks.h (default_estimated_poly_value): Declare.
24 * targhooks.c (default_estimated_poly_value): New function.
25 * target.h (estimated_poly_value): Likewise.
26 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
27 (wi::unary_traits): Delete.
28 (wi::binary_traits::signed_shift_result_type): Define for
29 offset_int << HOST_WIDE_INT, etc.
30 (generic_wide_int::operator <<=): Define for all types and use
31 wi::lshift instead of <<.
32 (wi::hwi_with_prec): Add a default constructor.
33 (wi::ints_for): New class.
34 (operator <<): Define for all wide-int types.
35 (operator /): New function.
36 (operator %): Likewise.
37 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
38 (ASSERT_MAYBE_NE_AT): New macros.
40 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
41 Dominik Vogt <vogt@linux.vnet.ibm.com>
44 * emit-rtl.c (init_emit): Remove ??? comment.
45 * explow.c (get_dynamic_stack_size): Take known alignment of stack
46 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
47 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
48 alignment of 3 virtual registers to BITS_PER_WORD.
50 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
52 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
54 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
55 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
57 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
60 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
62 2017-12-13 Jakub Jelinek <jakub@redhat.com>
64 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
65 with the exception of debug begin stmt markers appear before
69 * final.c (rest_of_handle_final): Call variable_tracking_main only
70 if !flag_var_tracking.
72 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
73 Jakub Jelinek <jakub@redhat.com>
77 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
78 labels debug stmts that can only appear after labels.
80 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
82 PR rtl-optimization/82398
83 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
84 EXPR_USEFULNESS in priority comparison.
86 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
88 PR rtl-optimization/83393
89 * combine.c (move_deaths): If reg_stat points to a too new insn in
90 last_death, do not use it: find the proper insn instead.
92 2017-12-12 Jeff Law <law@redhat.com>
94 PR tree-optimization/83298
95 PR tree-optimization/83362
96 PR tree-optimization/83383
97 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
98 push_value_range a public interface. Add new argument to
99 record_ranges_from_stmt.
100 * gimple-ssa-evrp-analyze.c
101 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
102 Update comments. Handle recording temporary equivalences.
103 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
104 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
105 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
106 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
107 gimple-ssa-evrp-analyze.h.
108 (record_temporary_equivalences_from_phis): Add new argument. When
109 the PHI arg is an SSA_NAME, set the result's range to the range
111 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
113 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
114 Pass it down to children as needed.
115 (thread_outgoing_edges): Likewise.
116 (thread_across_edge): Likewise. Push/pop range state as needed.
117 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
119 2017-12-12 Julia Koval <julia.koval@intel.com>
121 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
122 (PTA_CANNONLAKE): Remove PTA_CLWB.
124 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
127 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
128 (vcondv2div2df): Likewise.
129 (vconduv2dfv2di): Likewise.
131 2017-12-12 Jakub Jelinek <jakub@redhat.com>
133 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
134 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
135 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
136 ATTR_MATHFN_FPROUNDING_ERRNO.
138 2017-12-12 Richard Biener <rguenther@suse.de>
140 PR tree-optimization/83385
141 * graphite-scop-detection.c (get_order, order): Remove.
142 (bb_to_rpo): New global.
144 (build_scops): Sort pbbs in RPO order.
146 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
148 * combine.c (simplify_set): Do not transform subregs to zero_extends
149 if the destination mode is a vector mode.
151 2017-12-12 Jakub Jelinek <jakub@redhat.com>
153 PR tree-optimization/83359
154 * tree-cfg.h (fold_loop_internal_call): Declare.
155 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
156 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
157 (find_loop_dist_alias): New function.
158 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
159 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
162 PR tree-optimization/80631
163 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
164 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
165 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
166 hardcoding zero as the value if COND_EXPR is never true. For
167 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
168 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
169 hardcoding MAX_EXPR as the reduction operation.
170 (is_nonwrapping_integer_induction): Allow negative step.
171 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
172 vect_create_epilog_for_reduction, if no value is suitable, don't
173 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
175 2017-12-12 Richard Biener <rguenther@suse.de>
177 PR tree-optimization/81889
178 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
179 range info from the non-wrapping IV instead of just the range
182 2017-12-12 Julia Koval <julia.koval@intel.com>
184 * config.gcc: Add vaesintrin.h.
185 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
187 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
188 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
190 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
191 * config/i386/immintrin.h: Include vaesintrin.h.
192 * config/i386/sse.md (vaesdec_<mode>): New pattern.
193 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
194 _mm_aesdec_epi128): New intrinsics.
196 2017-12-12 Julia Koval <julia.koval@intel.com>
198 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
199 OPTION_MASK_ISA_VAES_UNSET): New.
200 (ix86_handle_option): Handle -mvaes.
201 * config/i386/cpuid.h: Define bit_VAES.
202 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
203 * config/i386/i386-c.c (__VAES__): New.
204 * config/i386/i386.c (ix86_target_string): Add -mvaes.
205 (ix86_valid_target_attribute_inner_p): Ditto.
206 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
207 * config/i386/i386.opt: Add -mvaes.
208 * doc/invoke.texi: Ditto.
210 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
212 * debug.h (gcc_debug_hooks): Add inline_entry.
213 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
214 * debug.c (do_nothing_debug_hooks): Likewise.
215 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
216 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
217 (dwarf2_lineno_debug_hooks): Likewise.
219 * common.opt (gstatement-frontiers): New, setting
220 debug_nonbind_markers_p.
221 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
222 * toplev.c (process_options): Autodetect value for debug statement
224 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
225 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
227 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
228 markers. Integrate source bind into debug stmt expand loop.
229 (pass_expand::execute): Check debug marker limit. Avoid deep
230 TER and expand debug locations for debug bind insns only.
231 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
233 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
234 * final.c (reemit_insn_block_notes): Take current block from
235 nonbind markers. Declare note where it's first set.
236 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
237 begin stmt markers if enabled.
238 (notice_source_line): Handle nonbind markers. Fail if their
239 location is unknown or that of builtins.
240 (rest_of_handle_final): Convert begin stmt markers to notes if
241 var-tracking didn't run.
242 (rest_of_clean_state): Skip begin stmt markers.
243 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
245 * function.c (allocate_struct_function): Set begin_stmt_markers.
246 * function.h (struct function): Add debug_marker_count counter
247 and debug_nonbind_markers flag.
248 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
249 * gimple-low.c (lower_function_body): Adjust
250 debug_nonbind_markers.
251 (lower_stmt): Drop or skip gimple debug stmts.
252 (lower_try_catch): Skip debug stmts.
253 * gimple.c (gimple_build_debug_begin_stmt): New.
254 (gimple_copy): Increment debug_marker_count if copying one.
255 * gimple.h (gimple_build_debug_begin_stmt): Declare.
256 * gimplify.c (rexpr_location): New.
257 (rexpr_has_location): New.
258 (warn_switch_unreachable_r): Handle gimple debug stmts.
259 (shortcut_cond_r): Call expr_location.
261 (find_goto_label): New.
262 (shortcut_cond_expr): Call expr_has_location, expr_location, and
264 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
266 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
267 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
268 (LANG_HOOKS_INITIALIZER): ... this.
269 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
270 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
272 (update_ebb_live_info): Skip debug insn markers.
273 * lra.c (debug_insn_static_data): Rename to...
274 (debug_bind_static_data): ... this.
275 (debug_marker_static_data): New.
276 (lra_set_insn_recog_data): Select one of the above depending
278 (lra_update_isn_regno_info): Don't assume debug insns have
280 (push_insns): Skip debug insns.
281 * lto-streamer-in.c (input_function): Drop debug stmts
282 depending on active options. Adjust debug_nonbind_markers.
283 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
284 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
285 begin stmt marker notes.
286 (print_insn): Likewise.
287 * recog.c (extract_insn): Recognize rtl for debug markers.
288 * rtl.def (DEBUG_MARKER): New.
289 * tree-inline.c: Include params.h.
290 (remap_gimple_stmt): Handle nonbind markers.
291 (maybe_move_debug_stmts_to_successors): Likewise.
292 (copy_debug_stmt): Likewise.
293 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
294 markers regardless of no side effects.
295 (tsi_link_before): Don't update container's side effects when adding
297 (tsi_link_after): Likewise.
298 (expr_first): Skip begin stmt markers.
299 (expr_last): Likewise.
300 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
301 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
302 Disregard nonbind markers.
303 * tree.c (make_node_stat): Don't set side effects for begin stmt
305 (build1_stat): Likewise.
306 * tree.def (DEBUG_BEGIN_STMT): New.
307 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
308 * var-tracking.c (delete_debug_insns): Renamed to...
309 (delete_vta_debug_insns): ... this.
310 (reemit_marker_as_note): New.
311 (vt_initialize): Reemit markers.
312 (delete_vta_debug_insns): Likewise.
313 (vt_debug_insns_local): Reemit or delete markers.
314 (variable_tracking_main_1): Likewise.
315 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
316 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
317 (gimple_debug_nonbind_marker_p): New.
318 (gimple_build_debug_bind): Adjust.
319 (gimple_build_debug_begin_stmt): New.
320 * doc/invoke.texi (max-debug-marker-count): New param.
321 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
322 (debug_parameter_ref, debug_marker): New.
323 (NOTE_INSN_BEGIN_STMT): New.
324 (DEBUG_INSN): Describe begin stmt markers.
326 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
327 without debug insns, we wouldn't, but clean up debug insns
328 after a control flow insn nevertheless.
330 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
331 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
332 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
333 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
334 (rtl_tidy_fallthru_edge): Likewise.
335 (rtl_verify_fallthru): Likewise.
336 (rtl_verify_bb_layout): Likewise.
337 (skip_insns_after_block): Likewise.
338 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
339 * dwarf2out.c: Include print-rtl.h.
340 (dwarf2out_next_real_insn): New.
341 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
342 Dump debug binds in asm comments.
343 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
344 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
345 callers to use gsi_start_nondebug_bb instead.
346 (gsi_after_labels): Skip gimple debug stmts.
347 (gsi_start_nondebug): New.
348 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
349 (proper_loop_form_for_interchange): Adjust.
350 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
351 * gimple.h (gimple_seq_last_nondebug_stmt): New.
352 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
353 (collect_fallthrough_labels): Likewise.
354 (should_warn_for_implicit_fallthrough): Likewise.
355 (warn_implicit_fallthrough_r): Likewise.
356 (expand_FALLTHROUGH_r): Likewise.
357 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
358 (graphite_copy_stmts_from_block): Skip nonbind markers.
359 * haifa-sched.c (sched_extend_bb): Skip debug insns.
360 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
361 * jump.c (clean_barriers): Skip debug insns.
362 * omp-expand.c (expand_parallel_call): Skip debug insns.
363 (expand_task_call): Likewise.
364 (remove_exit_barrier): Likewise.
365 (expand_omp_taskreg): Likewise.
366 (expand_omp_for_init_counts): Likewise.
367 (expand_omp_for_generic): Likewise.
368 (expand_omp_for_static_nochunk): Likewise.
369 (expand_omp_for_static_chunk): Likewise.
370 (expand_omp_simd): Likewise.
371 (expand_omp_taskloop_for_outer): Likewise.
372 (expand_omp_taskloop_for_inner): Likewise.
373 (expand_oacc_for): Likewise.
374 (expand_omp_sections): Likewise.
375 (expand_omp_single): Likewise.
376 (expand_omp_synch): Likewise.
377 (expand_omp_atomic_load): Likewise.
378 (expand_omp_atomic_store): Likewise.
379 (expand_omp_atomic_fetch_op): Likewise.
380 (expand_omp_atomic_pipeline): Likewise.
381 (expand_omp_atomic_mutex): Likewise.
382 (expand_omp_target): Likewise.
383 (grid_expand_omp_for_loop): Likewise.
384 (grid_expand_target_grid_body): Likewise.
385 (build_omp_regions_1): Likewise.
386 * omp-low.c (check_combined_parallel): Skip debug stmts.
387 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
388 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
389 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
391 * tree-cfg.c (make_blobs_1): Skip debug stmts.
392 (make_edges): Likewise.
393 (cleanup_dead_labels): Likewise.
394 (gimple_can_merge_blocks_p): Likewise.
395 (stmt_starts_bb_p): Likewise.
396 (gimple_block_label): Likewise.
397 (gimple_redirect_edge_and_branch): Likewise.
398 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
400 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
402 * tree-pretty-print (print_declaration): Omit initializer in slim
404 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
406 (eliminate_unnecessary_stmts): Stabilize block removal order.
407 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
408 * var-tracking.c (get_first_insn): New.
409 (vt_emit_notes): Call it.
410 (vt_initialize): Walk any insns before the first BB.
411 (delete_debug_insns): Likewise.
413 * gimple.h (enum gimple_debug_subcode): Add
414 GIMPLE_DEBUG_BEGIN_STMT.
415 (gimple_debug_begin_stmt_p): New.
416 (gimple_debug_nonbind_marker_p): New.
417 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
418 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
419 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
420 * insn-notes.def (BEGIN_STMT): New.
421 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
422 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
423 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
424 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
425 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
426 (INSN_DEBUG_MARKER_KIND): New.
427 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
428 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
429 (INSN_VAR_LOCATION_PTR): New.
430 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
432 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
433 deps only if debug bind insns are enabled.
434 (pass_expand::execute): Avoid deep TER and expand
435 debug locations for debug bind insns only.
436 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
437 debug stmts special handling down to debug bind stmts.
438 * combine.c (try_combine): Narrow debug insns special handling
439 down to debug bind insns.
440 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
441 Narrow debug insns preexisting special handling down to debug
443 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
444 handling down to debug bind insns.
445 * function.c (instantiate_virtual_regs): Skip debug markers,
446 adjust handling of debug binds.
447 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
448 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
449 * haifa-sched.c (schedule_insn): Narrow special handling of debug
450 insns to debug bind insns.
451 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
452 special handling of debug stmts to debug bind stmts.
453 * ipa-split.c (split_function): Likewise.
454 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
455 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
457 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
458 * regrename.c (build_def_use): Likewise.
459 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
460 (pass_cprop_hardreg): Narrow special casing of debug insns to
462 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
463 * reload1.c (reload): Likewise.
464 * sese.c (sese_insert_phis_for_liveouts): Narrow special
465 casing of debug stmts to debug bind stmts.
466 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
467 * ssa-iterators.h (num_imm_uses): Likewise.
468 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
469 debug stmts to debug bind stmts.
470 * tree-inline.c (tree_function_versioning): Narrow special casing
471 of debug stmts to debug bind stmts.
472 * tree-loop-distribution.c (generate_loops_for_partition):
473 Narrow special casing of debug stmts to debug bind stmts.
474 * tree-sra.c (analyze_access_subtree): Narrow special casing
475 of debug stmts to debug bind stmts.
476 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
477 stmts to debug bind stmts.
478 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
479 casing of debug stmts to debug bind stmts.
480 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
481 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
482 casing of debug stmts to debug bind stmts.
483 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
485 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
486 debug stmts to debug bind stmts.
487 (gimple_replace_ssa_lhs): Likewise.
488 (insert_debug_temp_for_var_def): Likewise.
489 (insert_debug_temps_for_defs): Likewise.
490 (reset_debug_uses): Likewise.
491 * tree-ssanames.c (release_ssa_name_fn): Likewise.
492 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
493 (adjust_debug_stmts): Likewise.
494 (adjust_phi_and_debug_stmts): Likewise.
495 (vect_do_peeling): Likewise.
496 * tree-vect-loop.c (vect_transform_loop): Likewise.
497 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
498 * var-tracking.c (adjust_mems): Narrow special casing of debug
499 insns to debug bind insns.
500 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
501 (compute_bb_dataflow, vt_find_locations): Likewise.
502 (vt_expand_loc, emit_notes_for_changes): Likewise.
503 (vt_init_cfa_base): Likewise.
504 (vt_emit_notes): Likewise.
505 (vt_initialize): Likewise.
506 (vt_finalize): Likewise.
508 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
509 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
510 (next_nonnote_nondebug_insn_bb): New.
511 (prev_nonnote_nondebug_insn_bb): New.
512 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
513 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
514 (prev_nonnote_nondebug_insn_bb): Declare.
515 (next_nonnote_nondebug_insn_bb): Declare.
516 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
517 * cfgrtl.c (get_last_bb_insn): Likewise.
518 * lra.c (push_insns): Likewise.
520 2017-12-11 David Malcolm <dmalcolm@redhat.com>
523 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
525 * selftest.c (selftest::test_runner::test_runner): New ctor.
526 (selftest::test_runner::~test_runner): New dtor.
527 * selftest.h (class selftest::test_runner): New class.
529 2017-12-11 Carl Love <cel@us.ibm.com>
531 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
532 vec_extract_fp32_from_shortl]): Add #defines.
533 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
534 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
535 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
536 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
537 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
538 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
539 * doc/extend.texi: Add documentation for the added builtins.
541 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
543 PR rtl-optimization/80693
544 PR rtl-optimization/81019
545 PR rtl-optimization/81020
546 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
547 are not mentioned in i3. Place the REG_UNUSED note on i2,
548 possibly modified to REG_DEAD, if it did not originate in i3.
550 2017-12-11 Jakub Jelinek <jakub@redhat.com>
552 * recog.c (store_data_bypass_p_1): New function.
553 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
554 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
556 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
558 PR rtl-optimization/83361
559 * ifcvt.c (if_convert): Call fixup_partitions.
561 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
563 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
564 early folding of splat_u{8,16,32}.
566 2017-12-11 Jakub Jelinek <jakub@redhat.com>
568 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
569 output_operand_lossage first argument with capital letter.
570 (aarch64_override_options): Don't start error and sorry first argument
573 2017-12-11 Andi Kleen <ak@linux.intel.com>
575 PR gcov-profile/83355
576 * auto-profile.c (string_table::get_index_by_decl): Don't
577 recurse when abstract origin points to itself.
579 2017-12-11 Bin Cheng <bin.cheng@arm.com>
581 PR tree-optimization/83320
582 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
583 (prune_datarefs_not_in_loop): Ditto.
585 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
587 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
589 2017-12-10 Jakub Jelinek <jakub@redhat.com>
591 PR tree-optimization/83337
592 * gimple-loop-interchange.cc (compute_access_stride): Handle
593 bitfield DRs properly.
595 2017-12-09 Jakub Jelinek <jakub@redhat.com>
597 PR tree-optimization/83338
598 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
599 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
600 vector of unsigned integers to vector of signed integers.
602 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
604 PR rtl-optimization/83317
605 * lra-constraints.c (process_address_1): Add insn code check.
607 2017-12-08 Michael Matz <matz@suse.de>
609 Fix PR tree-optimization/83323
610 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
611 head-controlled loops and loop BBs.
612 * common.opt (funroll-and-jam): Remove, instead ...
613 (floop-unroll-and-jam): ... reuse this option.
614 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
615 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
616 (-floop-unroll-and-jam): ... this option.
618 2017-12-08 Jakub Jelinek <jakub@redhat.com>
620 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
621 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
622 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
624 2017-12-08 Richard Biener <rguenther@suse.de>
627 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
628 handle accesses outside of zero-sized vars.
630 2017-12-08 Martin Jambor <mjambor@suse.cz>
632 PR tree-optimization/83141
633 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
634 test for MEM_REFs implicitely changing types with padding. Remove
636 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
638 2017-12-08 Jakub Jelinek <jakub@redhat.com>
640 * config/arc/arc.c (arc_attribute_table): Add exclusions to
642 * config/avr/avr.c (avr_attribute_table): Likewise.
643 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
644 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
645 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
646 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
647 of struct attribute_spec.
648 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
650 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
653 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
655 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
658 * profile-count.c (profile_count::from_gcov_type): Move from
659 profile-count.h; handle overflow.
660 * profile-count.h (profile_count::from_gcov_type): Move offline.
662 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
664 PR rtl-optimization/83304
665 * combine.c (move_deaths): If we do not know where a register died,
668 2017-12-08 Richard Biener <rguenther@suse.de>
670 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
671 Provide -fopt-info-loop feedback when we interchange in a nest.
673 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
675 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
676 for armv6 ARM CPU IDs.
678 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
680 * common/config/arm/arm-common.c: Include <algorithm>.
681 (INCLUDE_VECTOR): Define.
682 (compare_opt_names): New function.
683 (arm_rewrite_selected_arch): Only strip out extensions that can be
684 expressed through -mfpu. Sort the remaining extensions
687 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
689 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
690 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
691 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
692 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
694 2017-06-08 Tristan Gingold <gindold@adacore.com>
697 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
698 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
700 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
702 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
703 of struct attribute_spec.
705 2017-12-08 Julia Koval <julia.koval@intel.com>
707 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
708 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
709 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
710 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
711 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
712 _mm_maskz_dpwssds_epi32): Ditto.
714 2017-12-08 Richard Biener <rguenther@suse.de>
716 PR tree-optimization/81303
717 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
718 conditions try to create a comparison vector type matching
719 the data vector type.
720 (vectorizable_condition): Adjust.
721 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
722 Leave invariant conditions alone in case we can vectorize those.
724 2017-12-08 Julia Koval <julia.koval@intel.com>
726 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
727 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
728 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
729 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
730 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
732 2017-12-08 Julia Koval <julia.koval@intel.com>
734 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
735 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
736 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
737 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
738 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
739 _mm_maskz_dpbusds_epi32): New intrinsics.
741 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
743 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
745 (ld<bh>io_signed): Likewise.
746 (st<bhw>io): Likewise.
747 * config/nios2/predicates.md (ldstio_memory_operand): Allow
748 SMALL_INT12 constant integer operand.
750 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
752 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
754 (nios2_large_constant_allowed): ...this. Adjust uses.
755 (nios2_plus_symbolic_constant_p): Rename to...
756 (nios2_plus_large_constant_p): ...this. Adjust uses.
757 (nios2_legitimate_address_p): Correct CONST_INT handling.
758 (nios2_symbolic_memory_operand_p): Rename to...
759 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
760 (nios2_large_constant_p): Check for large constant integers too.
761 (nios2_split_large_constant): Handle constant integers.
762 (nios2_split_symbolic_memory_operand): Rename to...
763 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
764 (nios2_legitimize_constant_address): Handle constant integers.
765 (r0rel_constant_p): Handle small constant integers.
766 (nios2_print_operand_address): Handle r0-relative integer addresses.
767 * config/nios2/nios2-protos.h: Adjust for renamed functions.
768 * config/nios2/nios2.md: Adjust for renamed functions.
770 2017-12-07 Andrew Waterman <andrew@sifive.com>
772 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
773 (TARGET_HAVE_SRODATA_SECTION): New define.
774 (riscv_select_section): New function.
776 2017-12-08 Joseph Myers <joseph@codesourcery.com>
777 Alexander Monakov <amonakov@ispras.ru>
778 Jakub Jelinek <jakub@redhat.com>
781 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
783 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
785 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
786 a bool success value. Don't call output_operand_lossage here.
787 (aarch64_print_ldpstp_address): Return a bool success value.
788 (aarch64_print_operand_address): Call output_addr_const if
789 aarch64_print_address_internal fails.
790 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
791 'y'; call output_operand_lossage instead. Call output_operand_lossage
792 if aarch64_print_ldpstp_address fails.
794 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
796 * tree-vector-builder.h
797 (tree_vector_builder::binary_encoded_nelts): Declare.
798 * tree-vector-builder.c
799 (tree_vector_builder::binary_encoded_nelts): New function.
800 * fold-const.c (negate_expr_p): Likewise.
801 (operand_equal_p, fold_checksum_tree): Likewise.
802 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
803 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
804 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
805 (uniform_vector_p): Likewise.
806 * varasm.c (const_hash_1, compare_constant): Likewise.
807 * tree-ssa-ccp.c: Include tree-vector-builder.h.
808 (valid_lattice_transition): Operate directly on the VECTOR_CST
810 * ipa-icf.c: Include tree-vector-builder.h.
811 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
812 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
814 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
816 * tree.c (build_vector): Delete.
817 * tree.h (build_vector): Make static and move into the self-testing
820 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
822 * vector-builder.h (vector_builder::derived): New const overload.
823 (vector_builder::elt): New function.
824 * tree-vector-builder.h (tree_vector_builder::type): New function.
825 (tree_vector_builder::apply_step): Declare.
826 * tree-vector-builder.c (tree_vector_builder::apply_step): New
828 * gimple-fold.h (tree_vector_builder): Declare.
829 (gimple_build_vector): Take a tree_vector_builder instead of a
830 type and vector of elements.
831 * gimple-fold.c (gimple_build_vector): Likewise.
832 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
834 (get_initial_defs_for_reduction): Likewise.
835 (vectorizable_induction): Likewise.
837 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
839 * tree-vector-builder.h
840 (tree_vector_builder::new_binary_operation): Declare.
841 * tree-vector-builder.c
842 (tree_vector_builder::new_binary_operation): New function.
843 * fold-const.c (fold_relational_const): Use it.
844 (const_binop): Likewise. Check that both input vectors have
845 the same number of elements, thus excluding things like WIDEN_SUM.
846 Check whether it is possible to operate directly on the encodings
849 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
851 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
852 new_unary_operation, operating only on the encoded elements.
853 (const_unop): Likewise.
854 (exact_inverse): Likewise.
855 (distributes_over_addition_p): New function.
856 (const_binop): Use tree_vector_builder and new_unary_operation
857 for combinations of VECTOR_CST and INTEGER_CST. Operate only
858 on the encoded elements unless the encoding is strided and the
859 operation does not distribute over addition.
860 (fold_convert_const): Use tree_vector_builder and
861 new_unary_operation. Operate only on the encoded elements
862 for truncating integer conversions, or for non-stepped encodings.
864 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
866 * config/sparc/sparc.c: Include tree-vector-builder.h.
867 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
868 * expmed.c: Include tree-vector-builder.h.
869 (make_tree): Use tree_vector_builder instead of build_vector.
870 * fold-const.c: Include tree-vector-builder.h.
871 (const_binop): Use tree_vector_builder instead of build_vector.
872 (const_unop): Likewise.
873 (native_interpret_vector): Likewise.
874 (fold_vec_perm): Likewise.
875 (fold_ternary_loc): Likewise.
876 * gimple-fold.c: Include tree-vector-builder.h.
877 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
879 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
880 (simplify_vector_constructor): Use tree_vector_builder instead
882 * tree-vect-generic.c: Include tree-vector-builder.h.
883 (add_rshift): Use tree_vector_builder instead of build_vector.
884 (expand_vector_divmod): Likewise.
885 (optimize_vector_constructor): Likewise.
886 * tree-vect-loop.c: Include tree-vector-builder.h.
887 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
888 of build_vector. Explicitly use a stepped encoding for
890 * tree-vect-slp.c: Include tree-vector-builder.h.
891 (vect_get_constant_vectors): Use tree_vector_builder instead
893 (vect_transform_slp_perm_load): Likewise.
894 (vect_schedule_slp_instance): Likewise.
895 * tree-vect-stmts.c: Include tree-vector-builder.h.
896 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
897 (vect_gen_perm_mask_any): Likewise.
898 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
899 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
901 (build_vector_from_val): Likewise. Explicitly use a duplicate
904 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
906 * doc/generic.texi (VECTOR_CST): Describe new representation of
908 * vector-builder.h: New file.
909 * tree-vector-builder.h: Likewise.
910 * tree-vector-builder.c: Likewise.
911 * Makefile.in (OBJS): Add tree-vector-builder.o.
912 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
913 * tree-core.h (tree_base): Add a vector_cst field to the u union.
914 (tree_vector): Change the number of elements to
915 vector_cst_encoded_nelts.
916 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
917 (VECTOR_CST_ELTS): Delete.
918 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
919 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
920 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
921 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
922 (VECTOR_CST_ENCODED_ELT): Likewise.
923 (vector_cst_encoded_nelts): New function.
924 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
925 VECTOR_CST_NELTS_PER_PATTERN as arguments.
926 (vector_cst_int_elt, vector_cst_elt): Declare.
927 * tree.c: Include tree-vector-builder.h.
928 (tree_code_size): Abort if passed VECTOR_CST.
929 (tree_size): Update for new VECTOR_CST layout.
930 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
931 VECTOR_CST_NELTS_PER_PATTERN as arguments.
932 (build_vector): Use tree_vector_builder.
933 (vector_cst_int_elt, vector_cst_elt): New functions.
934 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
935 encoded elements and then create the vector in the canonical form.
936 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
937 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
938 (tree_c_tests): Call test_vector_cst_patterns.
939 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
941 (hash_tree): Likewise.
942 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
943 (streamer_write_tree_header): Likewise.
944 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
945 (streamer_alloc_tree): Likewise. Update call to make_vector.
946 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
948 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
950 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
951 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
952 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
954 2017-12-07 Bin Cheng <bin.cheng@arm.com>
955 Richard Biener <rguenther@suse.de>
957 PR tree-optimization/81303
958 * Makefile.in (gimple-loop-interchange.o): New object file.
959 * common.opt (floop-interchange): Reuse the option from graphite.
960 * doc/invoke.texi (-floop-interchange): Ditto. New document for
961 -floop-interchange and mention it for -O3.
962 * opts.c (default_options_table): Enable -floop-interchange at -O3.
963 * gimple-loop-interchange.cc: New file.
964 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
965 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
966 * passes.def (pass_linterchange): New pass.
967 * timevar.def (TV_LINTERCHANGE): New time var.
968 * tree-pass.h (make_pass_linterchange): New declaration.
969 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
970 interchange. Record IV before/after increment in new parameters.
971 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
972 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
974 (check_reduction_path): ...New function here.
975 * tree-vectorizer.h (check_reduction_path): New declaration.
977 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
980 PR rtl-optimization/80818
981 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
982 always early clobbered.
983 * lra-lives.c (process_bb_lives): Check input hard regs for early
984 clobbered non-operand hard reg.
986 2017-12-07 Jakub Jelinek <jakub@redhat.com>
989 * tree-cfg.c (verify_gimple_assign_binary): Don't require
990 types_compatible_p, just that TYPE_MODE is the same.
992 2017-12-07 Martin Sebor <msebor@redhat.com>
995 * attribs.c (empty_attribute_table): Initialize new member of
996 struct attribute_spec.
997 (decl_attributes): Add argument. Handle mutually exclusive
998 combinations of attributes.
999 (selftests::test_attribute_exclusions): New function.
1000 (selftests::attribute_c_tests): Ditto.
1001 * attribs.h (decl_attributes): Add default argument.
1002 * selftest.h (attribute_c_tests): Declare.
1003 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1004 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1006 * doc/extend.texi (Common Function Attributes): Update const and pure.
1007 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1008 of struct attribute_spec.
1009 * config/arc/arc.c (arc_attribute_table): Same.
1010 * config/arm/arm.c (arm_attribute_table): Same.
1011 * config/avr/avr.c ( avr_attribute_table): Same.
1012 * config/bfin/bfin.c (bfin_attribute_table): Same.
1013 * config/cr16/cr16.c (cr16_attribute_table): Same.
1014 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1015 * config/h8300/h8300.c (h8300_attribute_table): Same.
1016 * config/i386/i386.c (ix86_attribute_table): Same.
1017 * config/ia64/ia64.c (ia64_attribute_table): Same.
1018 * config/m32c/m32c.c (m32c_attribute_table): Same.
1019 * config/m32r/m32r.c (m32r_attribute_table): Same.
1020 * config/m68k/m68k.c (m68k_attribute_table): Same.
1021 * config/mcore/mcore.c (mcore_attribute_table): Same.
1022 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1023 * config/mips/mips.c (mips_attribute_table): Same.
1024 * config/msp430/msp430.c (msp430_attribute_table): Same.
1025 * config/nds32/nds32.c (nds32_attribute_table): Same.
1026 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1027 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1028 * config/rl78/rl78.c (rl78__attribute_table): Same.
1029 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1030 * onfig/rx/rx.c (rx_attribute_table): Same.
1031 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1032 * config/sh/sh.c (sh_attribute_table): Same.
1033 * config/sparc/sparc.c (sparc_attribute_table): Same.
1034 * config/spu/spu.c (spu_attribute_table): Same.
1035 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1036 * config/v850/v850.c (v850_attribute_table): Same.
1037 * config/visium/visium.c (visium_attribute_table): Same.
1039 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1042 * config/arm/arm.c (INCLUDE_STRING): Define.
1043 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1044 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1047 2017-12-07 Michael Matz <matz@suse.de>
1049 Add unroll and jam pass
1051 * gimple-loop-jam.c: New file.
1052 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1053 * common.opt (funroll-and-jam): New option.
1054 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1055 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1056 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1057 * passes.def: Add pass_loop_jam.
1058 * timevar.def (TV_LOOP_JAM): Add.
1059 * tree-pass.h (make_pass_loop_jam): Declare.
1060 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1061 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1062 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1063 to flow_loop_tree_node_add.
1064 (duplicate_subloops, copy_loops_to): Append to sibling list.
1065 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1066 * doc/invoke.texi (-funroll-and-jam): Document new option.
1067 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1069 2017-12-07 Richard Biener <rguenther@suse.de>
1071 PR tree-optimization/83296
1072 PR tree-optimization/67769
1073 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1074 flow sensitive info in an unrelated BB.
1075 (value_replacement): Use reset_flow_sensitive_info.
1076 (minmax_replacement): Reset flow sensitive info on the def
1077 we move. Do not reset flow sensitive info in the whole BB
1078 we move the stmt to.
1079 (abs_replacement): Likewise.
1081 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1084 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1085 rs6000_cpu to the given -mcpu=, or to the default processor.
1087 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1089 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1090 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1091 instead of rs6000_cpu_attr.
1092 (rs6000_adjust_cost): Ditto.
1093 (is_microcoded_insn): Ditto.
1094 (rs6000_adjust_priority): Ditto.
1095 (rs6000_issue_rate): Ditto.
1096 (rs6000_use_sched_lookahead): Ditto.
1097 (rs6000_use_sched_lookahead_guard): Ditto.
1098 (rs6000_sched_reorder): Ditto.
1099 (force_new_group): Ditto.
1100 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1101 (group_ending_nop): Ditto.
1103 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1105 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1106 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1107 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1108 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1110 (rs6000_reassociation_width): Ditto.
1111 (rs6000_emit_epilogue): Ditto.
1112 (rs6000_adjust_cost): Ditto.
1113 (is_microcoded_insn): Ditto.
1114 (is_cracked_insn): Ditto.
1115 (rs6000_adjust_priority): Ditto.
1116 (rs6000_sched_reorder): Ditto.
1117 (rs6000_sched_reorder2): Ditto.
1118 (insn_must_be_first_in_group): Ditto.
1119 (insn_must_be_last_in_group): Ditto.
1120 (rs6000_register_move_cost): Ditto.
1121 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1124 2017-12-07 Julia Koval <julia.koval@intel.com>
1126 * config.gcc: Add vaesintrin.h.
1127 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1129 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1130 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1132 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1133 * config/i386/immintrin.h: Include vaesintrin.h.
1134 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1135 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1136 _mm_aesdec_epi128): New intrinsics.
1138 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1140 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1141 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1142 c-family/c-spellcheck.cc.
1143 (best_macro_match::best_macro_match): Likewise.
1145 (struct edit_distance_traits<cpp_hashnode *>): Move to
1146 c-family/c-spellcheck.h.
1147 (class best_macro_match): Likewise.
1149 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1151 PR tree-optimization/83293
1152 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1153 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1154 might insert into empty bb.
1157 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1158 simplify for plus with :c added, and pointer_plus without that.
1159 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1160 with undefined overflow and the conversion is not widening,
1161 perform negation in utype and only convert to type afterwards.
1162 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1163 simplify for plus with :c added, and pointer_plus without that.
1164 If type is integral with undefined overflow and the conversion is
1165 not widening, perform minus in utype and only convert to type
1166 afterwards. Move the last pointer_diff_expr simplify into the
1169 2017-12-06 Martin Sebor <msebor@redhat.com>
1171 PR tree-optimization/82646
1172 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1173 strncpy, not maxlen.
1175 2017-12-06 Martin Sebor <msebor@redhat.com>
1177 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1180 PR tree-optimization/83075
1181 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1182 strncat/strncpy don't change length of source string.
1184 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1187 2017-11-29 Martin Aberg <maberg@gaisler.com>
1189 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1190 to prevent b2bst errata sequence.
1191 (sqrtdf2_fix): Likewise.
1193 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1195 PR tree-optimization/81945
1196 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1197 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1198 to dest_cfun has orig_loop_num set, either remap it to the new
1199 loop number if the loop got moved too, or clear it.
1201 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1203 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1205 (thunderx2t99_nothing): New insn reservation.
1206 (thunderx2t99_mrs): New insn reservation.
1207 (thunderx2t99_multiple): New insn reservation.
1208 (thunderx2t99_alu_basi): Add bfx to reservation.
1209 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1211 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1214 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1216 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1218 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1219 tree-ssa-pre.c::remove_dead_inserted_code.
1220 * tree-ssa-dce.h: New file.
1221 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1222 (remove_dead_inserted_code): Move and rename to function
1223 tree-ssa-dce.c::simple_dce_from_worklist.
1224 (pass_pre::execute): Update use.
1226 2017-12-05 Richard Biener <rguenther@suse.de>
1228 PR tree-optimization/83277
1229 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1230 to code-gen liveout vars.
1232 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1234 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1235 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1236 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1239 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1241 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1243 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1244 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1247 2017-12-05 Richard Biener <rguenther@suse.de>
1249 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1250 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1251 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1252 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1254 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1256 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1258 2017-12-05 Martin Liska <mliska@suse.cz>
1259 Jakub Jelinek <jakub@redhat.com>
1261 * doc/invoke.texi: Document the options.
1262 * flag-types.h (enum sanitize_code): Add
1263 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1264 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1265 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1266 * opts.c: Define new sanitizer options.
1267 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1268 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1270 2017-12-05 Julia Koval <julia.koval@intel.com>
1272 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1273 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1274 (ix86_handle_option): Handle -mavx512vnni.
1275 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1276 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1277 * config/i386/i386-c (__AVX512VNNI__): New.
1278 * config/i386/i386.c (ix86_target_string): Handle new option.
1279 (ix86_valid_target_attribute_inner_p): Handle new option.
1280 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1281 * config/i386/i386.opt (mavx512vnni): New option.
1283 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1286 * athlon.md: Disable for generic.
1287 * haswell.md: Enable for generic.
1288 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1289 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1291 (ix86_adjust_cost): Move generic to haswell path.
1293 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1295 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1296 instead of MEM_P in a couple more places. Fix formatting issues.
1298 2017-12-04 Jim Wilson <jimw@sifive.com>
1300 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1301 instead of GP_REG_LAST-1.
1302 (riscv_adjust_libcall_cfi_prologue): Likewise.
1303 (riscv_adjust_libcall_cri_epilogue): Likewise.
1304 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1307 2017-12-04 Luis Machado <luis.machado@linaro.org>
1309 * ipa-pure-const.c (check_decl): Add missing newline.
1310 (state_from_flags): Likewise.
1312 2017-12-04 Jeff Law <law@redhat.com>
1314 PR tree-optimizatin/78496
1315 * gimple-ssa-evrp-analyze.h
1316 (evrp_range_analyzer::get_vr_values): Simplify.
1317 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1318 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1319 and gimple-ssa-evrp-analyze.h.
1320 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1321 (simplify_stmt_for_jump_threading): Copy a blob of code from
1322 tree-vrp.c to use ranges to simplify statements.
1323 (dom_opt_dom_walker::before_dom_children): Call
1324 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1325 (dom_opt_dom_walker::after_dom_children): Similarly for
1326 evrp_range_analyzer::leave.
1327 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1330 * gimple-ssa-evrp-analyze.c
1331 (evrp_range_analyzer::extract_range_from_stmt): Always use
1332 vr_values::update_value_range so preexisting range info is
1333 medged with new range info, even if the new range is VR_VARYING.
1335 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1337 * combine.c: Adjust comment.
1338 (use_crosses_set_p): Delete.
1339 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1340 (try_combine): Ditto.
1342 2017-12-04 Richard Biener <rguenther@suse.de>
1344 PR tree-optimization/83255
1345 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1346 Re-add zero-iteration check.
1348 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1350 PR rtl-optimization/83245
1351 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1352 hard registers as earlyclobber, also if not in an asm.
1354 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1358 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1361 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1362 rs6000_cpu based on cpu_index, not tune_index.
1364 2017-12-04 Richard Biener <rguenther@suse.de>
1366 PR tree-optimization/83238
1367 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1368 code match comment, rejecting invalid SESE regions.
1370 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1372 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1373 require base operand is a REG_POINTER prior to reload on targets
1374 with non-equivalent space registers.
1376 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1378 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1379 (update_specialized_profile): Fix updating of counts.
1380 (perhaps_add_new_callers): Likewise.
1382 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1385 * x86-tune.def: Remove obsolette FIXMEs.
1386 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1387 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1388 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1390 (X86_TUNE_PAD_RETURNS): Disable for generic.
1392 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1394 PR tree-optimization/83170
1395 PR tree-optimization/83241
1396 * gimple-ssa-store-merging.c
1397 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1398 gimple_vuse (ins_stmt) in case it has changed.
1399 (imm_store_chain_info::output_merged_store): Likewise.
1401 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1405 * tree-cfg.c (pass_warn_function_return::execute): Handle
1406 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1407 with BUILTINS_LOCATION.
1411 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1412 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1413 instead of DECL_MODE.
1415 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1416 last argument to rtx pointer.
1417 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1418 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1419 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1420 depending on the chosen ISAs.
1421 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1422 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1423 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1425 * config/i386/sse.md (mov<mode>_internal): Likewise.
1426 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1428 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1430 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1432 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1434 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1435 parameters from prototype.
1436 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1437 parameters. Don't print a comment.
1438 (emit_fusion_gpr_load): Adjust.
1439 (emit_fusion_load_store): Adjust.
1440 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1441 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1442 comment on the second line.
1444 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1447 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1448 rs6000_cpu based on cpu_index, not tune_index.
1450 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1452 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1453 which_alternative instead of which_alternative + 1.
1454 (output_asm_insn): Print an extra tab if the template is short.
1456 2017-12-01 Jim Wilson <jimw@sifive.com>
1458 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1460 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1461 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1463 * doc/tm.texi: Regenerate.
1465 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1468 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1469 whether we can allocate pseudos before trying to fix an address.
1470 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1471 memory address is indexed or indirect.
1472 (floatuns_<mode>si2_hw2): Likewise.
1474 2017-12-01 Jason Merrill <jason@redhat.com>
1476 * Makefile.in (TAGS): Add c-family/*.cc.
1478 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1480 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1481 (call_value_insn): Likewise.
1482 (sibcall_insn): Likewise.
1483 (sibcall_value_insn): Likewise.
1484 (movsi_aarch64): Likewise.
1485 (movdi_aarch64): Likewise.
1486 (add_losym_): Likewise.
1487 (ldr_got_small_): Likewise.
1488 (ldr_got_small_sidi): Likewise.
1489 (ldr_got_small_28k_): Likewise.
1490 (ldr_got_small_28k_sidi): Likewise.
1491 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1492 Move output_addr_const to symbolic case. Add error check.
1494 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1496 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1497 * config/s390/s390.md: Disable MVC merging peephole if it would
1498 disable operand forwarding.
1499 (new peephole2): Split MVCs if it would turn them into up to 2
1502 2017-12-01 Richard Biener <rguenther@suse.de>
1504 PR tree-optimization/83232
1505 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1506 detection of same access. Instead of breaking the group here
1507 do not consider the duplicate. Add comment explaining real fix.
1509 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1511 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1513 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1515 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1517 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1519 * function.h (struct function): Remove cilk_frame_decl,
1520 is_cilk_function and calls_cilk_spawn fields.
1521 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1523 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1524 cilk_elemental field.
1525 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1526 * target.def: Adjust comment.
1527 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1528 Don't test cilk_elemental.
1530 PR tree-optimization/83233
1531 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1532 bswap_stat name for the struct.
1535 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1536 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1537 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1538 SWITCH_BREAK_LABEL_P set on the label.
1539 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1540 added for default case if it was missing and not all cases covered.
1541 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1542 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1544 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1545 to avoid -Wimplicit-fallthrough warning.
1546 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1547 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1549 PR tree-optimization/83221
1550 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1552 (init_reassoc): Formatting fix.
1555 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1556 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1560 * tree-cfg.c: Include asan.h.
1561 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1563 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1565 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1567 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1569 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1570 vpcmpeqd instruction.
1572 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1574 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1575 registers type for 128bit mode.
1577 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1579 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1581 * gengtype-state.c (read_state_pair): Likewise.
1582 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1583 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1584 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1585 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1586 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1587 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1588 * builtins.c (fold_builtin_3): Likewise.
1589 * graphite-scop-detection.c
1590 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1591 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1594 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1597 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1598 parameter. Return true if ignore_decl_rtl_set_p is true and other
1599 conditions are satisfied.
1600 * asan.h (asan_protect_global): Add new parameter.
1601 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1602 to asan_protect_global calls.
1604 2017-11-30 Jim Wilson <jimw@sifive.com>
1606 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1607 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1608 -mstrict-align, add info on default value. Delete redundant lines for
1609 -mabi. Add missing -mexplicit-relocs docs.
1611 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1613 * config/arc/arc.md (trap): New pattern.
1615 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1617 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1618 instruction to end into a delay slot.
1619 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1620 can be placed into a delay slot against reg_note.
1622 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1624 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1625 labels number of usages.
1627 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1629 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1631 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1633 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1636 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1637 multiplication by power of 2 constant into two shifts + comparison.
1639 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1642 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1644 2017-11-30 Richard Biener <rguenther@suse.de>
1646 PR tree-optimization/83202
1647 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1648 allow_peel argument and guard peeling.
1649 (canonicalize_loop_induction_variables): Likewise.
1650 (canonicalize_induction_variables): Pass false.
1651 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1652 peeling from cunrolli.
1654 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1656 * combine.c (try_combine): Print a message to dump file whenever
1657 I0, I1, or I2 cannot be combined into I3.
1659 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1661 PR rtl-optimization/83156
1662 PR rtl-optimization/82621
1663 * combine.c (try_combine): Don't split an I2 if one of the dests is
1664 set again before I3. Allow unused dests.
1666 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1668 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1670 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1672 PR rtl-optimization/80818
1673 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1674 recursively. Use insn code for clobber.
1675 (lra_set_insn_recog_data): Pass the new arg to
1676 collect_non_operand_hard_regs.
1677 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1679 (lra_update_insn_regno_info): Pass insn to
1680 add_regs_to_insn_regno_info.
1682 2017-11-29 Jim Wilson <jimw@sifive.com>
1683 Andrew Waterman <andrew@sifive.com>
1685 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1686 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1687 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1688 use SINGLE_SHIFT_COST.
1689 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1690 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1692 2017-11-29 Julia Koval <julia.koval@intel.com>
1694 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1695 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1696 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1697 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1698 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1699 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1700 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1701 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1702 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1703 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1704 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1705 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1706 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1707 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1708 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1709 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1710 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1711 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1712 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1713 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1714 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1715 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1716 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1717 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1718 __builtin_ia32_vpshldv_v2di_mask,
1719 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1720 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1721 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1723 2017-11-29 Julia Koval <julia.koval@intel.com>
1725 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1726 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1727 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1728 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1729 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1730 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1731 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1732 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1733 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1734 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1735 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1736 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1737 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1738 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1739 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1740 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1741 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1742 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1743 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1744 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1745 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1747 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1749 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1750 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1753 2017-11-29 Martin Aberg <maberg@gaisler.com>
1755 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1756 to prevent b2bst errata sequence.
1757 (sqrtdf2_fix): Likewise.
1759 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1761 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1762 (div_sqrt_insn_p): New function.
1763 (sparc_do_work_around_errata): Insert NOP instructions to
1764 prevent sequences that could trigger the TN-0013 errata for
1765 certain LEON3 processors.
1766 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1767 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1768 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1769 (in_branch_delay): Prevent div and sqrt in delay slot if
1771 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1773 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1775 * config/sparc/sparc.c (atomic_insn_p): New function.
1776 (sparc_do_work_around_errata): Insert NOP instructions to
1777 prevent sequences that could trigger the TN-0010 errata for
1779 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1780 instruction referable in atomic_insns_p.
1782 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1784 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1785 (atomic_compare_and_swap_leon3_1): Likewise.
1788 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1790 * config/sparc/sparc.c (fpop_insn_p): New function.
1791 (sparc_do_work_around_errata): Insert NOP instructions to
1792 prevent sequences that could trigger the TN-0012 errata for
1794 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1795 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1796 (in_branch_annul_delay): Prevent floating-point instructions
1797 in delay slot of annulled integer branch.
1799 2017-11-29 Richard Biener <rguenther@suse.de>
1801 PR tree-optimization/83202
1802 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1804 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1805 nodes vectorized to the same stmts multiple times.
1806 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1807 (vect_analyze_slp_instance): Adjust.
1808 (scalar_stmts_to_slp_tree_map_t): New typedef.
1809 (vect_schedule_slp_instance): Add a map recording the SLP node
1810 representing the vectorized stmts for a set of scalar stmts.
1811 Avoid code-generating redundancies.
1812 (vect_schedule_slp): Allocate map and pass it down.
1814 2017-11-29 Nathan Sidwell <nathan@acm.org>
1817 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1820 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1822 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1823 cases for printing LDP/STP memory addresses.
1824 (aarch64_print_address_internal): Renamed from
1825 aarch64_print_operand_address, added parameter, add Pmode check.
1826 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1827 (aarch64_print_operand_address): Indirect to
1828 aarch64_print_address_internal.
1829 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1830 'y' operand output specifier.
1832 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1835 * tree.c (build_simple_mem_ref_loc): Handle
1836 get_addr_base_and_unit_offset returning a MEM_REF.
1839 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1842 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1845 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1847 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1849 (layout::move_to_column): Likewise.
1851 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1854 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1855 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1857 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1858 Martin Jambor <mjambor@suse.cz>
1861 * tree.h (expr_type_first_operand_type_p): Declare
1862 * tree.c (expr_type_first_operand_type_p): New function.
1863 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1864 (ipa_value_from_jfunc): Adjust declaration.
1865 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1866 Use it as result type for arithmetics, unless it is NULL in which case
1867 be more conservative.
1868 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1869 ipa_get_jf_pass_through_result.
1870 (propagate_vals_across_pass_through): Likewise.
1871 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1872 is to propagate_vals_across_pass_through.
1873 (propagate_constants_across_call): Pass PARM_TYPE to
1874 propagate_scalar_across_jump_function.
1875 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1876 ipa_value_from_jfunc.
1877 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1878 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1879 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1880 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1881 target_type, pass it to ipa_value_from_jfunc.
1882 (update_indirect_edges_after_inlining): Pass parameter type to
1883 try_make_edge_direct_simple_call.
1885 2017-11-28 Jeff Law <law@redhat.com>
1887 * gimple-ssa-evrp-analyze.c
1888 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1889 refine ranges if scev_initialized_p returns true.
1890 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1892 2017-11-28 Julia Koval <julia.koval@intel.com>
1894 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1895 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1896 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1897 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1898 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1899 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1900 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1901 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1902 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1903 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1904 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1905 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1906 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1907 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1908 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1909 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1910 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1911 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1912 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1913 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1914 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1915 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1917 2017-11-28 Julia Koval <julia.koval@intel.com>
1919 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1920 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1921 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1922 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1923 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1924 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1925 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1926 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1927 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1928 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1929 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1930 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1931 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1932 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1933 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1934 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1935 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1936 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1937 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1938 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1939 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1940 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1941 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1942 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1943 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1944 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1945 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1946 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1947 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1948 __builtin_ia32_vpshld_v2di_mask): New builtins.
1949 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1950 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1952 2017-11-28 Richard Biener <rguenther@suse.de>
1954 PR tree-optimization/80776
1955 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1957 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1959 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1960 If the incoming edge is an effective fallthru because the other
1961 edge only reaches a __builtin_unreachable () then record ranges
1962 derived from the controlling condition in SSA info.
1963 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1964 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1966 2017-11-28 Olivier Hainque <hainque@adacore.com>
1968 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1969 on mingw build hosts.
1971 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1974 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1976 2017-11-28 Richard Biener <rguenther@suse.de>
1979 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1980 copies generated from memcpy use a character array as reference
1983 2017-11-28 Julia Koval <julia.koval@intel.com>
1984 Sebastian Peryt <sebastian.peryt@intel.com>
1986 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1987 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1988 cilk-common.o, cilk.h, cilk-common.c): Remove.
1990 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1991 * builtins.c (is_builtin_name): Remove cilkplus condition.
1992 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1993 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1994 cilk-builtins.def, cilkplus.def): Remove.
1995 * cif-code.def (CILK_SPAWN): Remove.
1996 * cilk-builtins.def: Delete.
1997 * cilk-common.c: Ditto.
1999 * cilkplus.def: Ditto.
2000 * config/darwin.h (fcilkplus): Delete.
2001 * cppbuiltin.c: Ditto.
2002 * doc/extend.texi: Remove cilkplus doc.
2003 * doc/generic.texi: Ditto.
2004 * doc/invoke.texi: Ditto.
2005 * doc/passes.texi: Ditto.
2006 * gcc.c (fcilkplus): Remove.
2007 * gengtype.c (cilk.h): Remove.
2008 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2010 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2012 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2014 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2015 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2016 cilkplus conditions.
2017 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2018 inline_read_section): Ditto.
2019 * ipa-inline-analysis.c (cilk.h): Remove.
2020 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2021 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2022 append_linker_options): Remove condition for fcilkplus.
2023 * lto/lto-lang.c (cilk.h): Remove.
2024 (lto_init): Remove condition for fcilkplus.
2025 * omp-expand.c (expand_cilk_for_call): Delete.
2026 (expand_omp_taskreg, expand_omp_for_static_chunk,
2027 expand_omp_for): Remove cilkplus
2029 (expand_cilk_for): Delete.
2030 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2031 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2032 execute_lower_omp, diagnose_sb_0): Ditto.
2033 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2034 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2035 * tree-nested.c: Ditto.
2036 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2037 (dump_generic_node): Ditto.
2038 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2039 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2041 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2043 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2045 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2046 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2047 for vector mode and !TARGET_SIMD.
2049 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2051 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2053 * tree.h (SWITCH_LABELS): Remove.
2054 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2055 assert SWITCH_BODY is non-NULL.
2056 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2058 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2060 PR tree-optimization/80788
2061 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2062 has TREE_OVERFLOW set, call drop_tree_overflow.
2064 2017-11-28 Richard Biener <rguenther@suse.de>
2066 PR tree-optimization/83158
2067 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2069 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2072 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2073 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2075 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2077 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2078 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2080 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2083 * varasm.c (compare_constant): Take the mode of the constants into
2084 account when comparing floating point constants.
2086 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2088 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2089 from explicit instantiation of debug_helper.
2090 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2092 2017-11-27 Richard Biener <rguenther@suse.de>
2094 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2097 2017-11-27 Richard Biener <rguenther@suse.de>
2099 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2100 (type_hash_canon): Also clear int_cst_hash_table entry for
2102 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2104 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2106 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2107 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2108 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2110 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2112 * hash-map.h (gt_cleare_cache): Avoid UB.
2114 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2116 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2117 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2118 (decide_unroll_runtime_iterations): Likewise.
2119 (decide_unroll_stupid): Likewise.
2121 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2124 * config/i386/i386.md: Add a loop with incssp.
2126 2017-11-27 Martin Jambor <mjambor@suse.cz>
2128 PR tree-optimization/81248
2129 * tree-sra.c (splice_param_accesses): Remove size check.
2130 (decide_one_param_reduction): Fix size check.
2131 * gimple-pretty-print.c (dump_profile): Silence warning.
2132 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2134 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2137 * dbxout.c (lastlineno): New variable.
2138 (dbx_debug_hooks): Use dbxout_switch_text_section as
2139 switch_text_section debug hook.
2140 (dbxout_function_end): Switch to current_function_section
2141 rather than function_section. If crtl->has_bb_partition,
2142 output just one N_FUN, depending on in_cold_section_p.
2143 (dbxout_source_line): Remember last lineno in lastlineno.
2144 (dbxout_switch_text_section): New function.
2145 (dbxout_function_decl): Adjust dbxout_block caller.
2146 (dbx_block_with_cold_children): New function.
2147 (dbxout_block): Return true if any LBRAC/RBRAC have been
2148 emitted. Use dbx_block_with_cold_children at depth == 0
2149 in second partition. Add PARENT_BLOCKNUM argument, pass
2150 it optionally adjusted to children. Output LBRAC/RBRAC
2151 around recursive call only if the block is in the current
2152 partition, if not and anything was output, emit empty
2154 * final.c (final_scan_insn): Compute cold_function_name
2155 before calling switch_text_section debug hook. Call
2156 that hook even if dwarf2out_do_frame if not emitting
2160 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2161 TREE_READONLY decls.
2163 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2165 PR rtl-optimization/82488
2166 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2168 2017-11-26 Julia Koval <julia.koval@intel.com>
2170 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2172 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2175 2017-11-26 Julia Koval <julia.koval@intel.com>
2177 * config/i386/driver-i386.c (host_detect_local_cpu):
2178 Detect skylake-avx512.
2180 2017-11-26 Julia Koval <julia.koval@intel.com>
2182 * config.gcc: Add -march=cannonlake.
2183 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2184 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2185 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2186 (PTA_CANNONLAKE): New.
2187 (processor_target_table): Add cannonlake.
2188 (ix86_option_override_internal): Ditto.
2189 (fold_builtin_cpu): Ditto.
2190 (get_builtin_code_for_version): Handle cannonlake.
2191 (M_INTEL_COREI7_CANNONLAKE): New.
2192 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2193 * doc/invoke.texi: Add -march=cannonlake.
2195 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2197 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2198 (try_init_one_plugin): Alternative implementation for MinGW.
2199 * Makefile.in (plugin_implib): New.
2200 (gengtype-lex.c): Fix broken AIX workaround.
2201 * configure: Regenerate.
2202 * doc/plugins.texi: Document support for MinGW.
2204 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2206 PR rtl-optimization/81553
2207 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2208 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2209 is a shift where C1 has different mode than the whole shift, use C1's
2210 mode for MULT rather than the shift's mode.
2213 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2214 builtins not enabled in the currently selected ISA.
2216 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2218 PR tree-optimization/71026
2219 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2220 (insert_reciprocals): Change to insert reciprocals before a division
2221 by a square and to insert the square of a reciprocal.
2222 (execute_cse_reciprocals_1): Change to consider division by a square.
2223 (register_division_in): Add importance parameter.
2225 2017-11-24 Richard Biener <rguenther@suse.de>
2227 PR tree-optimization/82402
2228 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2229 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2231 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2233 * match.pd (0-ptr): New transformation.
2235 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2238 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2239 when skipping checking.
2241 2017-11-24 Richard Biener <rguenther@suse.de>
2243 PR tree-optimization/83128
2244 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2245 (vn_reference_lookup_3): Likewise.
2247 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2250 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2251 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2254 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2255 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2256 check for the latter with gimple_call_builtin_p. Do not handle
2257 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2259 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2261 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2264 2017-11-23 Julia Koval <julia.koval@intel.com>
2266 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2267 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2268 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2269 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2270 _mm512_maskz_expandloadu_epi16): New intrinsics.
2271 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2272 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2273 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2274 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2275 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2276 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2277 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2278 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2279 _mm256_maskz_expandloadu_epi8): New intrinsics.
2280 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2281 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2282 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2283 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2284 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2285 * config/i386/sse.md (VI248_VLBW): New iterator.
2286 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2288 2017-11-23 Julia Koval <julia.koval@intel.com>
2290 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2291 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2292 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2293 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2294 _mm512_mask_compressstoreu_epi16): New.
2295 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2296 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2297 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2298 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2299 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2300 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2301 _mm256_mask_compressstoreu_epi8): New.
2302 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2303 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2304 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2305 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2306 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2307 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2308 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2309 __builtin_ia32_compresshi128_mask,
2310 __builtin_ia32_compressstoreuqi512_mask,
2311 __builtin_ia32_compressstoreuhi512_mask,
2312 __builtin_ia32_compressstoreuqi256_mask,
2313 __builtin_ia32_compressstoreuqi128_mask,
2314 __builtin_ia32_compressstoreuhi256_mask,
2315 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2316 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2318 (ix86_expand_special_args_builtin): Handle new types.
2319 (s4fma_expand): Handle new builtin array.
2320 * config/i386/immintrin.h: Include new headers.
2321 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2322 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2324 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2326 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2327 fixes. Declare temp and g variables at the top in order to avoid
2328 {} in most of the cases.
2330 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2332 * match.pd (ptr-0): New transformation.
2334 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2336 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2337 (struct addr_mode_cost_table): New.
2338 (struct tune_params): Add field addr_mode_costs.
2339 * config/arm/arm.c (generic_addr_mode_costs): New.
2340 (arm_slowmul_tune): Initialise addr_mode_costs field.
2341 (arm_fastmul_tune): Likewise.
2342 (arm_strongarm_tune): Likewise.
2343 (arm_xscale_tune): Likewise.
2344 (arm_9e_tune): Likewise.
2345 (arm_marvell_pj4_tune): Likewise.
2346 (arm_v6t2_tune): Likewise.
2347 (arm_cortex_tune): Likewise.
2348 (arm_cortex_a8_tune): Likewise.
2349 (arm_cortex_a7_tune): Likewise.
2350 (arm_cortex_a15_tune): Likewise.
2351 (arm_cortex_a35_tune): Likewise.
2352 (arm_cortex_a53_tune): Likewise.
2353 (arm_cortex_a57_tune): Likewise.
2354 (arm_exynosm1_tune): Likewise.
2355 (arm_xgene1_tune): Likewise.
2356 (arm_cortex_a5_tune): Likewise.
2357 (arm_cortex_a9_tune): Likewise.
2358 (arm_cortex_a12_tune): Likewise.
2359 (arm_cortex_a73_tune): Likewise.
2360 (arm_v7m_tune): Likewise.
2361 (arm_cortex_m7_tune): Likewise.
2362 (arm_v6m_tune): Likewise.
2363 (arm_fa726te_tune): Likewise.
2364 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2366 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2368 * config/arm/arm.c (arm_mem_costs): New function.
2369 (arm_rtx_costs_internal): Use arm_mem_costs.
2371 2017-11-23 Mark Wielaard <mark@klomp.org>
2373 * dwarf2out.c (init_sections_and_labels): Use generation to create
2374 unique ranges_section_label and ranges_base_label. Return generation.
2375 (output_rnglists): Add generation argument. Use generation to create
2376 unique ranges labels.
2377 (dwarf2out_finish): Get generation from init_sections_and_labels
2378 and pass generation to output_rnglists.
2380 2017-11-23 Mike Stump <mikestump@comcast.net>
2381 Eric Botcazou <ebotcazou@adacore.com>
2383 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2384 * cfgloop.h (struct loop): Add unroll field.
2385 * function.h (struct function): Add has_unroll bitfield.
2386 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2387 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2388 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2389 (pass_rtl_unroll_loops::gate): Likewise.
2390 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2391 for which loop->unroll==1.
2392 (decide_unroll_constant_iterations): Use note for consistency and
2393 take loop->unroll into account. Return early if loop->unroll is set.
2394 Fix thinko in existing test.
2395 (decide_unroll_runtime_iterations): Use note for consistency and
2396 take loop->unroll into account.
2397 (decide_unroll_stupid): Likewise.
2398 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2399 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2400 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2402 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2403 (print_loop): Print loop->unroll if set.
2404 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2405 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2406 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2408 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2409 loop->unroll is set and smaller than the trip count. Otherwise bypass
2410 entirely the heuristics if loop->unroll is set. Remove dead note.
2411 Fix off-by-one bug in other note.
2412 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2413 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2415 (tree_unroll_loops_completely): Make static.
2416 (pass_complete_unroll::execute): Use correct type for variable.
2417 (pass_complete_unrolli::execute): Fix formatting.
2418 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2420 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2422 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2423 enable when TARGET_PREFER_AVX128 is set.
2425 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2427 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2429 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2430 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2432 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2434 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2435 frequencies. Fix estimation of aggregate parameters.
2437 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2439 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2442 2017-11-23 Tom de Vries <tom@codesourcery.com>
2444 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2446 2017-11-23 Tom de Vries <tom@codesourcery.com>
2448 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2451 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2454 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2455 sibcall_value_pcrel_fdpic): Use local variable instead of
2457 (calli_tbr_rel): Add missing operand 2.
2458 (call_valuei_tbr_rel): Add missing operand 3.
2460 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2463 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2464 bitpos/bitsize covering the whole destination, use store_expr only if
2465 the complex mode is the same. Otherwise, use expand_normal and if
2466 it returns CONCAT, subreg each part separately instead of trying to
2467 subreg the whole result.
2469 2017-11-23 Richard Biener <rguenther@suse.de>
2471 PR tree-optimization/23094
2472 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2473 come here from walking over backedges in the first iteration.
2474 (vn_reference_lookup_3): Skip clobbers that store the same value.
2476 2017-11-23 Richard Biener <rguenther@suse.de>
2478 PR tree-optimization/81403
2479 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2480 a block we need a leader relative to.
2481 (phi_translate_1): For nary processing require a leader from
2482 get_representative_for given we run expression simplification
2483 using match-and-simplify. Remove previous fix.
2485 2017-11-22 Jeff Law <law@redhat.com>
2487 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2488 Use new method allocate_value_range rather than accessing the
2489 vrp_value_range_pool data member directly.
2490 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2491 to use extract_range_from_stmt method to avoid need for
2492 extract_range_from_assignment method.
2493 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2494 method rather than setting values_propgated data member directly.
2495 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2496 and values propagated data members and extract_range_from_assignment
2497 method. Reorder private data members to conform to standards.
2498 Add new methods set_lattice_propagation_complete and
2499 allocate_value_range.
2501 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2503 PR rtl-optimization/83030
2504 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2505 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2506 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2507 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2508 CROSSING_JUMP_P flag is already set before setting it.
2509 * cfgrtl.c (fixup_partition_crossing): Likewise.
2510 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2513 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2515 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2516 Handle the case where both arguments are using gen_const_vec_series.
2518 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2521 * pretty-print.c (pp_format): Move quoting implementation to
2522 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2523 to pass address of "quote" local.
2524 (pp_begin_quote): New function.
2525 (pp_end_quote): New function.
2526 * pretty-print.h (printer_fn): Convert penultimate param from bool
2528 (pp_begin_quote): New decl.
2529 (pp_end_quote): New decl.
2530 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2531 param from bool to bool *.
2532 * tree-diagnostic.h (default_tree_printer): Likewise.
2534 2017-11-22 Jeff Law <law@redhat.com>
2536 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2537 Initialize vr_values.
2538 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2539 vr_values via vr_values class instance rather than delegators.
2540 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2541 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2542 (evrp_range_analyzer::push_value_range): Likewise.
2543 (evrp_range_analyzer::pop_value_range): Likewise.
2544 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2545 most delegators. Those remaining are exposed as public interfaces.
2546 Make vr_values a pointer and private.
2547 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2549 (evrp_range_analyzer::get_vr_value): New method.
2550 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2551 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2552 (class evrp_dom_walker): Attach evrp_folder class, initialize
2553 it in the ctor. Remove temporary delegators.
2554 (evrp_dom_walker::before_dom_children): Call methods in attached
2555 evrp_range_analyzer class via class instance pointer. Use
2556 free value_range_constant_singleton to remove need for
2557 op_with_constant_singleton_value delegator method. Do not
2558 create a vrp_prop class instance for every call! Narrow
2559 scope of a couple variables.
2560 (evrp_dom_walker::cleanup): Call methods in attached
2561 evrp_range_analyzer class via class instance pointer.
2562 * vr-values.h (class vr_values): Privatize many methods and
2565 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2567 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2568 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2569 * expr.c (expand_expr_real_2): Likewise.
2570 * fold-const.c (const_unop): Likewise.
2571 * optabs-tree.c (optab_for_tree_code): Likewise.
2572 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2573 * tree-inline.c (estimate_operator_cost): Likewise.
2574 * tree-pretty-print.c (dump_generic_node): Likewise.
2575 (op_code_prio): Likewise.
2576 (op_symbol_code): Likewise.
2577 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2578 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2579 * internal-fn.c (direct_internal_fn_optab): New function.
2580 (direct_internal_fn_array, direct_internal_fn_supported_p
2581 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2582 * fold-const-call.c (fold_const_reduction): New function.
2583 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2585 * tree-vect-loop.c: Include internal-fn.h.
2586 (reduction_code_for_scalar_code): Rename to...
2587 (reduction_fn_for_scalar_code): ...this and return an internal
2589 (vect_model_reduction_cost): Take an internal_fn rather than
2591 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2593 (vectorizable_reduction): Use internal functions rather than tree
2594 codes for the reduction operation. Update calls to the functions
2596 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2597 Use calls to internal functions rather than REDUC tree codes.
2598 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2600 2017-11-22 Olivier Hainque <hainque@adacore.com>
2602 * config/vxworks.c (vxworks_override_options): Pick default
2603 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2604 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2605 DWARF_GNAT_ENCODINGS_DEFAULT.
2606 * config/vxworksae.h: Likewise.
2608 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2610 PR tree-optimization/83104
2611 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2614 2017-11-22 Marek Polacek <polacek@redhat.com>
2615 H.J. Lu <hongjiu.lu@intel.com>
2616 Jason Merrill <jason@redhat.com>
2621 * calls.c (initialize_argument_information): Call
2622 warn_parameter_passing_abi target hook.
2623 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2624 argument onto stack.
2625 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2626 * common.opt: Update -fabi-version description.
2627 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2628 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2630 (ix86_is_empty_record): New function.
2631 (ix86_warn_parameter_passing_abi): New function.
2632 (TARGET_EMPTY_RECORD_P): Redefine.
2633 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2634 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2635 * doc/tm.texi: Regenerated.
2636 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2637 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2638 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2639 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2640 instead of int_size_in_bytes.
2641 * expr.c (copy_blkmode_to_reg): Likewise.
2642 * function.c (aggregate_value_p): Return 0 for empty types.
2643 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2645 (locate_and_pad_parm): Call arg size_in_bytes instead
2647 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2648 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2649 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2651 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2652 (std_gimplify_va_arg_expr): Skip empty records. Call
2653 arg_size_in_bytes instead size_in_bytes.
2654 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2655 * tree-core.h (tree_type_common): Add empty_flag.
2656 (tree_decl_common): Update comments.
2657 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2659 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2660 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2662 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2663 * tree.c (default_is_empty_type): New function.
2664 (default_is_empty_record): New function.
2665 (arg_int_size_in_bytes): New function.
2666 (arg_size_in_bytes): New function.
2667 (get_ultimate_context): New function.
2668 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2669 TRANSLATION_UNIT_WARN_EMPTY_P.
2670 (default_is_empty_record, arg_int_size_in_bytes,
2671 arg_size_in_bytes, get_ultimate_context): Declare.
2673 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2675 * config/arm/arm.c (cmse_clear_registers): New function.
2676 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2677 code by call to cmse_clear_registers.
2678 (cmse_nonsecure_entry_clear_before_return): Likewise.
2680 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2682 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2683 (vdot_s32, vdotq_s32): New.
2684 (vdot_lane_u32, vdotq_lane_u32): New.
2685 (vdot_lane_s32, vdotq_lane_s32): New.
2688 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2691 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2692 for unsigned values with fewer HWIs than the precision.
2693 (test_overflow): New function.
2694 (wide_int_cc_tests): Call it.
2696 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2698 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2699 has the same mode class as Pmode.
2701 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2703 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2704 padding_bits_to_clear_ptr.
2705 (cmse_nonsecure_entry_clear_before_return): Likewise.
2707 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2709 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2710 auto_sbitap instead of integer bitfield to control register needing
2713 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2715 PR tree-optimization/83044
2716 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2717 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2718 that need the upper bound. Subtract offset from
2719 get_addr_base_and_unit_offset only if positive and subtract it
2720 before division by eltsize rather than after it.
2723 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2724 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2725 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2729 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2730 expand_binop with *mul_widen_optab, make sure at least one of the
2731 operands doesn't have VOIDmode.
2734 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2736 PR rtl-optimization/82044
2737 PR tree-optimization/82042
2738 * dse.c (record_store): Check for overflow.
2739 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2740 add_wild_read instead of clear_rhs_from_active_local_stores on
2741 overflow and log it into dump_file.
2743 2017-11-22 Richard Biener <rguenther@suse.de>
2745 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2746 fake edges to exit when looking for a place to insert.
2747 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2749 (insert_into_preds_of_block): Commit edge insertion immediately,
2750 assert that doesn't require new BBs.
2751 (fini_pre): Release expressions.
2752 (pass_pre::execute): Shuffle things around a bit, if the fn
2753 is too large do not compute AVAIL either as this is really the
2756 2017-11-22 Richard Biener <rguenther@suse.de>
2758 PR tree-optimization/83089
2759 * tree-if-conv.c (pass_if_conversion::execute): If anything
2760 changed reset SCEV and free the number of iteration estimates.
2762 2017-11-21 Martin Sebor <msebor@redhat.com>
2764 PR tree-optimization/82945
2765 * calls.h (warn_nonstring_bound): Remove unused function.
2767 2017-11-21 Martin Sebor <msebor@redhat.com>
2769 PR tree-optimization/82945
2770 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2771 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2772 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2774 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2775 * calls.h (get_attr_nonstring_decl): Declare new function.
2776 * doc/extend.texi (attribute nonstring): Update.
2777 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2778 get_attr_nonstring_decl and handle it.
2779 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2780 detection of nul-termination.
2781 (strlen_to_stridx): Change to a pointer.
2782 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2783 (pass_strlen::execute): Same.
2785 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2787 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2788 for the new option -mprefer-vector-width=[none|128|256|512].
2789 * config/i386/i386.c (ix86_target_string): remove old style options
2790 -mprefer-avx256 and make -mprefer-avx128 as alias.
2791 (ix86_option_override_internal): Apply defaults for the
2792 -mprefer-vector-width=[128|256] option.
2793 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2794 Implement macros to work with -mprefer-vector-width=.
2795 * config/i386/i386.opt: Implemented option
2796 -mprefer-vector-width=[none|128|256|512].
2797 * doc/invoke.texi: Documentation for
2798 -mprefer-vector-width=[none|128|256|512].
2800 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2802 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2804 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2806 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2808 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2810 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2812 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2813 * expr.c (expand_expr_real_2): Likewise.
2814 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2815 fold_binary_loc): Likewise.
2816 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2817 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2818 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2819 MINUS_EXPR transformations.
2820 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2821 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2822 * tree-inline.c (estimate_operator_cost): Likewise.
2823 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2824 op_symbol_code): Likewise.
2825 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2826 * vr-values.c (extract_range_from_binary_expr): Likewise.
2827 * varasm.c (initializer_constant_valid_p_1): Likewise.
2828 * tree.def: New tree code POINTER_DIFF_EXPR.
2830 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2832 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2833 integer suffix to movbe mnemonic.
2834 (*bswaphi2_movbe): Ditto.
2835 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2837 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2840 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2841 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2842 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2844 2017-11-21 Martin Liska <mliska@suse.cz>
2846 * tree-inline.c (expand_call_inline): Remove not needed
2849 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2850 Jakub Jelinek <jakub@redhat.com>
2853 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2854 Remove static keyword from f variable.
2856 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2858 PR tree-optimization/83086
2859 * gimple-ssa-store-merging.c
2860 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2861 rather than n.base_addr.
2863 2017-11-21 Martin Liska <mliska@suse.cz>
2865 PR rtl-optimization/82044
2866 PR tree-optimization/82042
2867 * dse.c (check_mem_read_rtx): Check for overflow.
2869 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2871 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2874 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2876 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2877 and its corresponding call arg location note.
2879 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2880 Andrew Burgess <andrew.burgess@embecosm.com>
2882 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2884 (arc_return_slot_offset): Likewise.
2885 (arc_eh_return_address_location): New declaration.
2886 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2887 (MUST_SAVE_REGISTER): Add exception handler case.
2888 (MUST_SAVE_RETURN_ADDR): Likewise.
2889 (arc_frame_pointer_required): Likewise.
2890 (arc_frame_pointer_needed): New function.
2891 (arc_compute_frame_size): Changed.
2892 (arc_expand_prologue): Likewise.
2893 (arc_expand_epilogue): Likewise.
2894 (arc_initial_elimination_offset): Likewise.
2895 (arc_return_slot_offset): Delete.
2896 (arc_eh_return_address_location): New function.
2897 (arc_builtin_setjmp_frame_value): Likewise.
2898 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2899 (EH_RETURN_STACKADJ_RTX): Define.
2900 (EH_RETURN_HANDLER_RTX): Likewise.
2901 * config/arc/arc.md (eh_return): Delete.
2903 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2905 * print-tree.h (debug_vec_tree): Remove prototype.
2906 * gdbinit.in (pvt): Remove macro.
2908 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2910 PR tree-optimization/83047
2911 * gimple-ssa-store-merging.c
2912 (imm_store_chain_info::output_merged_store): If the loads with the
2913 same vuse are in different basic blocks, for load_gsi pick a load
2914 location that is dominated by the other loads.
2917 * config/i386/i386.c (ix86_memmodel_check): Start
2918 -Winvalid-memory-model diagnostics with lowercase letter.
2921 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2922 set in_cold_section_p to the partition containing loc_list->first.
2923 When seeing loc_list->last_before_switch node, update secname and
2924 perform range_across_switch second partition handling only after that.
2927 * run-rtl-passes.c: Include debug.h.
2928 (run_rtl_passes): Call debug_hooks->assembly_start.
2929 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2933 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2934 OPTAB_DIRECT in calls to expand_simple_binop.
2936 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2939 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2941 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2944 * spellcheck.h (best_match::blithely_get_best_candidate): New
2947 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2949 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2950 with lower case letter.
2952 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2954 * config/i386/i386.md (bswaphi2): New expander.
2955 (*bswaphi2_movbe): New insn pattern.
2956 (bswaphi -> rorhi pepehole2): New peephole pattern.
2958 2017-11-20 Jeff Law <law@redhat.com>
2960 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2961 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2962 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2963 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2964 gimple-ssa-evrp-analyze.h.
2966 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2967 set BB_VISITED here.
2968 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2970 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2972 * tree-predcom.c: Add general comment on Store-Store chains.
2973 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2975 (get_chain_last_ref_at): Rename into...
2976 (get_chain_last_write_at): ...this.
2977 (get_chain_last_write_before_load): New function.
2978 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2979 CT_STORE_STORE when write reference is added.
2980 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2981 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2982 (initialize_root_vars_store_elim_1): Ditto.
2983 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2984 definition is created.
2985 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2986 chain by replacing it with dominant stored value.
2988 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2990 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2992 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2994 * vr-values.c (extract_range_from_binary_expr): Use a full range
2997 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2999 * config/arm/arm.md (R4_REGNUM): Define constant.
3000 (nonsecure_call_internal): Remove r4 clobber.
3001 (nonsecure_call_value_internal): Likewise.
3002 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3003 clobber and resequence match_operands.
3004 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3005 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3006 (nonsecure_call_value_reg_thumb2): Likewise.
3008 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3010 PR tree-optimization/78821
3011 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3012 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3013 to the address of the base rather than the base itself.
3014 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3015 (find_bswap_or_nop_finalize): New function.
3016 (find_bswap_or_nop): Use it.
3017 (bswap_replace): Return a tree rather than bool, change first
3018 argument from gimple * to gimple_stmt_iterator, allow inserting
3019 into an empty sequence, allow ins_stmt to be NULL - then emit
3020 all stmts into gsi. Fix up MEM_REF address gimplification.
3021 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3022 (struct store_immediate_info): Add N and INS_STMT non-static
3024 (store_immediate_info::store_immediate_info): Initialize them
3025 from newly added ctor args.
3026 (merged_store_group::apply_stores): Formatting fixes. Sort by
3028 (stmts_may_clobber_ref_p): For stores call also
3029 refs_anti_dependent_p.
3030 (gather_bswap_load_refs): New function.
3031 (imm_store_chain_info::try_coalesce_bswap): New method.
3032 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3033 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3034 (imm_store_chain_info::output_merged_store): Fail if number of
3035 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3036 and NOP_EXPR rhs_code.
3037 (pass_store_merging::process_store): Compute n and ins_stmt, if
3038 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3039 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3041 (pass_store_merging::execute): Calculate dominators.
3043 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3044 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3045 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3046 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3047 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3048 class pass_optimize_bswap, bswap_replace,
3049 pass_optimize_bswap::execute): Moved to ...
3050 * gimple-ssa-store-merging.c: ... this file.
3051 Include optabs-tree.h.
3052 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3053 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3054 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3055 anonymous namespace, remove static keywords.
3056 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3057 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3059 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3062 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3064 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3066 * vec.h (debug_helper): New function.
3067 (DEFINE_DEBUG_VEC): New macro.
3068 * hash-set.h (debug_helper): New function.
3069 (DEFINE_DEBUG_HASH_SET): New macro.
3070 * cfg.c (debug_slim (edge)): New function.
3071 Call DEFINE_DEBUG_VEC for edges.
3072 Call DEFINE_DEBUG_HASH_SET for edges.
3073 * cfghooks.c (debug_slim (basic_block)): New function.
3074 Call DEFINE_DEBUG_VEC for basic blocks.
3075 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3076 * print-tree.c (debug_slim): New function to handle trees.
3077 Call DEFINE_DEBUG_VEC for trees.
3078 Call DEFINE_DEBUG_HASH_SET for trees.
3079 (debug (vec<tree, va_gc>) &): Remove.
3080 (debug (<vec<tree, va_gc>) *): Remove.
3081 * print-rtl.c (debug_slim): New function to handle const_rtx.
3082 Call DEFINE_DEBUG_VEC for rtx_def.
3083 Call DEFINE_DEBUG_VEC for rtx_insn.
3084 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3085 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3086 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3087 (debug (vec<rtx_insn *> *ptr): Remove.
3088 (debug_insn_vector): Remove.
3089 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3091 2017-11-20 Tom de Vries <tom@codesourcery.com>
3093 PR rtl-optimization/82020
3094 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3095 IF_THEN_ELSE condition.
3097 2017-11-19 Jeff Law <law@redhat.com>
3099 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3100 of degenerates resulting from ignoring an edge.
3102 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3105 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3107 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3110 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3111 for uninitialied counts.
3113 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3116 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3118 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3121 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3123 2017-11-19 Tom de Vries <tom@codesourcery.com>
3125 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3128 2017-11-19 Tom de Vries <tom@codesourcery.com>
3130 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3133 2017-11-19 Tom de Vries <tom@codesourcery.com>
3135 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3138 2017-11-19 Tom de Vries <tom@codesourcery.com>
3140 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3142 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3144 2017-11-19 Tom de Vries <tom@codesourcery.com>
3146 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3148 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3149 semicolon after MCORE_EXPORT_NAME call.
3151 2017-11-19 Tom de Vries <tom@codesourcery.com>
3154 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3155 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3157 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3159 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3160 (cgraph_edge::clone): Cleanup updating of profile.
3161 * ipa-cp.c (update_profiling_info): Likewise.
3162 * ipa-inline-transform.c (inline_transform): Likewise.
3163 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3164 * ipa-split.c (execute_split_functions): Do not split when function
3166 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3167 * profile-count.c (profile_count::dump): Add global0.
3168 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3170 (profile_count::to_sreal_scale): Likewise.
3171 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3172 (profile_count::combine_with_ipa_count): New function.
3173 * profile-count.h (profile_guessed_global0adjusted): New.
3174 (profile_count::adjusted_zero): New.
3175 (profile_count::global0adjusted): New.
3176 (profile_count::combine_with_ipa_count): New.
3177 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3178 correct profile of return block of split functions.
3179 (copy_cfg_body): Remove unused profile_count.
3180 (copy_body): Likewise.
3181 (expand_call_inline): Update.
3182 (tree_function_versioning): Update.
3184 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3186 * hash-set.h (hash_set::empty): New.
3187 * tree-ssa-threadbackward.h: Delete.
3188 * tree-ssa-threadbackward.c (class thread_jumps): New.
3189 Move max_threaded_paths into class.
3190 (fsm_find_thread_path): Remove arguments that are now in class.
3191 (profitable_jump_thread_path): Rename to...
3192 (thread_jumps::profitable_jump_thread_path): ...this.
3193 (convert_and_register_jump_thread_path): Rename to...
3194 (thread_jumps::convert_and_register_current_path): ...this.
3195 (check_subpath_and_update_thread_path): Rename to...
3196 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3197 (register_jump_thread_path_if_profitable): Rename to...
3198 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3199 (handle_phi): Rename to...
3200 (thread_jumps::handle_phi): ...this.
3201 (handle_assignment): Rename to...
3202 (thread_jumps::handle_assignment): ...this.
3203 (fsm_find_control_statement_thread_paths): Rename to...
3204 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3205 (find_jump_threads_backwards): Rename to...
3206 (thread_jumps::find_jump_threads_backwards): ...this.
3207 Initialize path local data.
3208 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3209 from within thread_jumps class.
3210 (pass_early_thread_jumps::execute): Same.
3212 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3214 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3216 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3218 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3219 to canonical location.
3221 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3224 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3226 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3228 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3230 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3231 rcrt1.o%s/grcrt1.o%s for -static-pie.
3233 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3235 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3236 ix86_shift_rotate_cost): Break out from ...
3237 (ix86_rtx_costs): ... here.
3238 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3241 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3243 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3244 when entry block was promoted unlikely.
3245 (estimate_bb_frequencies): Increase frequency scale.
3246 * profile-count.h (profile_count): Export precision info.
3248 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3250 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3251 disturbin profile of entry block.
3253 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3255 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3258 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3260 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3261 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3263 2017-11-17 Jeff Law <law@redhat.com>
3265 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3266 from evrp_dom_walker class. Various methods moved into new class.
3267 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3268 (evrp_range_analyzer::enter): New method.
3269 (evrp_range_analyzer::leave): New method.
3270 (evrp_dom_walker): Remove delegators no longer needed by this class.
3271 Replace vr_values data member with evrp_range_analyzer
3273 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3274 method extracted from evrp_dom_walker::before_dom_children.
3275 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3276 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3278 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3279 Add private copy constructor and move assignment operators.
3280 Privatize methods and class data where trivially possible.
3281 (evrp_dom_walker::cleanup): New function, extracted from
3282 execute_early_vrp. Simplify access to class data.
3284 * vr-values.h (get_output_for_vrp): Prototype.
3285 * vr-values.c (get_output_for_vrp): New function extracted from
3286 vrp_visit_assignment_or_call and extract_range_from_stmt.
3287 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3289 2017-11-17 Luis Machado <luis.machado@linaro.org>
3291 * config/aarch64/aarch64.c
3292 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3293 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3294 tune_params::AUTOPREFETCHER_WEAK.
3296 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3299 * config/arm/arm.c (arm_valid_target_attribute_rec):
3300 Parse "arch=" and "+<ext>".
3301 (arm_valid_target_attribute_tree): Re-init global options.
3302 (arm_option_override): Make non-static.
3303 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3304 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3305 (__ARM_FEATURE_CRC32): Support undef.
3306 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3307 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3309 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3311 * gdbinit.in (break-on-diagnostic): New command.
3313 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3315 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3316 return to indirect jump for EH return if control-flow
3317 protection is enabled. Change explicit 'false' argument in
3318 pro_epilogue_adjust_stack with a value of
3320 * config/i386/i386.md (simple_return_indirect_internal):
3321 Remove SImode restriction to support 64-bit.
3323 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3325 * combine.c (added_notes_insn): New.
3326 (try_combine): Handle added_notes_insn like added_links_insn.
3327 Rewrite return value code.
3328 (distribute_notes): Set added_notes_insn to the earliest insn we added
3331 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3333 PR rtl-optimization/82621
3334 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3335 dest of one of those SETs is unused.
3337 2017-11-17 Richard Biener <rguenther@suse.de>
3340 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3341 * tree-pretty-print.c (dump_generic_node): Handle
3342 annot_expr_parallel_kind.
3343 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3344 * gimplify.c (gimple_boolify): Likewise.
3346 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3348 * config.gcc (extra_headers): Add cet.h for x86 targets.
3349 * config/i386/cet.h: New file.
3350 * doc/install.texi: Add --enable-cet/--disable-cet.
3352 2017-11-17 Richard Biener <rguenther@suse.de>
3354 PR tree-optimization/83017
3355 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3356 (gen_parallel_loop): Properly count iterations.
3357 (parallelize_loops): Handle loop->can_be_parallel independent
3358 of flag_loop_parallelize_all. Make static profitability test match
3360 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3361 * invoke.texi (parloops-min-per-thread): Document.
3363 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3365 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3366 upstreaming review comments.
3368 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3370 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3371 with fast unaligned access.
3372 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3374 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3376 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3377 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3379 2017-11-17 Richard Biener <rguenther@suse.de>
3381 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3382 folding of references.
3384 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3387 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3388 of replacing call to strncmp with corresponding call to strcmp when
3391 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3393 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3394 option prefer-avx256 for skylake-avx512 configuration.
3395 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3396 (get_builtin_code_for_version): Ditto.
3398 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3399 Monk Chiang <sh.chiang04@gmail.com>
3401 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3402 (FIXED_REGISTERS): Reserve more register numbers.
3403 (CALL_USED_REGISTERS): Likewise.
3404 (REG_ALLOC_ORDER): Likewise.
3405 (REG_CLASS_CONTENTS): Likewise.
3406 (REGISTER_NAMES): Likewise.
3408 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3409 Kito Cheng <kito.cheng@gmail.com>
3411 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3413 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3415 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3417 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3418 parameter list for vec_splats.
3420 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3422 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3424 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3426 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3429 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3430 with the required version of automake 1.11.6.
3432 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3434 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3436 (DU_C2_3_power9): Correct reservation combinations.
3437 (FP_DIV_power9, VEC_DIV_power9): New.
3438 (power9-alu): Split out rotate/shift...
3439 (power9-rot): ...to here, correct dispatch resource.
3440 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3442 (power9-fp): Correct latency.
3443 (power9-sdiv): Add div/sqrt resource.
3444 (power9-ddiv): Correct latency, add div/sqrt resource.
3445 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3446 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3448 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3450 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3452 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3453 switch statement mapping KF built-ins to TF built-ins if we don't
3454 have the proper ISA 3.0 assembler support.
3456 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3458 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3459 (lower_emutls_function_body): Do not compute it.
3461 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3463 * ipa-split.c (split_bb_info): Turn time to sreal.
3464 (split_point): Likewise.
3465 (dump_split_point): Likewise.
3466 (fine_split_points): Likewise.
3467 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3469 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3471 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3474 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3476 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3478 (expensive_function_p): Remove useless assert.
3479 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3481 2017-11-16 Martin Sebor <msebor@redhat.com>
3483 PR tree-optimization/82588
3484 PR tree-optimization/82583
3485 * tree-vrp.c (check_array_ref): Handle flexible array members,
3486 string literals, and inner indices.
3487 (search_for_addr_array): Add detail to diagnostics.
3489 2017-11-16 Nathan Sidwell <nathan@acm.org>
3493 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3494 (SET_DECL_ASSEMBLER_NAME): Forward to
3495 overwrite_decl_assembler_name.
3496 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3497 (overwrite_decl_assembler_name): Declare.
3498 * tree.c (overwrite_decl_assembler_name): New.
3499 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3500 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3501 (LANG_HOOKS_INITIALIZER): Add it.
3502 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3503 * langhooks.c (lhd_set_decl_assembler_name): Use
3504 SET_DECL_ASSEMBLER_NAME.
3505 (lhd_overwrite_decl_assembler_name): Default implementation.
3507 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3508 Jackson Woodruff <jackson.woodruff@arm.com>
3510 PR tree-optimization/71026
3511 * match.pd: Canonicalize constant multiplies in division.
3513 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3515 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3516 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3517 Remove OPT_fomit_frame_pointer entry.
3518 * common/config/arc/arc-common.c: Likewise.
3519 * common/config/arm/arm-common.c: Likewise.
3520 * common/config/avr/avr-common.c: Likewise.
3521 * common/config/c6x/c6x-common.c: Likewise.
3522 * common/config/cr16/cr16-common.c: Likewise.
3523 * common/config/cris/cris-common.c: Likewise.
3524 * common/config/epiphany/epiphany-common.c: Likewise.
3525 * common/config/fr30/fr30-common.c: Likewise.
3526 * common/config/frv/frv-common.c: Likewise.
3527 * common/config/ia64/ia64-common.c: Likewise.
3528 * common/config/iq2000/iq2000-common.c: Likewise.
3529 * common/config/lm32/lm32-common.c: Likewise.
3530 * common/config/m32r/m32r-common.c: Likewise.
3531 * common/config/mcore/mcore-common.c: Likewise.
3532 * common/config/microblaze/microblaze-common.c: Likewise.
3533 * common/config/mips/mips-common.c: Likewise.
3534 * common/config/mmix/mmix-common.c: Likewise.
3535 * common/config/mn10300/mn10300-common.c: Likewise.
3536 * common/config/nios2/nios2-common.c: Likewise.
3537 * common/config/pa/pa-common.c: Likewise.
3538 * common/config/pdp11/pdp11-common.c: Likewise.
3539 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3540 * common/config/riscv/riscv-common.c: Likewise.
3541 * common/config/rs6000/rs6000-common.c: Likewise.
3542 * common/config/rx/rx-common.c: Likewise.
3543 * common/config/s390/s390-common.c: Likewise.
3544 * common/config/sh/sh-common.c: Likewise.
3545 * common/config/sparc/sparc-common.c: Likewise.
3546 * common/config/tilegx/tilegx-common.c: Likewise.
3547 * common/config/tilepro/tilepro-common.c: Likewise.
3548 * common/config/v850/v850-common.c: Likewise.
3549 * common/config/visium/visium-common.c: Likewise.
3550 * common/config/xstormy16/xstormy16-common.c: Likewise.
3551 * common/config/xtensa/xtensa-common.c: Likewise.
3552 * invoke.texi (-fomit-frame-pointer): Update documentation.
3554 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3556 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3558 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3560 * cfg.c (scale_bbs_frequencies_int,
3561 cale_bbs_frequencies_gcov_type): Remove.
3562 * cfg.h (scale_bbs_frequencies_int,
3563 cale_bbs_frequencies_gcov_type): Remove.
3565 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3567 * tree-ssa-loop-manip.c
3568 (scale_dominated_blocks_in_loop): Update to profile counts.
3569 (tree_transform_and_unroll_loop): Likewise.
3571 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3573 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3574 scale_bbs_frequencies_int.
3576 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3578 * final.c (compute_alignments): Use counts rather than frequencies.
3580 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3582 * cfgloopanal.c: Include sreal.h
3583 (average_num_loop_insns): Use counts and sreal for accounting.
3585 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3587 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3590 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3591 Kito Cheng <kito.cheng@gmail.com>
3593 * config/nds32/constraints.md: Provide more constraints.
3594 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3595 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3596 support constraints usage.
3598 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3600 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3602 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3603 Kito Cheng <kito.cheng@gmail.com>
3605 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3606 * config/nds32/nds32.opt: Refine the layout.
3607 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3608 TARGET_EXT_STRING): Support new options.
3609 * config/nds32/nds32.h: Likewise.
3610 * config/nds32/nds32.md: Likewise.
3611 * config/nds32/nds32-predicates.c: Likewise.
3612 * config/nds32/constraints.md: Likewise.
3613 * common/config/nds32/nds32-common.c: Likewise.
3615 2017-11-16 Julia Koval <julia.koval@intel.com>
3618 * config/i386/gfniintrin.h: Add sse check.
3619 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3621 2017-11-16 Julia Koval <julia.koval@intel.com>
3623 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3624 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3625 (ix86_handle_option): Handle -mavx512vbmi2.
3626 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3627 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3628 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3629 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3630 (ix86_valid_target_attribute_inner_p): Ditto.
3631 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3632 * config/i386/i386.opt (mavx512vbmi2): New option.
3633 * doc/invoke.texi: Add new option.
3635 2017-11-16 Julia Koval <julia.koval@intel.com>
3637 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3638 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3639 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3640 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3641 _mm512_gf2p8mul_epi8): New intrinsics.
3642 * config/i386/i386-builtin-types.def
3643 (V64QI_FTYPE_V64QI_V64QI): New type.
3644 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3645 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3646 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3647 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3648 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3649 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3651 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3653 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3654 explicitly as a stream of bytes.
3656 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3658 * config/rs6000/altivec.h (vec_xst_be): New #define.
3659 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3660 and externalize from *altivec_vperm_<mode>_internal.
3661 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3663 (XL_BE_V8HI): Likewise.
3664 (XL_BE_V4SI): Likewise.
3665 (XL_BE_V4SI): Likewise.
3666 (XL_BE_V2DI): Likewise.
3667 (XL_BE_V4SF): Likewise.
3668 (XL_BE_V2DF): Likewise.
3669 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3670 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3671 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3672 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3673 VSX_BUILTIN_VEC_XST_BE.
3674 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3675 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3677 (altivec_init_builtins): Replace conditional calls to def_builtin
3678 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3679 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3680 with unconditional calls. Remove calls to def_builtin for
3681 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3682 __builtin_vec_xst_be.
3683 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3684 to define_expand, and add alternate RTL generation for P8.
3685 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3686 vsx_ld_elemrev_v8hi.
3687 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3688 add alternate RTL generation for P8.
3689 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3690 vsx_ld_elemrev_v16qi.
3691 (vsx_st_elemrev_v8hi): Convert define_insn
3692 to define_expand, and add alternate RTL generation for P8.
3693 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3694 vsx_st_elemrev_v8hi.
3695 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3696 add alternate RTL generation for P8.
3697 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3698 vsx_st_elemrev_v16qi.
3700 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3703 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3704 TARGET_AVX512ER check.
3705 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3706 neither -mzeroupper nor -mno-zeroupper is used and
3707 TARGET_EMIT_VZEROUPPER is set.
3708 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3709 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3711 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3713 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3714 folding of vector compares.
3715 (fold_build_vec_cmp): New helper function.
3716 (fold_compare_helper): New helper function.
3717 (builtin_function_type): Add compare builtins to the list of functions
3718 having unsigned arguments. Cosmetic updates to comment indentation.
3719 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3720 the not+eq combination.
3722 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3724 PR tree-optimization/82726
3725 PR tree-optimization/70754
3726 * tree-predcom.c (order_drefs_by_pos): New function.
3727 (combine_chains): Move code setting has_max_use_after to...
3728 (try_combine_chains): ...here. New parameter. Sort combined chains
3729 according to position information.
3730 (tree_predictive_commoning_loop): Update call to above function.
3731 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3733 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3735 PR tree-optimization/82726
3737 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3739 PR tree-optimization/70754
3740 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3741 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3742 combined stmt before it if not NULL.
3743 (combine_chains): Process refs reversely and compute dominance point
3747 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3749 PR tree-optimization/79663
3750 * tree-predcom.c (combine_chains): Process refs in reverse order
3751 only for ZERO length chains, and add explaining comment.
3753 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3755 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3756 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3757 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3759 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3761 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3763 2017-11-15 Martin Liska <mliska@suse.cz>
3765 * tree-cfg.c (pass_warn_function_return::execute):
3766 Compare warn_return_type for greater than zero.
3768 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3772 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3773 to return true on Xeon and not on Xeon Phi.
3774 (ix86_check_avx256_register): Changed to ...
3775 (ix86_check_avx_upper_register): ... this. Add extra check for
3776 VALID_AVX512F_REG_OR_XI_MODE.
3777 (ix86_avx_u128_mode_needed): Changed
3778 ix86_check_avx256_register to ix86_check_avx_upper_register.
3779 (ix86_check_avx256_stores): Changed to ...
3780 (ix86_check_avx_upper_stores): ... this. Changed
3781 ix86_check_avx256_register to ix86_check_avx_upper_register.
3782 (ix86_avx_u128_mode_after): Changed
3783 avx_reg256_found to avx_upper_reg_found. Changed
3784 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3785 (ix86_avx_u128_mode_entry): Changed
3786 ix86_check_avx256_register to ix86_check_avx_upper_register.
3787 (ix86_avx_u128_mode_exit): Ditto.
3788 * config/i386/i386.h: (host_detect_local_cpu): New define.
3790 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3792 * config/arm/xgene1.md (xgene1): Split into automatons
3793 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3794 (xgene1_f_load): Adjust reservations and/or types.
3795 (xgene1_f_store): Likewise.
3796 (xgene1_load_pair): Likewise.
3797 (xgene1_store_pair): Likewise.
3798 (xgene1_fp_load1): Likewise.
3799 (xgene1_load1): Likewise.
3800 (xgene1_store1): Likewise.
3801 (xgene1_move): Likewise.
3802 (xgene1_alu): Likewise.
3803 (xgene1_simd): Likewise.
3804 (xgene1_bfm): Likewise.
3805 (xgene1_neon_load1): Likewise.
3806 (xgene1_neon_store1): Likewise.
3807 (xgene1_neon_logic): Likewise.
3808 (xgene1_neon_st1): Likewise.
3809 (xgene1_neon_ld1r): Likewise.
3810 (xgene1_alu_cond): Added.
3811 (xgene1_shift_reg): Likwise.
3812 (xgene1_bfx): Likewise.
3813 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3815 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3818 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3820 (can_widen_mult_without_libcall): New function.
3821 (expand_mul_overflow): If only checking unsigned mul overflow,
3822 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3823 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3824 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3825 (expand_DIVMOD): Formatting fix.
3826 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3827 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3828 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3830 PR tree-optimization/82977
3831 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3832 constructed temporary to strlen_to_stridx.put.
3834 2017-11-15 Martin Liska <mliska@suse.cz>
3836 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3837 * configure: Regenerate.
3839 2017-11-15 Martin Liska <mliska@suse.cz>
3842 * config/sh/sh-mem.cc: Use proper probability for
3845 2017-11-14 Jeff Law <law@redhat.com>
3847 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3848 the red zone for stack_clash_protection_final_dynamic_probe targets
3849 when the total dynamic stack size is zero bytes.
3851 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3852 blocks is post order.
3854 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3856 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3857 * final.c (rest_of_clean_state): Set it for the
3858 -fcompare-debug dump.
3859 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3860 class when TDF_COMPARE_DEBUG is set.
3862 * dwarf2out.c (gen_producer_string): Discard
3865 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3868 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3869 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3870 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3871 (__TGMATH_REAL_2_3): Remove macros.
3872 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3873 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3875 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3877 (remquo): Define using __TGMATH_REAL_3.
3879 2017-11-14 Jeff Law <law@redhat.com>
3881 * vr-values.c: New file with contents extracted from tree-vrp.c.
3882 * Makefile.in (OBJS): Add vr-values.o
3883 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3884 (set_value_range, set_and_canonicalize_value_range): Likewise.
3885 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3886 (value_range_constant_singleton, symbolic_range_p): Likewise.
3887 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3888 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3889 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3890 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3891 (find_case_label_range, find_case_label_index): Likewise.
3892 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3893 (range_int_cst_singleton_p, value_inside_range): Likewise.
3894 (get_single_symbol): Likewise.
3895 (switch_update): Move structure definition here.
3896 (to_remove_edges, to_update_switch_stmts): Provide externs.
3897 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3898 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3899 (vrp_val_is_min, set_value_range): Likewise.
3900 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3901 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3902 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3903 (range_is_nonnull, range_int_cst_p): Likewwise.
3904 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3905 (get_single_symbol, operand_less_p): Likewise
3906 (compare_values_warnv, compare_values): Likewise.
3907 (value_inside_range, value_range_constant_singleton): Likewise.
3908 (zero_nonzero_bitgs_from_vr): Likewise.
3909 (extract_range_from_binary_expr_1): Likewise.
3910 (overflow_comparison_p): Likewise.
3911 (to_remove_edges, to_update_switch_stmts): Likewise.
3912 (find_case_label-index, find_case_label_range): Likewise.
3913 (switch_update, set_value_range_to_nonnegative): Remove.
3914 (set_value_range_to_truthvalue): Likewise.
3915 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3916 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3917 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3918 (find_case_label_ranges, test_for_singularity): Likewise.
3919 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3920 (x_vr_values): Move to its remaining use site.
3922 2017-11-10 Jeff Law <law@redhat.com>
3924 * vr-values.h (VR_INITIALIZER): Move #define here.
3925 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3926 * Makefile.in (OBJS): Add tree-evrp.o
3927 * tree-vrp.h (assert_info): Move structure definition here.
3928 (set_value_range_to_varying): Prototype.
3929 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3930 (infer_value_range, register_edge_assert_for): Likewise.
3931 (stmt_interesting_for_vrp): Likewise.
3932 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3933 (set_value_range_to_varying): No longer static.
3934 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3935 (infer_value_range, register_edge_assert_for): Likewise.
3937 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3939 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3940 generate the XXBRD instruction.
3942 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3943 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3944 as being compatible if -mabi=ieeelongdouble.
3945 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3946 to setup float128 built-ins with hardware support.
3947 (BU_FLOAT128_HW_2): Likewise.
3948 (BU_FLOAT128_HW_3): Likewise.
3949 (BU_FLOAT128_HW_VSX_1): Likewise.
3950 (BU_FLOAT128_HW_VSX_2): Likewise.
3951 (scalar_extract_expq): Change float128 built-in functions to
3952 accommodate having both KFmode and TFmode functions. Use the
3953 KFmode variant as the default.
3954 (scalar_extract_sigq): Likewise.
3955 (scalar_test_neg_qp): Likewise.
3956 (scalar_insert_exp_q): Likewise.
3957 (scalar_insert_exp_qp): Likewise.
3958 (scalar_test_data_class_qp): Likewise.
3959 (sqrtf128_round_to_odd): Delete processing the round to odd
3960 built-in functions as special built-in functions, and define them
3961 as float128 built-ins. Use the KFmode variant as the default.
3962 (truncf128_round_to_odd): Likewise.
3963 (addf128_round_to_odd): Likewise.
3964 (subf128_round_to_odd): Likewise.
3965 (mulf128_round_to_odd): Likewise.
3966 (divf128_round_to_odd): Likewise.
3967 (fmaf128_round_to_odd): Likewise.
3968 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3969 support for KFmode and TFmode xststdcqp calls.
3970 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3971 point, switch the built-in handlers for the get/set float128
3972 exponent, get float128 mantissa, float128 test built-ins, and the
3973 float128 round to odd built-in functions. Eliminate creating the
3974 float128 round to odd built-in functions as special built-ins.
3975 (rs6000_init_builtins): Eliminate special creation of the float128
3976 round to odd built-in functions.
3977 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3978 function insns to support both TFmode and KFmode variants.
3979 (xsxsigqp_<mode>): Likewise.
3980 (xsiexpqpf_<mode>): Likewise.
3981 (xsiexpqp_<mode>): Likewise.
3982 (xststdcqp_<mode>): Likewise.
3983 (xststdcnegqp_<mode>): Likewise.
3984 (xststdcqp_<mode>): Likewise.
3986 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3988 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3989 unused path_in_freq_ptr parameter.
3990 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3992 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3994 * ipa-inline.c (edge_badness): Dump sreal frequency.
3995 (compute_inlined_call_time): Match natural implementaiton ...
3996 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3997 forgotten division by CGRAPH_FREQ_BASE.
3999 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4001 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4002 Solaris 11. Update comment.
4003 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4005 * config/sol2.h (STARTFILE_SPEC): Likewise.
4006 * configure: Regenerate.
4008 2017-11-14 Carl Love <cel@us.ibm.com>
4010 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4011 le_ and be_ prefixes to swap* variables. Remove
4012 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4015 2017-11-14 Jason Merrill <jason@redhat.com>
4017 Support GTY((cache)) on hash_map.
4018 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4019 (ggc_cache_remove): Override it instead of ggc_mx.
4020 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4021 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4022 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4023 (simple_cache_map_traits): Override maybe_mx.
4024 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4025 (hash_map): Friend gt_cleare_cache.
4026 (gt_cleare_cache): New.
4027 * tree.h (tree_cache_traits): New hash_map traits class.
4028 (tree_cache_map): New typedef.
4030 2017-11-14 Richard Biener <rguenther@suse.de>
4032 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4033 paramter and handling.
4034 (cleanup_control_flow_bb): Likewise.
4035 (cleanup_control_flow_pre): New helper performing a DFS walk
4036 to call cleanup_control_flow_bb in PRE order.
4037 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4038 via cleanup_control_flow_pre.
4040 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4042 * config/aarch64/aarch64-simd.md
4043 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4044 (*aarch64_simd_bsl<mode>_alt): Likewise.
4045 (aarch64_simd_bsldi_internal): New.
4046 (aarch64_simd_bsldi_alt): Likewise.
4048 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4050 * tracer.c (better_p): Do not compare frequencies.
4051 * reg-stack.c (better_edge): Likewise.
4052 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4055 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4057 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4058 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4059 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4060 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4061 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4062 * predict.c (maybe_hot_frequency_p): Inline to ...
4063 (maybe_hot_count_p): ... here; rewrite to counts.
4064 (counts_to_freqs): Rename to ...
4065 (update_max_bb_count): ... this one.
4066 (expensive_function_p): Use counts.
4067 (estimate_bb_frequencies): Update.
4068 (rebuild_frequencies): Update.
4069 * predict.h (counts_to_freqs): Rename to ...
4070 (update_max_bb_count): ... this one.
4071 * profile.c (compute_branch_probabilities): Add debug info
4072 * tree-inline.c (expand_call_inline): Update debug info.
4073 (optimize_inline_calls): Use update_max_bb_count..
4074 (tree_function_versioning): Use update_max_bb_count..
4075 * value-prof.c (gimple_value_profile_transformations):
4076 Do not use update_max_bb_count.
4078 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4080 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4081 always use frequencies.
4083 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4085 * bb-reorder.c: Remove frequencies from comments.
4086 (better_edge_p): Use profile counts.
4087 (find_traces): Dump profile counts.
4088 (rotate_loop): Use profile counts.
4089 (find_traces_1_round): Likewise.
4090 (connect_better_edge_p): Use counts instead of probabilities for
4092 (copy_bb_p): Drop early check for non-0 frequency.
4093 (sanitize_hot_paths): Update comments.
4095 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4097 * ipa-split.c (struct split_point): Add count.
4098 (consider_split): Do not compute incoming frequency; compute incoming
4099 count and store it to split_point.
4100 (split_function): Set count of the call to split part correctly.
4102 2017-11-13 Carl Love <cel@us.ibm.com>
4104 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4106 2017-11-13 Tom Tromey <tom@tromey.com>
4108 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4110 2017-11-13 Carl Love <cel@us.ibm.com>
4112 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4113 Add support for builtins:
4114 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4115 vector {un,}signed {char,int,short},
4116 vector {un,}signed {char,int,short}) arguments.
4117 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4118 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4119 Add BU_P9V_AV_2 expansions for the builtins.
4120 * config/rs6000/altivec.h (vec_first_match_index,
4121 vec_first_mismatch_index, vec_first_match_or_eos_index,
4122 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4123 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4124 new extern declaration.
4125 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4126 * config/rs6000/vsx.md (first_match_index_<mode>,
4127 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4128 first_mismatch_or_eos_index_<mode>): Add define expand.
4129 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4130 * doc/extend.texi: Update the built-in documenation file for the new
4133 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4135 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4136 into the min/max operations for _Float<N> and _Float<N>X types.
4138 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4141 * dwarf2out.c (do_eh_frame): New static variable.
4142 (dwarf2out_begin_prologue): Set it.
4143 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4145 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4147 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4149 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4151 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4154 * debug.h (dwarf2out_do_eh_frame): Declare.
4155 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4156 (dwarf2out_do_frame): Use it.
4157 (dwarf2out_do_cfi_asm): Likewise.
4158 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4159 (dwarf2out_assembly_start): Likewise.
4160 (dwarf2out_begin_prologue): Fix comment.
4161 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4162 if the target needs either debug or unwind DWARF2 info.
4163 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4164 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4165 -ftrapping-math, -ftrapv and -fwrapv.
4167 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4169 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4170 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4171 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4172 (estimate_edge_size_and_time): Likewise.
4173 (ipa_merge_fn_summary_after_inlining): Likewise.
4174 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4175 (compute_uninlined_call_time): Use sreal_frequency.
4176 (compute_inlined_call_time): Likewise.
4177 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4178 * profile-count.c: Include sreal.h.
4179 (profile_count::to_sreal_scale): New.
4180 * profile-count.h: Forward declare sreal.
4181 (profile_count::to_sreal_scale): Declare.
4183 2017-11-13 Nathan Sidwell <nathan@acm.org>
4185 * diagnostic.c (maybe_line_and_column): New.
4186 (diagnostic_get_location_text): Use it.
4187 (diagnostic_report_current_module): Likewise.
4188 (test_diagnostic_get_location_text): Add tests.
4190 2017-11-13 Luis Machado <luis.machado@linaro.org>
4192 * doc/md.texi (Specifying processor pipeline description): Fix
4193 incorrect latency for the div instruction example.
4195 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4197 PR tree-optimization/78821
4198 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4199 that bit_not_p is the same.
4200 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4201 (split_group): Count precisely bit_not_p bits in each statement.
4202 (invert_op): New function.
4203 (imm_store_chain_info::output_merged_store): Use invert_op to
4204 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4205 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4207 2017-11-13 Martin Liska <mliska@suse.cz>
4209 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4210 (struct source_info): Likewise.
4211 (add_branch_counts): Likewise.
4212 (add_line_counts): Likewise.
4213 (function_summary): Likewise.
4214 (output_intermediate_line): Likewise.
4215 (generate_results): Likewise.
4217 2017-11-13 Martin Liska <mliska@suse.cz>
4219 * gcov.c (struct block_info): Remove typedef for block_t.
4220 (struct line_info): Likewise.
4221 (line_info::has_block): Likewise.
4222 (EXIT_BLOCK): Likewise.
4223 (unblock): Likewise.
4224 (circuit): Likewise.
4225 (get_cycles_count): Likewise.
4226 (process_file): Likewise.
4227 (read_graph_file): Likewise.
4228 (solve_flow_graph): Likewise.
4229 (find_exception_blocks): Likewise.
4230 (add_line_counts): Likewise.
4231 (accumulate_line_info): Likewise.
4232 (output_line_details): Likewise.
4234 2017-11-13 Martin Liska <mliska@suse.cz>
4236 * gcov.c (struct arc_info): Remove typedef for arc_t.
4237 (struct line_info): Likewise.
4238 (add_branch_counts): Likewise.
4239 (output_branch_count): Likewise.
4240 (function_info::~function_info): Likewise.
4241 (circuit): Likewise.
4242 (output_intermediate_line): Likewise.
4243 (read_graph_file): Likewise.
4244 (solve_flow_graph): Likewise.
4245 (find_exception_blocks): Likewise.
4246 (add_line_counts): Likewise.
4247 (accumulate_line_info): Likewise.
4248 (output_line_details): Likewise.
4249 (output_function_details): Likewise.
4251 2017-11-13 Martin Liska <mliska@suse.cz>
4253 * gcov.c (struct function_info): Remove typedef for function_t.
4254 (struct source_info): Likewise.
4255 (source_info::get_functions_at_location): Likewise.
4256 (solve_flow_graph): Likewise.
4257 (find_exception_blocks): Likewise.
4258 (add_line_counts): Likewise.
4259 (output_intermediate_file): Likewise.
4260 (process_file): Likewise.
4261 (generate_results): Likewise.
4262 (release_structures): Likewise.
4263 (read_graph_file): Likewise.
4264 (read_count_file): Likewise.
4265 (accumulate_line_counts): Likewise.
4266 (output_lines): Likewise.
4268 2017-11-13 Martin Liska <mliska@suse.cz>
4270 * gcov.c (function_info::function_info): Remove num_counts
4271 and add vector<gcov_type>.
4272 (function_info::~function_info): Use the vector.
4273 (process_file): Likewise.
4274 (read_graph_file): Likewise.
4275 (read_count_file): Likewise.
4276 (solve_flow_graph): Likewise.
4278 2017-11-13 Martin Liska <mliska@suse.cz>
4280 * gcov.c (function_info::is_artificial): New function.
4281 (process_file): Erase all artificial early.
4282 (generate_results): Skip as all artificial are already
4285 2017-11-13 Martin Liska <mliska@suse.cz>
4287 * gcov.c (read_graph_file): Store to global vector of functions.
4288 (read_count_file): Iterate the vector.
4289 (process_file): Likewise.
4290 (generate_results): Likewise.
4291 (release_structures): Likewise.
4293 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4295 PR tree-optimization/82954
4296 * gimple-ssa-store-merging.c
4297 (imm_store_chain_info::coalesce_immediate_stores): If
4298 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4300 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4302 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4303 Upddate call to ENDIAN_LANE_N.
4304 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4305 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4306 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4307 and use aarch64_endian_lane_rtx.
4308 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4310 2017-11-12 Tom de Vries <tom@codesourcery.com>
4312 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4314 2017-11-12 Tom de Vries <tom@codesourcery.com>
4316 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4318 2017-11-12 Tom de Vries <tom@codesourcery.com>
4320 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4322 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4323 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4324 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4326 2017-11-11 Martin Sebor <msebor@redhat.com>
4329 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4332 * prefic.c (translate_name): Replace strncpy with memcpy to
4333 avoid -Wstringop-truncation.
4335 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4337 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4339 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4341 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4343 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4345 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4348 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4350 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4351 enable generating XXBRH if the value is in a vector register.
4352 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4353 value is in a vector register.
4354 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4355 register to register bswap64's instead of doing the GPR sequence
4356 used on previous machines.
4357 (bswapdi2_xxbrd): New insn.
4358 (bswapdi2_reg): Disallow on ISA 3.0.
4359 (register to register bswap64 splitter): Do not split the insn on
4360 ISA 3.0 systems that use XXBRD.
4362 2017-11-10 Martin Sebor <msebor@redhat.com>
4365 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4367 (find_subframework_file): Same.
4369 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4371 * auto-profile.c (afdo_indirect_call): Drop frequency.
4372 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4373 (cgraph_node::create_edge): Drop frequency argument.
4374 (cgraph_node::create_indirect_edge): Drop frequency argument.
4375 (cgraph_edge::make_speculative): Drop frequency arguments.
4376 (cgraph_edge::resolve_speculation): Do not update frequencies
4377 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4378 (cgraph_node::dump): Check consistency in IPA mode.
4379 (cgraph_edge::maybe_hot_p): Use IPA counter.
4380 (cgraph_edge::verify_count_and_frequency): Rename to ...
4381 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4382 (cgraph_node::verify_node): Update.
4383 * cgraph.h (struct cgraph_edge): Drop frequency.
4384 (cgraph_edge::frequency): New function.
4385 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4387 (cgraph_edge::rebuild_edges): Likewise.
4388 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4389 (duplicate_thunk_for_node): Do not pass frequency.
4390 (cgraph_node::create_clone): Scale only counts.
4391 (cgraph_node::create_virtual_clone): Do not pass frequency.
4392 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4393 (cgraph_node::create_version_clone): Do not pass frequency.
4394 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4395 (cgraph_node::expand_thunk): Do not pass frequency.
4396 (cgraph_node::create_wrapper): Do not pass frequency.
4397 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4399 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4400 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4401 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4402 (gather_caller_stats): Use frequency function.
4403 (ipcp_cloning_candidate_p): Use frequency function.
4404 (ipcp_propagate_stage): Use frequency function.
4405 (get_info_about_necessary_edges): Use frequency function.
4406 (update_profiling_info): Update only IPA profile.
4407 (update_specialized_profile): Use frequency functoin.
4408 (perhaps_add_new_callers): Update only IPA profile.
4409 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4410 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4411 (dump_ipa_call_summary): Use frequency function.
4412 (estimate_edge_size_and_time): Use frequency function.
4413 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4414 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4415 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4416 (update_noncloned_counts): ... ths one; scale counts only.
4417 (clone_inlined_nodes): Do not scale frequency.
4418 (inline_call): Do not pass frequency.
4419 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4420 (compute_inlined_call_time): Use IPA profile.
4421 (want_inline_small_function_p): Use IPA profile.
4422 (want_inline_self_recursive_call_p): Use IPA profile.
4423 (edge_badness): Use IPA profile.
4424 (lookup_recursive_calls): Use IPA profile.
4425 (recursive_inlining): Do not pass frequency.
4426 (resolve_noninline_speculation): Do not update frequency.
4427 (inline_small_functions): Collect max of IPA profile.
4428 (dump_overall_stats): Dump IPA porfile.
4429 (dump_inline_stats): Dump IPA porfile.
4430 (ipa_inline): Collect IPA stats.
4431 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4432 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4433 (ipa_propagate_frequency): Use frequency function.
4434 (ipa_profile): Cleanup.
4435 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4436 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4437 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4438 (input_node): Do not stream frequency.
4439 (input_edge): Do not stream frequency.
4440 (merge_profile_summaries): Scale only IPA profiles.
4441 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4442 * predict.c (drop_profile): Do not recompute frequency.
4443 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4444 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4445 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4446 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4447 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4448 * tree-inline.c (copy_bb): Do not scale frequency.
4449 (expand_call_inline): Do not scale frequency.
4450 (tree_function_versioning): Do not scale frequency.
4451 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4453 2017-11-10 Julia Koval <julia.koval@intel.com>
4455 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4456 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4457 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4458 (_mm256_maskz_gf2p8affine_epi64_epi8)
4459 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4460 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4461 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4462 (__builtin_ia32_vgf2p8affineqb_v32qi)
4463 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4464 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4466 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4470 (arm_option_override): Refactor.
4471 (arm_option_reconfigure_globals): New.
4472 (arm_options_perform_arch_sanity_checks): New.
4473 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4475 (arm_options_perform_arch_sanity_checks): Likewise
4477 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4479 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4480 (power9-qpmul): New.
4481 * rs6000/rs6000.md ("type" attr): Add qmul.
4482 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4483 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4484 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4486 2017-11-10 Martin Sebor <msebor@redhat.com>
4489 * builtins.c (compute_objsize): Handle arrays that
4490 compute_builtin_object_size likes to fail for. Make extern.
4491 * builtins.h (compute_objsize): Declare.
4492 (check_strncpy_sizes): New function.
4493 (expand_builtin_strncpy): Call check_strncpy_sizes.
4494 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4495 -Wstringop-truncation.
4496 (gimple_fold_builtin_strncat): Same.
4497 * gimple.c (gimple_build_call_from_tree): Set call location.
4498 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4499 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4500 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4501 (handle_builtin_strlen): Use strlen_to_stridx.
4502 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4504 Use strlen_to_stridx.
4505 (pass_strlen::execute): Release strlen_to_stridx.
4506 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4507 (-Wstringop-truncation): Document new option.
4509 2017-11-10 Martin Liska <mliska@suse.cz>
4511 PR gcov-profile/82702
4512 * gcov.c (main): Handle intermediate files in a different
4514 (get_gcov_intermediate_filename): New function.
4515 (output_gcov_file): Remove support of intermediate files.
4516 (generate_results): Allocate intermediate file.
4517 (release_structures): Clean-up properly fn_end.
4518 (output_intermediate_file): Start iterating with line 1.
4520 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4522 PR tree-optimization/82929
4523 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4524 ops_swapped_p non-static data member.
4525 (store_immediate_info::store_immediate_info): Clear it.
4526 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4527 ops set ops_swapped_p.
4528 (count_multiple_uses): Handle ops_swapped_p.
4530 2017-11-10 Martin Liska <mliska@suse.cz>
4532 * coverage.c (coverage_init): Stream information about
4533 support of has_unexecuted_blocks.
4534 * doc/gcov.texi: Document that.
4535 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4536 * gcov.c (read_graph_file): Likewise.
4537 (output_line_beginning): Fix a small issue with
4540 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4542 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4543 reference of trivial component.
4545 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4548 * gimple-ssa-store-merging.c
4549 (pass_store_merging::terminate_all_aliasing_chains): For
4550 gimple_store_p stmts also call refs_output_dependent_p.
4552 PR rtl-optimization/82913
4553 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4556 2017-11-09 Jeff Law <law@redhat.com>
4558 * vr-values.h: New file with vr_values class.
4559 * tree-vrp.c: Include vr-values.h
4560 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4561 data objects into the vr_values class.
4562 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4563 (get_value_range): Make it a member function within vr_values class.
4564 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4565 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4566 (op_with_constant_singleton_value_range): Likewise.
4567 (extract_range_for_var_from_comparison_expr): Likewise.
4568 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4569 (extract_range_from_binary_expr): Likewise.
4570 (extract_range_from_unary_expr): Likewise.
4571 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4572 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4573 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4574 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4575 (compare_name_with_value, compare_names): Likewise.
4576 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4577 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4578 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4579 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4580 (extract_range_from_phi_node): Likewise.
4581 (simplify_truth_ops_using_ranges): Likewise.
4582 (simplify_div_or_mod_using_ranges): Likewise.
4583 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4584 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4585 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4586 (simplify_float_conversion_using_ranges): Likewise.
4587 (simplify_internal_call_using_ranges): Likewise.
4588 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4589 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4590 poitner via x_vr_values for calls into gimple folder.
4591 (vrp_initialize_lattice): Make this the vr_values ctor.
4592 (vrp_free_lattice): Make this the vr_values dtor.
4593 (set_vr_value): New function.
4594 (class vrp_prop): Add vr_values data member. Add various member
4595 functions as well as member functions that delegate to vr_values.
4596 (check_array_ref): Make a member function within vrp_prop class.
4597 (search_for_addr_array, vrp_initialize): Likewise.
4598 (vrp_finalize): Likewise. Revamp to avoid direct access to
4599 vr_value, values_propagated, etc.
4600 (check_array_bounds): Extract vrp_prop class instance pointer from
4601 walk info structure. Use it to call member functions.
4602 (check_all_array_refs): Make a member function within vrp_prop class.
4603 Smuggle class instance pointer via walk info structure.
4604 (x_vr_values): New local static.
4605 (vrp_valueize): Use x_vr_values to get class instance.
4606 (vr_valueize_1): Likewise.
4607 (class vrp_folder): Add vr_values data member. Add various member
4608 functions as well as member functions that delegate to vr_values.
4609 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4610 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4611 class instance from vr_values. Use it to call member functions.
4612 (vrp_dom_walker): Add vr_values data member.
4613 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4614 instance via x_vr_values.
4615 (identify_jump_threads): Accept vr_values as argument. Store
4616 it into the walker structure.
4617 (evrp_dom_walker): Add vr_values class data member. Add various
4619 (evrp_dom_walker::try_find_new_range): Use vr_values data
4620 member to access the memory allocator.
4621 (evrp_dom_walker::before_dom_children): Store vr_values class
4622 instance into the vrp_folder class.
4623 (evrp_dom_walker::push_value_range): Rework to avoid direct
4624 access to num_vr_values and vr_value.
4625 (evrp_dom_walker::pop_value_range): Likewise.
4626 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4627 Use vr_values to get to dump_all_value_ranges member function.
4628 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4629 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4630 Pass vr_values class instance down to identify_jump_threads.
4631 Remove call to vrp_free_lattice.
4632 (debug_all_value_ranges): Remove.
4634 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4635 (vrp_folder): Likewise.
4637 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4638 Get it from the existing bitmap instead.
4639 (vrp_intersect_ranges_1): Likewise.
4641 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4643 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4645 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4646 set bit_not_p to it.
4647 (imm_store_chain_info::coalesce_immediate_stores): Break group
4648 if bit_not_p is different.
4649 (count_multiple_uses, split_group,
4650 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4651 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4652 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4653 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4654 to store_immediate_info ctor.
4656 2017-11-09 Jim Wilson <jimw@sifive.com>
4658 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4659 (scan_prog_file): Likewise.
4661 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4663 * bb-reorder.c (max_entry_frequency): Remove.
4664 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4665 connect_traces, push_to_next_round_p): Remove prototypes.
4666 (find_traces_1_round): Use counts only.
4667 (push_to_next_round_p): Likewise.
4668 (find_traces): Likewise.
4669 (rotate_loop): Likewise.
4670 (find_traces_1_round): Likewise.
4671 (connect_traces): Likewise.
4672 (edge_order): Likewise.
4674 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4676 * config/arm/arm.c (output_return_instruction): Add comments to
4677 indicate requirement for cmse_nonsecure_entry return to account
4678 for the size of clearing instruction output here.
4679 (thumb_exit): Likewise.
4680 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4681 return in hardfloat mode.
4683 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4685 * config/rs6000/rs6000.c (machine_function): Add a bool,
4686 "toc_is_wrapped_separately".
4687 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4688 if it wasn't explicitly set or unset, we are optimizing for speed, and
4689 doing separate shrink-wrapping.
4690 (rs6000_get_separate_components): Enable the TOC component if
4691 saving the TOC register in the prologue.
4692 (rs6000_components_for_bb): Handle the TOC component.
4693 (rs6000_emit_prologue_components): Store the TOC register where needed.
4694 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4695 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4697 2017-11-09 Martin Jambor <mjambor@suse.cz>
4699 * ipa-param-manipulation.c: New file.
4700 * ipa-param-manipulation.h: Likewise.
4701 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4702 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4703 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4704 (ipa_parm_adjustment): Likewise.
4705 (ipa_parm_adjustment_vec): Likewise.
4706 (ipa_get_vector_of_formal_parms): Moved declaration to
4707 ipa-param-manipulation.h.
4708 (ipa_get_vector_of_formal_parm_types): Likewise.
4709 (ipa_modify_formal_parameters): Likewise.
4710 (ipa_modify_call_arguments): Likewise.
4711 (ipa_combine_adjustments): Likewise.
4712 (ipa_dump_param_adjustments): Likewise.
4713 (ipa_modify_expr): Likewise.
4714 (ipa_get_adjustment_candidate): Likewise.
4715 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4716 ipa-param-manipulation.c.
4717 (ipa_get_vector_of_formal_parm_types): Likewise.
4718 (ipa_modify_formal_parameters): Likewise.
4719 (ipa_modify_call_arguments): Likewise.
4720 (ipa_modify_expr): Likewise.
4721 (get_ssa_base_param): Likewise.
4722 (ipa_get_adjustment_candidate): Likewise.
4723 (index_in_adjustments_multiple_times_p): Likewise.
4724 (ipa_combine_adjustments): Likewise.
4725 (ipa_dump_param_adjustments): Likewise.
4726 * tree-sra.c: Also include ipa-param-manipulation.h
4727 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4730 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4731 Alan Hayward <alan.hayward@arm.com>
4732 David Sherwood <david.sherwood@arm.com>
4734 * doc/sourcebuild.texi (vect_masked_store): Document.
4736 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4737 Alan Hayward <alan.hayward@arm.com>
4738 David Sherwood <david.sherwood@arm.com>
4740 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4742 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4743 Alan Hayward <alan.hayward@arm.com>
4744 David Sherwood <david.sherwood@arm.com>
4746 * doc/sourcebuild.texi (vect_variable_length): Document.
4748 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4749 Alan Hayward <alan.hayward@arm.com>
4750 David Sherwood <david.sherwood@arm.com>
4752 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4754 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4755 Alan Hayward <alan.hayward@arm.com>
4756 David Sherwood <david.sherwood@arm.com>
4758 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4760 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4761 Alan Hayward <alan.hayward@arm.com>
4762 David Sherwood <david.sherwood@arm.com>
4764 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4765 previously undocumented selectors.
4766 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4768 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4770 * doc/rtl.texi (const_vector): Say that elements can be
4771 const_wide_ints too.
4772 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4773 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4774 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4775 * optabs.c (expand_vector_broadcast): Likewise.
4777 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4778 Alan Hayward <alan.hayward@arm.com>
4779 David Sherwood <david.sherwood@arm.com>
4781 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4782 scaled index even if the unscaled address was invalid.
4783 Don't increase the complexity of using a scale in that case.
4785 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4786 Alan Hayward <alan.hayward@arm.com>
4787 David Sherwood <david.sherwood@arm.com>
4789 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4790 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4791 UNITS_PER_WORD bytes.
4792 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4793 based on REGMODE_NATURAL_SIZE of the inner mode.
4794 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4795 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4796 * expmed.c (lowpart_bit_field_p): Divide the value up into
4797 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4798 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4799 whether something is likely to occupy more than one register.
4801 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4804 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4805 profile_count::adjust_for_ipa_scaling.
4806 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4807 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4809 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4811 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4813 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4814 (split_group): Add total_orig and total_new arguments, estimate the
4815 number of statements related to the store group without store merging
4816 and with store merging.
4817 (imm_store_chain_info::output_merged_store): Adjust split_group
4818 callers, punt if estimated number of statements with store merging
4819 is not smaller than estimated number of statements without it.
4821 (handled_load): Remove has_single_use checks.
4822 (pass_store_merging::process_store): Likewise.
4824 2017-11-09 Richard Biener <rguenther@suse.de>
4826 PR tree-optimization/82902
4827 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4829 2017-11-09 Martin Liska <mliska@suse.cz>
4832 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4835 2017-11-09 Martin Liska <mliska@suse.cz>
4837 PR tree-optimization/82669
4838 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4840 2017-11-09 Martin Liska <mliska@suse.cz>
4842 PR gcov-profile/48463
4843 * coverage.c (coverage_begin_function): Output also end locus
4844 of a function and information whether the function is
4846 * gcov-dump.c (tag_function): Parse and print the information.
4847 * gcov.c (INCLUDE_MAP): Add include.
4848 (INCLUDE_SET): Likewise.
4849 (struct line_info): Move earlier in the source file because
4850 of vector<line_info> in function_info structure.
4851 (line_info::line_info): Likewise.
4852 (line_info::has_block): Likewise.
4853 (struct source_info): Add new member index.
4854 (source_info::get_functions_at_location): New function.
4855 (function_info::group_line_p): New function.
4856 (output_intermediate_line): New function.
4857 (output_intermediate_file): Use the mentioned function.
4858 (struct function_start): New.
4859 (struct function_start_pair_hash): Likewise.
4860 (process_file): Add code that identifies group functions.
4861 Assign lines either to global or function scope.
4862 (generate_results): Skip artificial functions.
4863 (find_source): Assign index for each source file.
4864 (read_graph_file): Read new flag artificial and end_line.
4865 (add_line_counts): Assign it either to global of function scope.
4866 (accumulate_line_counts): Isolate core of the function to
4867 accumulate_line_info and call it for both function and global
4869 (accumulate_line_info): New function.
4870 (output_line_beginning): Fix GNU coding style.
4871 (print_source_line): New function.
4872 (output_line_details): Likewise.
4873 (output_function_details): Likewise.
4874 (output_lines): Iterate both source (global) scope and function
4876 (struct function_line_start_cmp): New class.
4877 * doc/gcov.texi: Reflect changes in documentation.
4879 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4882 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4883 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4884 and similarly for not instead of neg.
4886 2017-11-08 Andi Kleen <ak@linux.intel.com>
4888 * config/i386/i386.opt: Add -mforce-indirect-call.
4889 * config/i386/predicates.md: Check for flag_force_indirect_call.
4890 * doc/invoke.texi: Document -mforce-indirect-call
4892 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4894 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4896 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4897 config/riscv/riscv.c (predict.h): New include.
4898 (riscv_slow_unaligned_access_p): No longer static.
4899 (riscv_block_move_straight): Add require.
4900 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4902 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4905 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4906 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4907 nonimmediate_operand predicate for operand 1 instead of
4910 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4912 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4914 * config/aarch64/constraints.md (Uml): New constraint.
4915 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4918 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4920 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4921 of two vec_duplicates into a vec_concat.
4923 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4925 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4926 Simplify vec_merge of vec_duplicate and vec_concat.
4927 * config/aarch64/constraints.md (Utq): New constraint.
4928 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4931 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4933 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4934 Simplify vec_merge of vec_duplicate and const_vector.
4935 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4937 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4938 mode iterator. Update predicate on operand 1 to
4939 handle non-const_vec constants. Delete constraints.
4940 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4942 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4944 PR tree-optimization/78821
4945 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4947 (store_operand_info::store_operand_info): Initialize it to false.
4948 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4949 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4950 store in the group, and if chain_info is non-NULL, to ignore altogether
4952 (compatible_load_p): Fail if bit_not_p does not match.
4953 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4954 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4955 (pass_store_merging::process_store): Adjust
4956 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4957 call terminate_all_aliasing_chains newly when adding a store into
4958 a chain with non-NULL chain_info.
4960 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4962 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4964 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4966 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4968 (aarch64_layout_frame): Initialise emit_frame_chain.
4969 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4970 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4972 2017-11-08 Martin Liska <mliska@suse.cz>
4974 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4975 of gimple_call_internal_p.
4977 2017-11-07 Tom de Vries <tom@codesourcery.com>
4979 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4981 2017-11-07 Tom de Vries <tom@codesourcery.com>
4983 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4986 2017-11-08 Martin Liska <mliska@suse.cz>
4989 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4991 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4993 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4995 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4998 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4999 (*cmp<mode>_ccz_1): New insn with $k alternative.
5002 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5003 RTX_COMM_COMPARE as commutative as well.
5004 (ix86_binary_operator_ok): Formatting fix.
5005 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5006 *<code><mode>3<mask_name><round_saeonly_name>,
5007 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5008 *<s>mul<mode>3_highpart<mask_name>,
5009 *vec_widen_umult_even_v16si<mask_name>,
5010 *vec_widen_umult_even_v8si<mask_name>,
5011 *vec_widen_umult_even_v4si<mask_name>,
5012 *vec_widen_smult_even_v16si<mask_name>,
5013 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5014 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5015 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5016 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5017 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5018 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5019 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5020 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5021 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5022 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5023 ix86_binary_operator_ok. Formatting fixes.
5024 (*<plusminus_insn><mode>3<mask_name><round_name>,
5025 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5028 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5030 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5031 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5032 eq and ne if TARGET_ISEL.
5033 (cmp): New code_iterator.
5034 (UNS, UNSU_, UNSIK): New code_attrs.
5035 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5036 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5038 ("eq<mode>3"): ... this.
5039 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5041 ("ne<mode>3"): ... this.
5043 2017-11-07 Julia Koval <julia.koval@intel.com>
5046 * common/config/i386/i386-common.c
5047 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5048 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5049 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5050 * config/i386/i386.opt: Ditto.
5051 * config/i386/i386.c (ix86_target_string): Ditto.
5052 (ix86_option_override_internal): Ditto.
5053 (ix86_init_mpx_builtins): Move MPX to args2.
5054 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5055 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5056 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5057 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5058 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5059 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5060 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5062 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5065 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5066 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5067 (zero-extendsidi peephole2): Remove peephole.
5069 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5072 * stor-layout.c: Include attribs.h.
5073 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5074 explicit lookup of "aligned" attribute.
5076 2017-11-07 Andrew Waterman <andrew@sifive.com>
5078 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5079 (riscv_expand_block_move): Likewise.
5080 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5082 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5083 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5084 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5085 (riscv_adjust_block_mem): Likewise.
5086 (riscv_block_move_loop): Likewise.
5087 (riscv_expand_block_move): Likewise.
5088 * config/riscv/riscv.md (movmemsi): New pattern.
5090 2017-11-07 Michael Clark <michaeljclark@mac.com>
5092 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5093 (MUSL_DYNAMIC_LINKER): Likewise.
5095 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5097 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5098 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5101 2017-11-07 Richard Biener <rguenther@suse.de>
5103 * match.pd: Fix build.
5105 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5106 Jackson Woodruff <jackson.woodruff@arm.com>
5108 PR tree-optimization/71026
5109 * match.pd: Canonicalize negate in division.
5111 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5114 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5116 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5118 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5119 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5121 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5123 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5124 non-scalar integral types.
5125 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5126 (-(A-B), -(~A)): New transformations.
5128 2017-11-07 Tom de Vries <tom@codesourcery.com>
5130 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5131 semicolon after "do {} while (0)".
5132 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5133 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5134 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5135 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5136 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5138 2017-11-07 Tom de Vries <tom@codesourcery.com>
5140 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5141 after "do {} while (0)".
5142 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5143 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5144 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5145 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5146 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5148 2017-11-07 Tom de Vries <tom@codesourcery.com>
5151 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5153 (arm_rtx_costs_internal): Add missing semicolon after
5154 HANDLE_NARROW_SHIFT_ARITH call.
5156 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5158 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5159 disable isel if it was not set explicitly.
5161 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5163 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5164 (add_type_attribute) likewise.
5166 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5168 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5169 of struct ix86_frame.
5170 (ix86_initial_elimination_offset): Likewise.
5171 (ix86_expand_split_stack_prologue): Likewise.
5173 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5175 * tree-vrp.h (enum value_range_type): Update stale comment.
5177 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5178 Alan Hayward <alan.hayward@arm.com>
5179 David Sherwood <david.sherwood@arm.com>
5181 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5182 (aarch64_expand_vec_perm_const): Take the number of units too.
5183 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5184 (aarch64_expand_vec_perm_const): Likewise.
5185 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5186 (vec_perm<mode>): Update accordingly.
5188 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5189 Alan Hayward <alan.hayward@arm.com>
5190 David Sherwood <david.sherwood@arm.com>
5192 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5193 Take the number of units too.
5194 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5195 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5196 but check for a vector mode before rather than after the call.
5197 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5198 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5199 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5200 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5201 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5202 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5203 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5204 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5205 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5206 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5207 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5208 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5209 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5210 (aarch64_sqdmull2_n<mode>): Update accordingly.
5212 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5213 Alan Hayward <alan.hayward@arm.com>
5214 David Sherwood <david.sherwood@arm.com>
5216 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5217 the number of units too.
5218 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5219 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5220 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5221 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5222 (vec_store_lanesxi<mode>): Update accordingly.
5224 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5225 Alan Hayward <alan.hayward@arm.com>
5226 David Sherwood <david.sherwood@arm.com>
5228 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5229 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5230 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5231 of units rather than the mode.
5232 * config/aarch64/iterators.md (nunits): New mode attribute.
5233 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5234 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5235 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5236 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5237 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5238 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5239 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5240 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5241 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5242 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5243 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5244 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5245 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5246 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5247 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5248 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5249 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5250 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5251 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5252 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5253 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5254 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5255 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5256 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5257 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5258 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5259 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5260 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5261 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5262 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5263 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5264 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5265 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5266 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5267 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5268 (aarch64_simd_vec_setv2di): Likewise.
5270 2017-11-06 Carl Love <cel@us.ibm.com>
5272 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5274 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5275 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5277 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5279 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5280 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5281 Add power 8 macro expansions.
5282 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5283 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5284 power 8 instructions. (VSX_XXBR): Add iterator.
5286 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5288 * config/arm/arm.md (predicable_short_it): Change default to "no",
5289 improve documentation, remove uses that are identical to the default.
5290 (enabled_for_depr_it): Rename to enabled_for_short_it.
5291 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5292 Remove default uses.
5293 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5294 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5295 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5296 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5298 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5301 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5302 float128 helper macros, which are no longer used after deleting
5303 the old 'q' built-in functions, and moving the round to odd
5304 built-in functions to being special built-in functions.
5305 (BU_FLOAT128_2): Likewise.
5306 (BU_FLOAT128_1_HW): Likewise.
5307 (BU_FLOAT128_2_HW): Likewise.
5308 (BU_FLOAT128_3_HW): Likewise.
5309 (FABSQ): Delete old 'q' built-in functions.
5310 (COPYSIGNQ): Likewise.
5311 (SQRTF128_ODD): Move round to odd built-in functions to be
5312 special built-in functions, so that we can handle
5313 -mabi=ieeelongdouble.
5314 (TRUNCF128_ODD): Likewise.
5315 (ADDF128_ODD): Likewise.
5316 (SUBF128_ODD): Likewise.
5317 (MULF128_ODD): Likewise.
5318 (DIVF128_ODD): Likewise.
5319 (FMAF128_ODD): Likewise.
5320 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5321 built-in names to 'f128'.
5322 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5323 old 'q' built-in functions, as the machine independent code for
5324 'f128' built-in functions handles this.
5325 (rs6000_expand_builtin): Add expansion for float128 round to odd
5326 functions, keying off on -mabi=ieeelongdouble of whether to use
5327 the KFmode or TFmode variant.
5328 (rs6000_init_builtins): Initialize the _Float128 round to odd
5330 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5331 _Float128 'q' built-in functions are now mapped into the new
5332 'f128' built-in functions.
5334 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5336 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5337 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5339 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5342 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5343 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5344 * toplev.c: Include "ipa-fnsummary.h".
5345 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5347 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5349 PR tree-optimization/82838
5350 * gimple-ssa-store-merging.c
5351 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5352 on a separate gimple_seq which is then appended to seq.
5354 2017-11-06 Jeff Law <law@redhat.com>
5357 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5358 (get_probe_interval): New functions.
5359 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5360 (ix86_adjust_stack_and_probe): Likewise.
5361 (output_adjust_stack_and_probe): Likewise.
5362 (ix86_emit_probe_stack_range): Likewise.
5363 (ix86_expand_prologue): Likewise.
5365 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5367 PR tree-optimization/82816
5368 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5369 if the modes of the two types are the same.
5370 (convert_plusminus_to_widen): Likewise.
5372 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5374 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5376 (usadv16qi): New define_expand.
5377 (usadv8hi): New define_expand.
5379 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5382 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5384 (inline_transform): Likewise.
5385 * predict.c (counts_to_freqs): Remove useless conditional.
5386 * profile-count.h (profile_count::apply_scale): Move sanity check.
5387 * tree-inline.c (copy_bb): Always scale.
5388 (copy_cfg_body): Likewise.
5390 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5393 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5395 (*cmp_ite0): Add enabled_for_depr_it attribute.
5396 (*cmp_ite1): Likewise.
5398 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5400 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5403 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5405 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5406 return the value by pointer.
5407 (extract_range_from_multiplicative_op_1): Update accordingly.
5408 Return as soon as an operation fails.
5410 2017-11-05 Tom de Vries <tom@codesourcery.com>
5413 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5414 (DEF_SANITIZER_BUILTIN): ... here.
5415 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5416 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5418 2017-11-05 Tom de Vries <tom@codesourcery.com>
5421 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5423 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5424 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5425 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5427 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5428 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5429 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5431 2017-11-05 Tom de Vries <tom@codesourcery.com>
5434 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5437 2017-11-04 Michael Clark <michaeljclark@mac.com>
5439 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5440 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5442 (*addsi3_extended): Likewise.
5443 (*addsi3_extended2): Likewise.
5444 (<optab>si3): Likewise.
5445 (<optab>di3): Likewise.
5446 (<optab><mode>3): Likewise.
5447 (<*optabe>si3_internal): Likewise.
5448 (zero_extendqi<SUPERQI:mode>2): Likewise.
5449 (*add<mode>hi3): Likewise.
5450 (*xor<mode>hi3): Likewise.
5451 (<optab>di3): Likewise.
5452 (*<optab>si3_extend): Likewise.
5453 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5454 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5455 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5457 2017-11-04 Andrew Waterman <andrew@sifive.com>
5459 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5460 TARGET_STRICT_ALIGN based upon -mtune argument.
5462 2017-11-04 Andrew Waterman <andrew@sifive.com>
5464 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5466 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5468 * config/i386/i386.c (choose_basereg): Use optional scratch
5469 register and add assertion.
5470 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5471 needed, and don't allocate stack.
5472 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5473 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5474 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5476 2017-11-03 Jeff Law <law@redhat.com>
5478 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5479 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5480 to probe at the start of a noreturn function.
5482 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5484 PR tree-optimization/78821
5485 * gimple-ssa-store-merging.c: Update the file comment.
5486 (MAX_STORE_ALIAS_CHECKS): Define.
5487 (struct store_operand_info): New type.
5488 (store_operand_info::store_operand_info): New constructor.
5489 (struct store_immediate_info): Add rhs_code and ops data members.
5490 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5491 and op1r arguments to the ctor, initialize corresponding data members.
5492 (struct merged_store_group): Add load_align_base and load_align
5494 (merged_store_group::merged_store_group): Initialize them.
5495 (merged_store_group::do_merge): Update them.
5496 (merged_store_group::apply_stores): Pick the constant for
5497 encode_tree_to_bitpos from one of the two operands, or skip
5498 encode_tree_to_bitpos if neither operand is a constant.
5499 (class pass_store_merging): Add process_store method decl. Remove
5500 bool argument from terminate_all_aliasing_chains method decl.
5501 (pass_store_merging::terminate_all_aliasing_chains): Remove
5502 var_offset_p argument and corresponding handling.
5503 (stmts_may_clobber_ref_p): New function.
5504 (compatible_load_p): New function.
5505 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5506 if there is overlap and rhs_code is not INTEGER_CST. For
5507 non-overlapping stores terminate group if rhs is not mergeable.
5508 (get_alias_type_for_stmts): Change first argument from
5509 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5510 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5511 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5513 (get_location_for_stmts): Change first argument from
5514 auto_vec<gimple *> & to vec<gimple *> &.
5515 (struct split_store): Remove orig_stmts data member, add orig_stores.
5516 (split_store::split_store): Create orig_stores rather than orig_stmts.
5517 (find_constituent_stmts): Renamed to ...
5518 (find_constituent_stores): ... this. Change second argument from
5519 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5520 to info structures rather than the statements.
5521 (split_group): Rename ALLOW_UNALIGNED argument to
5522 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5523 it. Adjust find_constituent_stores caller.
5524 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5525 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5526 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5527 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5528 (mem_valid_for_store_merging): New function.
5529 (handled_load): New function.
5530 (pass_store_merging::process_store): New method.
5531 (pass_store_merging::execute): Use process_store method. Adjust
5532 terminate_all_aliasing_chains caller.
5534 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5536 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5537 Return true for more constants, symbols and label references.
5538 (aarch64_valid_floating_const): Remove unused function.
5540 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5543 * config/aarch64/aarch64.c (aarch64_layout_frame):
5544 Undo forcing of LR at bottom of frame.
5546 2017-11-03 Jeff Law <law@redhat.com>
5549 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5550 for int_registers_saved.
5552 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5553 extracted from tree-ssa-dom.c.
5554 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5555 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5556 (record_equivalences_from_incoming_edge): Add additional argument
5557 to single_pred_edge_ignoring_loop_edges call.
5558 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5559 (uncprop_dom_walker::before_dom_children): Add additional argument
5560 to single_pred_edge_ignoring_loop_edges call.
5561 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5562 single_pred_edge_ignoring_loop_edges rather than open coding.
5563 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5565 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5567 * match.pd (-(-A)): Rewrite.
5569 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5571 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5572 (rs6000_emit_int_cmove): New declaration.
5573 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5574 (rs6000_emit_sISEL): Delete.
5575 (rs6000_emit_int_cmove): Make non-static.
5576 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5577 instead of rs6000_emit_sISEL.
5579 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5581 * asan.c (create_cond_insert_point): Maintain profile.
5582 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5584 * basic-block.h (struct basic_block_def): Remove frequency.
5585 (EDGE_FREQUENCY): Use to_frequency
5586 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5588 (find_traces): Update to use to_frequency.
5589 (find_traces_1_round): Likewise; use only IPA counts.
5590 (bb_to_key): Likewise.
5591 (connect_traces): Use IPA counts only.
5592 (copy_bb_p): Update to use to_frequency.
5593 (fix_up_crossing_landing_pad): Likewise.
5594 (sanitize_hot_paths): Likewise.
5595 * bt-load.c (basic_block_freq): Likewise.
5596 * cfg.c (init_flow): Set count_max to uninitialized.
5597 (check_bb_profile): Remove frequencies; check counts.
5598 (dump_bb_info): Do not dump frequencies.
5599 (update_bb_profile_for_threading): Update counts only.
5600 (scale_bbs_frequencies_int): Likewise.
5601 (MAX_SAFE_MULTIPLIER): Remove.
5602 (scale_bbs_frequencies_gcov_type): Update counts only.
5603 (scale_bbs_frequencies_profile_count): Update counts only.
5604 (scale_bbs_frequencies): Update counts only.
5605 * cfg.h (struct control_flow_graph): Add count-max.
5606 (update_bb_profile_for_threading): Update prototype.
5607 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5608 (find_many_sub_basic_blocks): Likewise.
5609 * cfgcleanup.c (try_forward_edges): Likewise.
5610 (try_crossjump_to_edge): Likewise.
5611 * cfgexpand.c (expand_gimple_cond): Likewise.
5612 (expand_gimple_tailcall): Likewise.
5613 (construct_init_block): Likewise.
5614 (construct_exit_block): Likewise.
5615 * cfghooks.c (verify_flow_info): Check consistency of counts.
5616 (dump_bb_for_graph): Do not dump frequencies.
5617 (split_block_1): Do not update frequencies.
5618 (split_edge): Do not update frequencies.
5619 (make_forwarder_block): Do not update frequencies.
5620 (duplicate_block): Do not update frequencies.
5621 (account_profile_record): Do not update frequencies.
5622 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5623 for global heuristics.
5624 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5625 (expected_loop_iterations_unbounded): Use counts only.
5626 * cfgloopmanip.c (scale_loop_profile): Simplify.
5627 (create_empty_loop_on_edge): Simplify
5629 (duplicate_loop_to_header_edge): Simplify
5630 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5631 (update_br_prob_note): Take care of removing note when profile
5633 (relink_block_chain): Do not dump frequency.
5634 (rtl_account_profile_record): Use to_frequency.
5635 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5636 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5637 (cgraph_update_edges_for_call_stmt_node): Likewise.
5638 (cgraph_edge::verify_count_and_frequency): Update.
5639 (cgraph_node::verify_node): Temporarily disable frequency verification.
5640 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5641 to_cgraph_frequency.
5642 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5643 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5645 (cgraph_node::expand_thunk): Update profile.
5646 * except.c (dw2_build_landing_pads): Do not update frequency.
5647 * final.c (compute_alignments): Use to_frequency.
5648 (dump_basic_block_info): Do not dump frequency.
5649 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5650 (dump_gimple_bb_header): Do not dump frequency.
5651 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5653 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5654 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5655 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5656 (init_before_recovery): Do not update frequency.
5657 (sched_create_recovery_edges): Do not update frequency.
5658 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5659 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5660 (ipa_cp_c_finalize): Set max_count to uninitialized.
5661 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5662 (param_change_prob): Use counts.
5663 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5665 * ipa-split.c (consider_split): Use to_frequency.
5666 (split_function): Use to_frequency.
5667 * ira-build.c (loop_compare_func): Likewise.
5668 (mark_loops_for_removal): Likewise.
5669 (mark_all_loops_for_removal): Likewise.
5670 * loop-doloop.c (doloop_modify): Do not update frequency.
5671 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5673 * lto-streamer-in.c (input_function): Update count_max.
5674 * omp-expand.c (expand_omp_taskreg): Update count_max.
5675 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5676 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5677 (maybe_hot_count_p): Use ipa counts only.
5678 (maybe_hot_bb_p): Simplify.
5679 (maybe_hot_edge_p): Simplify.
5680 (probably_never_executed): Do not take frequency argument.
5681 (probably_never_executed_bb_p): Do not pass frequency.
5682 (probably_never_executed_edge_p): Likewise.
5683 (combine_predictions_for_bb): Check that profile is nonzero.
5684 (propagate_freq): Do not set frequency.
5685 (drop_profile): Simplify.
5686 (counts_to_freqs): Simplify.
5687 (expensive_function_p): Use to_frequency.
5688 (propagate_unlikely_bbs_forward): Simplify.
5689 (determine_unlikely_bbs): Simplify.
5690 (estimate_bb_frequencies): Add hack to silence graphite issues.
5691 (compute_function_frequency): Use ipa counts.
5692 (pass_profile::execute): Update.
5693 (rebuild_frequencies): Use counts only.
5694 (force_edge_cold): Use counts only.
5695 * profile-count.c (profile_count::dump): Dump new count types.
5696 (profile_count::differs_from_p): Check compatiblity.
5697 (profile_count::to_frequency): New function.
5698 (profile_count::to_cgraph_frequency): New function.
5699 * profile-count.h (struct function): Declare.
5700 (enum profile_quality): Add profile_guessed_local and
5701 profile_guessed_global0.
5702 (class profile_proability): Decrease number of bits to 29;
5703 update from_reg_br_prob_note and to_reg_br_prob_note.
5704 (class profile_count: Update comment; decrease number of bits
5705 to 61. Check compatibility.
5706 (profile_count::compatible_p): New private member function.
5707 (profile_count::ipa_p): New member function.
5708 (profile_count::operator<): Handle global zero correctly.
5709 (profile_count::operator>): Handle global zero correctly.
5710 (profile_count::operator<=): Handle global zero correctly.
5711 (profile_count::operator>=): Handle global zero correctly.
5712 (profile_count::nonzero_p): New member function.
5713 (profile_count::force_nonzero): New member function.
5714 (profile_count::max): New member function.
5715 (profile_count::apply_scale): Handle IPA scalling.
5716 (profile_count::guessed_local): New member function.
5717 (profile_count::global0): New member function.
5718 (profile_count::ipa): New member function.
5719 (profile_count::to_frequency): Declare.
5720 (profile_count::to_cgraph_frequency): Declare.
5721 * profile.c (OVERLAP_BASE): Delete.
5722 (compute_frequency_overlap): Delete.
5723 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5724 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5725 * sched-ebb.c (rank): Use counts only.
5726 * shrink-wrap.c (handle_simple_exit): Use counts only.
5727 (try_shrink_wrapping): Use counts only.
5728 (place_prologue_for_one_component): Use counts only.
5729 * tracer.c (find_best_predecessor): Use to_frequency.
5730 (find_trace): Use to_frequency.
5731 (tail_duplicate): Use to_frequency.
5732 * trans-mem.c (expand_transaction): Do not update frequency.
5733 * tree-call-cdce.c: Do not update frequency.
5734 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5735 (gimple_merge_blocks): Likewise.
5736 (gimple_split_edge): Likewise.
5737 (gimple_duplicate_sese_region): Likewise.
5738 (gimple_duplicate_sese_tail): Likewise.
5739 (move_sese_region_to_fn): Likewise.
5740 (gimple_account_profile_record): Likewise.
5741 (insert_cond_bb): Likewise.
5742 * tree-complex.c (expand_complex_div_wide): Likewise.
5743 * tree-eh.c (lower_resx): Update profile.
5744 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5746 (initialize_cfun): Do not initialize frequencies
5747 (freqs_to_counts): Delete.
5748 (copy_cfg_body): Ignore count parameter.
5749 (copy_body): Update.
5750 (expand_call_inline): Update count_max.
5751 (optimize_inline_calls): Update count_max.
5752 (tree_function_versioning): Update count_max.
5753 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5754 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5756 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5757 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5758 (try_peel_loop): Likewise.
5759 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5761 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5762 (tree_transform_and_unroll_loop): Do not use frequencies
5763 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5764 Use reliable prediction only.
5765 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5766 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5767 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5768 probability scaling.
5769 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5770 not update frequency
5771 (any_remaining_duplicated_blocks): Likewise.
5772 (update_profile): Likewise.
5773 (estimated_freqs_path): Delete.
5774 (freqs_to_counts_path): Delete.
5775 (clear_counts_path): Delete.
5776 (ssa_fix_duplicate_block_edges): Likewise.
5777 (duplicate_thread_path): Likewise.
5778 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5779 * tree-tailcall.c (decrease_profile): Do not update frequency.
5780 (eliminate_tail_call): Likewise.
5781 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5782 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5783 (optimize_mask_stores): Likewise.
5784 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5785 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5786 (ubsan_expand_ptr_ifn): Update profile.
5787 * value-prof.c (gimple_ic): Simplify.
5788 * value-prof.h (gimple_ic): Update prototype.
5789 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5790 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5792 (want_inline_self_recursive_call_p): Likewise.
5793 (resolve_noninline_speculation): Only cummulate defined counts.
5794 (inline_small_functions): Use nonzero_p.
5795 (ipa_inline): Do not access freed node.
5797 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5799 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5800 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5802 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5804 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5805 non-legitimate address.
5807 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5809 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5810 (*lt0_<mode>di, *lt0_<mode>si): New.
5812 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5814 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5815 TARGET_PAIRED_FLOAT.
5817 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5818 Jim Wilson <jim.wilson@linaro.org>
5820 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5821 * config/aarch64/aarch64-tune.md: Regenerated.
5822 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5823 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5825 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5827 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5828 (arc_expand_prologue): Restore blink for millicode.
5829 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5831 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5834 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5835 gen_vec_duplicate after forcing the scalar into a register.
5837 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5839 * combine (try_combine): Print the insns input to try_combine to the
5842 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5845 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5846 Remove second argument from aarch64_process_target_attr call.
5847 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5849 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5851 (aarch64_handle_attr_arch): Remove second argument.
5852 (aarch64_handle_attr_cpu): Ditto.
5853 (aarch64_handle_attr_tune): Ditto.
5854 (aarch64_handle_attr_isa_flags): Ditto.
5855 (aarch64_process_one_target_attr): Ditto.
5856 (aarch64_process_target_attr): Ditto.
5857 (aarch64_option_valid_attribute_p): Remove second argument.
5858 on aarch64_process_target_attr call.
5860 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5862 * diagnostic.c: Include "selftest-diagnostic.h".
5863 (selftest::assert_location_text): New function.
5864 (selftest::test_diagnostic_get_location_text): New function.
5865 (selftest::diagnostic_c_tests): Call it.
5867 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5869 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5870 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5871 (class selftest::test_diagnostic_context): Move to...
5872 * selftest-diagnostic.c: New file.
5873 * selftest-diagnostic.h: New file.
5875 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5877 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5878 offset range for FT32B.
5879 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5880 * config/ft32/ft32.md: Add TARGET_NOPM.
5881 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5882 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5884 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5886 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5888 2017-11-02 Jeff Law <law@redhat.com>
5890 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5891 virtual keyword on FINAL OVERRIDE members.
5893 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5894 virtuals together. Add virtual destructor.
5895 (substitute_and_fold_engine): Similarly.
5897 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5899 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5901 2017-11-02 Richard Biener <rguenther@suse.de>
5903 PR tree-optimization/82795
5904 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5906 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5908 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5910 (gcc_GAS_CHECK_FEATURE): Remove.
5911 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5912 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5913 * configure: Regenerate.
5915 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5917 * config/arc/arc.c (hwloop_optimize): Account for empty
5920 2017-11-02 Richard Biener <rguenther@suse.de>
5923 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5924 Truncate ARRAY_REF index and element size.
5926 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5928 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5930 2017-11-01 Jeff Law <law@redhat.com>
5932 * tree-ssa-ccp.c (ccp_folder): New class derived from
5933 substitute_and_fold_engine.
5934 (ccp_folder::get_value): New member function.
5935 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5936 (ccp_fold_stmt): Remove prototype.
5937 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5938 * tree-ssa-copy.c (copy_folder): New class derived from
5939 substitute_and_fold_engine.
5940 (copy_folder::get_value): Renamed from get_value.
5941 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5942 * tree-vrp.c (vrp_folder): New class derived from
5943 substitute_and_fold_engine.
5944 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5945 (vrp_folder::get_value): New member function.
5946 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5947 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5948 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5949 provide a class interface to folder/substitute routines.
5950 (ssa_prop_fold_stmt_fn): Remove typedef.
5951 (ssa_prop_get_value_fn): Likewise.
5952 (subsitute_and_fold): Remove prototype.
5953 (replace_uses_in): Likewise.
5954 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5955 Renamed from replace_uses_in. Call the virtual member function
5956 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5957 (substitute_and_fold_dom_walker): Remove initialization of
5958 data member entries for calbacks. Add substitute_and_fold_engine
5959 member and initialize it.
5960 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5961 member functions for get_value, replace_phi_args_in c
5962 replace_uses_in, and fold_stmt calls.
5963 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5964 substitute_and_fold. Remove assert. Update ctor call.
5966 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5967 (ssa_prop_visit_phi_fn): Likewise.
5968 (class ssa_propagation_engine): New class to provide an interface
5970 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5972 (ssa_prop_visit_phi): Likewise.
5973 (ssa_propagation_engine::simulate_stmt): Moved into class.
5974 Call visit_phi/visit_stmt from the class rather than via
5975 file scoped static variables.
5976 (ssa_propagation_engine::simulate_block): Moved into class.
5977 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5978 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5979 set file scoped statics for the visit_stmt/visit_phi callbacks.
5980 * tree-complex.c (complex_propagate): New class derived from
5981 ssa_propagation_engine.
5982 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5983 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5984 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5986 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5987 ssa_propagation_engine.
5988 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5989 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5990 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5991 * tree-ssa-copy.c (copy_prop): New class derived from
5992 ssa_propagation_engine.
5993 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5994 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5995 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5996 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5997 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5998 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5999 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6001 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6003 PR rtl-optimization/82778
6004 PR rtl-optimization/82597
6005 * compare-elim.c (struct comparison): Add in_a_setter field.
6006 (find_comparison_dom_walker::before_dom_children): Remove killed
6007 bitmap and df_simulate_find_defs call, instead walk the defs.
6008 Compute last_setter and initialize in_a_setter. Merge definitions
6009 with first initialization for a few variables.
6010 (try_validate_parallel): Use insn_invalid_p instead of
6011 recog_memoized. Return insn rather than just the pattern.
6012 (try_merge_compare): Fix up comment. Don't uselessly test if
6013 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6015 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6018 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6019 Alan Hayward <alan.hayward@arm.com>
6020 David Sherwood <david.sherwood@arm.com>
6022 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6023 aarch64_hard_regno_nregs to get the number of registers
6026 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6027 Alan Hayward <alan.hayward@arm.com>
6028 David Sherwood <david.sherwood@arm.com>
6030 * config/aarch64/constraints.md (Upl): Rename to...
6032 * config/aarch64/aarch64.md
6033 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6036 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6037 Alan Hayward <alan.hayward@arm.com>
6038 David Sherwood <david.sherwood@arm.com>
6040 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6041 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6044 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6045 Alan Hayward <alan.hayward@arm.com>
6046 David Sherwood <david.sherwood@arm.com>
6048 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6049 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6050 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6052 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6053 of a single element.
6054 * config/aarch64/iterators.md: Add a comment above the permute
6055 unspecs to say that they are generated directly by
6056 aarch64_expand_vec_perm_const.
6057 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6059 2017-11-01 Nathan Sidwell <nathan@acm.org>
6061 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6063 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6065 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6066 and medany code models, and describe what they do.
6068 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6070 Revert accidental duplicate:
6072 * combine.c (can_change_dest_mode): Reject changes in
6073 REGMODE_NATURAL_SIZE.
6075 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6077 PR rtl-optimization/64682
6078 PR rtl-optimization/69567
6079 PR rtl-optimization/69737
6080 PR rtl-optimization/82683
6081 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6082 register mentioned in the note, drop the note, unless it came from I3,
6083 in which case it should go to I3 again.
6085 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6087 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6088 and return false if not.
6089 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6091 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6093 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6094 range is known to be empty.
6096 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6097 Alan Hayward <alan.hayward@arm.com>
6098 David Sherwood <david.sherwood@arm.com>
6100 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6101 and CONST_VECTOR_NUNITS instead of computing the number of units from
6102 the byte sizes of the vector and element.
6103 (simplify_binary_operation_1): Likewise.
6104 (simplify_const_binary_operation): Likewise.
6105 (simplify_ternary_operation): Likewise.
6107 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6108 Alan Hayward <alan.hayward@arm.com>
6109 David Sherwood <david.sherwood@arm.com>
6111 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6112 (int_mem_offset): ...this new function.
6113 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6114 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6115 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6118 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6119 Alan Hayward <alan.hayward@arm.com>
6120 David Sherwood <david.sherwood@arm.com>
6122 * lower-subreg.c (interesting_mode_p): New function.
6123 (compute_costs, find_decomposable_subregs, decompose_register)
6124 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6125 (resolve_clobber, dump_choices): Use it.
6127 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6128 Alan Hayward <alan.hayward@arm.com>
6129 David Sherwood <david.sherwood@arm.com>
6131 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6133 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6134 Alan Hayward <alan.hayward@arm.com>
6135 David Sherwood <david.sherwood@arm.com>
6137 * alias.c (find_base_value, find_base_term): Only process integer
6138 truncations. Check the precision rather than the size.
6140 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6141 Alan Hayward <alan.hayward@arm.com>
6142 David Sherwood <david.sherwood@arm.com>
6144 * machmode.h (is_narrower_int_mode): New function
6145 * optabs.c (expand_float, expand_fix): Use it.
6146 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6148 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6149 Alan Hayward <alan.hayward@arm.com>
6150 David Sherwood <david.sherwood@arm.com>
6152 * rtl.h (narrower_subreg_mode): New function.
6153 * ira-color.c (update_costs_from_allocno): Use it.
6155 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6156 Alan Hayward <alan.hayward@arm.com>
6157 David Sherwood <david.sherwood@arm.com>
6159 * optabs-query.h (convert_optab_p): New function, split out from...
6160 (convert_optab_handler): ...here.
6161 (widening_optab_handler): Delete.
6162 (find_widening_optab_handler): Remove permit_non_widening parameter.
6163 (find_widening_optab_handler_and_mode): Likewise. Provide an
6164 override that operates on mode class wrappers.
6165 * optabs-query.c (widening_optab_handler): Delete.
6166 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6167 parameter. Assert that the two modes are the same class and that
6168 the "from" mode is narrower than the "to" mode. Use
6169 convert_optab_handler instead of widening_optab_handler.
6170 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6171 instead of widening_optab_handler.
6172 * expr.c (expand_expr_real_2): Update calls to
6173 find_widening_optab_handler.
6174 * optabs.c (expand_widen_pattern_expr): Likewise.
6175 (expand_binop_directly): Take the insn_code as a parameter.
6176 (expand_binop): Only call find_widening_optab_handler for
6177 conversion optabs; use optab_handler otherwise. Update calls
6178 to find_widening_optab_handler and expand_binop_directly.
6179 Use convert_optab_handler instead of widening_optab_handler.
6180 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6181 find_widening_optab_handler and use scalar_mode rather than
6183 (convert_plusminus_to_widen): Likewise.
6185 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6186 Alan Hayward <alan.hayward@arm.com>
6187 David Sherwood <david.sherwood@arm.com>
6189 * machmode.h (fixed_size_mode): New class.
6190 * rtl.h (get_pool_mode): Return fixed_size_mode.
6191 * gengtype.c (main): Add fixed_size_mode.
6192 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6193 (get_raw_arg_mode): Likewise.
6194 * doc/tm.texi: Regenerate.
6195 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6196 * targhooks.c (default_get_reg_raw_mode): Likewise.
6197 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6198 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6199 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6200 (msp430_get_raw_result_mode): Likewise.
6201 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6202 * dbxout.c (dbxout_parms): Require fixed-size modes.
6203 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6204 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6205 * omp-low.c (lower_oacc_reductions): Likewise.
6206 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6207 (simplify_subreg): Update accordingly.
6208 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6209 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6210 that aren't fixed-size.
6211 (get_pool_mode): Return a fixed_size_mode.
6212 (output_constant_pool_2): Take a fixed_size_mode.
6214 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6215 Alan Hayward <alan.hayward@arm.com>
6216 David Sherwood <david.sherwood@arm.com>
6218 * doc/rtl.texi (vec_series): Document.
6219 (const): Say that the operand can be a vec_series.
6220 * rtl.def (VEC_SERIES): New rtx code.
6221 * rtl.h (const_vec_series_p_1): Declare.
6222 (const_vec_series_p): New function.
6223 * emit-rtl.h (gen_const_vec_series): Declare.
6224 (gen_vec_series): Likewise.
6225 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6226 (gen_vec_series): Likewise.
6227 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6228 * simplify-rtx.c (simplify_unary_operation): Handle negations
6230 (simplify_binary_operation_series): New function.
6231 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6232 (test_vector_ops_series): New function.
6233 (test_vector_ops): Call it.
6234 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6235 gen_const_vec_series.
6236 (altivec_lvsr): Likewise.
6237 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6239 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6240 Alan Hayward <alan.hayward@arm.com>
6241 David Sherwood <david.sherwood@arm.com>
6243 * doc/rtl.texi (const): Update description of address constants.
6244 Say that vector constants are allowed too.
6245 * common.md (E, F): Use CONSTANT_P instead of checking for
6247 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6248 checking for CONST_VECTOR.
6249 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6251 * expr.c (expand_expr_real_2): Check for vector modes instead
6252 of checking for CONST_VECTOR.
6253 * rtl.h (const_vec_p): New function.
6254 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6255 (unwrap_const_vec_duplicate): Handle them here too.
6257 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6258 David Malcolm <dmalcolm@redhat.com>
6259 Alan Hayward <alan.hayward@arm.com>
6260 David Sherwood <david.sherwood@arm.com>
6262 * rtl.h (vec_duplicate_p): New function.
6263 * selftest-rtl.c (assert_rtx_eq_at): New function.
6264 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6265 (assert_rtx_eq_at): Declare.
6266 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6267 * selftest-run-tests.c (selftest::run_tests): Call it.
6268 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6269 (simplify_unary_operation_1): Recursively handle vector duplicates.
6270 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6272 (simplify_subreg): Handle subregs of vector duplicates.
6273 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6274 (selftest::simplify_rtx_c_tests): New functions.
6276 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6277 Alan Hayward <alan.hayward@arm.com>
6278 David Sherwood <david.sherwood@arm.com>
6280 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6281 (gen_vec_duplicate): Likewise.
6282 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6284 (gen_const_vector): ...here.
6285 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6286 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6287 whose elements are all equal.
6288 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6289 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6290 (simplify_relational_operation): Likewise.
6291 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6293 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6294 (aarch64_expand_vector_init): Likewise.
6295 * config/arm/arm.c (neon_vdup_constant): Likewise.
6296 (neon_expand_vector_init): Likewise.
6297 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6298 (arm_block_set_unaligned_vect): Likewise.
6299 (arm_block_set_aligned_vect): Likewise.
6300 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6301 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6302 (expand_vec_perm_even_odd_pack): Likewise.
6303 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6304 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6305 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6306 gen_const_vec_duplicate.
6307 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6308 * config/mips/mips.c (mips_gen_const_int_vector): Use
6309 gen_const_vec_duplicate.
6310 (mips_expand_vector_init): Use CONST0_RTX.
6311 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6312 (define_split): Use gen_const_vec_duplicate.
6313 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6314 (define_split): Use gen_const_vec_duplicate.
6315 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6316 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6317 * config/spu/spu.c (spu_const): Likewise.
6319 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6320 Alan Hayward <alan.hayward@arm.com>
6321 David Sherwood <david.sherwood@arm.com>
6323 * combine.c (can_change_dest_mode): Reject changes in
6324 REGMODE_NATURAL_SIZE.
6326 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6328 * configure.ac (--enable-libssp): New.
6329 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6330 trying to determine target-specific default. Adjust indentation.
6331 * configure: Regenerated.
6332 * doc/install.texi (Configuration): Expand --disable-libssp
6335 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6337 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6340 2017-10-31 Martin Jambor <mjambor@suse.cz>
6343 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6345 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6347 * auto-profile.c (autofdo_source_profile::read): Use
6348 UNKNOWN_LOCATION rather than 0.
6349 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6350 (warning_at): ...this overload.
6351 (warning_at_rich_loc_n): Rename to...
6352 (warning_n): ...this overload.
6353 (error_at_rich_loc): Rename to...
6354 (error_at): ...this overload.
6355 (pedwarn_at_rich_loc): Rename to...
6356 (pedwarn): ...this overload.
6357 (permerror_at_rich_loc): Rename to...
6358 (permerror): ...this overload.
6359 (inform_at_rich_loc): Rename to...
6360 (inform): ...this overload.
6361 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6362 (diagnostic_n_impl_richloc): Rename to...
6363 (diagnostic_n_impl): ...this rich_location *-based decl.
6364 (inform_at_rich_loc): Rename to...
6365 (inform): ...this, and add an assertion.
6366 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6367 (warning_at_rich_loc): Rename to...
6368 (warning_at): ...this, and add an assertion.
6369 (warning_at_rich_loc_n): Rename to...
6370 (warning_n): ...this, and add an assertion.
6371 (warning_n): Update location_t-based implementation for removal of
6372 location_t-based diagnostic_n_impl.
6373 (pedwarn_at_rich_loc): Rename to...
6374 (pedwarn): ...this, and add an assertion.
6375 (permerror_at_rich_loc): Rename to...
6376 (permerror): ...this, and add an assertion.
6377 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6378 (error_at_rich_loc): Rename to...
6379 (error_at): ...this, and add an assertion.
6380 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6381 (driver::do_spec_on_infiles): Likewise.
6382 * substring-locations.c (format_warning_va): Update for renaming
6383 of inform_at_rich_loc.
6385 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6387 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6388 _Float<N>X built-in functions so that the variant without the
6389 "__builtin_" prefix is only enabled for the GNU C and Objective C
6390 languages when they are in non-strict ANSI/ISO mode.
6391 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6392 * target.def (floatn_builtin_p): Add a target hook to control
6393 whether _Float<N> and _Float<N>X built-in functions without the
6394 "__builtin_" prefix are enabled, and return true for C and
6395 Objective C in the default hook. Include langhooks.h in
6397 * targhooks.h (default_floatn_builtin_p): Likewise.
6398 * targhooks.c (default_floatn_builtin_p): Likewise.
6399 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6400 floatn_builtin_p target hook.
6401 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6403 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6404 Eric Botcazou <ebotcazou@adacore.com>
6406 PR rtl-optimization/81803
6407 * lra-constraints.c (curr_insn_transform): Also reload the whole
6408 register for a strict subreg no wider than a word if this is for
6409 a WORD_REGISTER_OPERATIONS target.
6411 2017-10-31 Jason Merrill <jason@redhat.com>
6413 * gdbinit.in: Skip over inlines from timevar.h.
6415 2017-10-31 Martin Liska <mliska@suse.cz>
6417 * doc/gcov.texi: Document new option.
6418 * gcov.c (print_usage): Likewise print it.
6419 (process_args): Support the argument.
6420 (format_count): New function.
6421 (format_gcov): Use the function.
6423 2017-10-31 Martin Liska <mliska@suse.cz>
6425 * gcov.c (struct name_map): do not use typedef.
6426 Define operator== and operator<.
6427 (name_search): Remove.
6428 (name_sort): Remove.
6429 (main): Do not allocate names.
6430 (process_file): Add vertical space.
6431 (generate_results): Use std::find.
6432 (release_structures): Do not release memory.
6433 (find_source): Use std::find.
6435 2017-10-31 Martin Liska <mliska@suse.cz>
6437 * gcov.c (struct line_info): Remove it's typedef.
6438 (line_info::line_info): Add proper ctor.
6439 (line_info::has_block): Do not use a typedef.
6440 (struct source_info): Do not use typedef.
6441 (circuit): Likewise.
6442 (get_cycles_count): Likewise.
6443 (output_intermediate_file): Iterate via vector iterator.
6444 (add_line_counts): Use std::vector methods.
6445 (accumulate_line_counts): Likewise.
6446 (output_lines): Likewise.
6448 2017-10-31 Martin Liska <mliska@suse.cz>
6450 * gcov.c (struct source_info): Remove typedef.
6451 (source_info::source_info): Add proper ctor.
6452 (accumulate_line_counts): Use struct, not it's typedef.
6453 (output_gcov_file): Likewise.
6454 (output_lines): Likewise.
6455 (main): Do not allocate an array.
6456 (output_intermediate_file): Use size of vector container.
6457 (process_file): Resize the vector.
6458 (generate_results): Do not preallocate, use newly added vector
6460 (release_structures): Do not release sources.
6461 (find_source): Use vector methods.
6462 (add_line_counts): Do not use typedef.
6464 2017-10-31 Martin Liska <mliska@suse.cz>
6466 * doc/gcov.texi: Document that.
6467 * gcov.c (add_line_counts): Mark lines with a non-executed
6469 (output_line_beginning): Handle such lines.
6470 (output_lines): Pass new argument.
6471 (output_intermediate_file): Print it in intermediate format.
6473 2017-10-31 Martin Liska <mliska@suse.cz>
6475 * color-macros.h: New file.
6476 * diagnostic-color.c: Factor out color related to macros to
6478 * doc/gcov.texi: Document -k option.
6479 * gcov.c (INCLUDE_STRING): Include string.h.
6480 (print_usage): Add -k option.
6481 (process_args): Parse it.
6482 (pad_count_string): New function.
6483 (output_line_beginning): Likewise.
6484 (DEFAULT_LINE_START): New macro.
6485 (output_lines): Support color output.
6487 2017-10-31 Martin Liska <mliska@suse.cz>
6489 PR gcov-profile/82633
6490 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6491 their interaction with GCOV infrastructure.
6492 * configure.ac: Add -fkeep-{inline,static}-functions to
6494 * configure: Regenerate.
6496 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6499 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6501 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6504 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6505 into a register if it does not fit into an immediate offset field.
6507 2017-10-31 Olivier Hainque <hainque@adacore.com>
6509 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6511 2017-10-31 Julia Koval <julia.koval@intel.com>
6513 * config.gcc: Add gfniintrin.h.
6514 * config/i386/gfniintrin.h: New.
6515 * config/i386/i386-builtin-types.def
6516 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6517 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6518 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6519 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6520 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6521 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6522 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6523 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6524 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6525 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6526 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6527 * config/i386/immintrin.h: Include gfniintrin.h.
6528 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6530 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6532 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6534 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6536 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6537 (arm_ashldi3_1bit): Remove pattern.
6538 (ashrdi3): Remove shift by 1 expansion.
6539 (arm_ashrdi3_1bit): Remove pattern.
6540 (lshrdi3): Remove shift by 1 expansion.
6541 (arm_lshrdi3_1bit): Remove pattern.
6542 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6543 cost of ashldi3 by 1.
6544 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6545 (<shift>di3_neon): Likewise.
6547 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6549 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6550 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6551 and (*aarch64_simd_mov<VQ:mode>).
6552 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6553 pattern alternative.
6554 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6555 attributes to match pattern alternative.
6557 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6559 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6560 * config/rs6000/emmintrin.h: New file.
6561 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6563 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6565 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6566 * (movdi_vfp_cortexa8): Remove pattern.
6568 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6570 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6572 (Specific, alpha*-dec-osf5.1): Remove.
6573 (Specific, mips-sgi-irix5): Remove.
6574 (Specific, mips-sgi-irix6): Remove.
6576 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6579 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6580 arguments to clear_bit_region_be.
6582 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6584 * gimplify.c: Include memmodel.h.
6586 2017-10-30 Martin Jambor <mjambor@suse.cz>
6588 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6589 condition whether loop should be executed at all.
6591 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6593 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6594 gimple folding of vec_madd() intrinsics.
6595 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6596 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6597 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6599 2017-10-30 Richard Biener <rguenther@suse.de>
6601 PR tree-optimization/82762
6603 2017-10-23 Richard Biener <rguenther@suse.de>
6605 PR tree-optimization/82129
6607 2017-08-01 Richard Biener <rguenther@suse.de>
6609 PR tree-optimization/81181
6610 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6611 (compute_antic): ... end of iteration here.
6613 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6615 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6616 -std=iso9899:2017 and -std=gnu17.
6617 * doc/standards.texi (C Language): Document C17 support.
6618 * doc/cpp.texi (Overview): Mention -std=c17.
6619 (Standard Predefined Macros): Document C11 and C17 values of
6620 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6621 * doc/extend.texi (Inline): Do not list individual options for
6622 standards newer than C99.
6623 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6625 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6628 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6630 * asan.c (asan_finish_file): Align asan globals array by shadow
6633 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6636 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6637 (struct store_immediate_info): Add bitregion_start and bitregion_end
6639 (store_immediate_info::store_immediate_info): Add brs and bre
6640 arguments and initialize bitregion_{start,end} from those.
6641 (struct merged_store_group): Add bitregion_start, bitregion_end,
6642 align_base and mask fields. Drop unnecessary struct keyword from
6643 struct store_immediate_info. Add do_merge method.
6644 (clear_bit_region_be): Use memset instead of loop storing zeros.
6645 (merged_store_group::do_merge): New method.
6646 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6647 stores as long as the surrounding bitregions have no gaps.
6648 (merged_store_group::merge_overlapping): Use do_merge.
6649 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6650 is byte aligned, rather than requiring that start and width are
6651 byte aligned. Drop unnecessary struct keyword from
6652 struct store_immediate_info. Allocate and populate also mask array.
6653 Make start of the arrays relative to bitregion_start rather than
6654 start and size them according to bitregion_{end,start} difference.
6655 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6656 struct store_immediate_info.
6657 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6658 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6659 struct keyword from struct store_immediate_info.
6660 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6661 between stores as long as the surrounding bitregions have no gaps.
6663 (struct split_store): Add orig non-static data member.
6664 (split_store::split_store): Initialize orig to false.
6665 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6666 if there is exactly a single original stmt. Change stmts argument
6667 to pointer from reference, if NULL, don't push anything to it. Add
6668 first argument, use it to optimize skipping over orig stmts that
6669 are known to be before bitpos already. Simplify.
6670 (split_group): Return unsigned int count how many stores are or
6671 would be needed rather than a bool. Add allow_unaligned argument.
6672 Change split_stores argument from reference to pointer, if NULL,
6673 only do a dry run computing how many stores would be produced.
6674 Rewritten algorithm to use both alignment and misalign if
6675 !allow_unaligned and handle bitfield stores with gaps.
6676 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6677 from bitregion_start instead of start. Compute allow_unaligned
6678 here, if true, do 2 split_group dry runs to compute which one
6679 produces fewer stores and prefer aligned if equal. Punt if
6680 new count is bigger or equal than original before emitting any
6681 statements, rather than during that. Remove no longer needed
6682 new_ssa_names tracking. Replace num_stmts with
6683 split_stores.length (). Use 32-bit stack allocated entries
6684 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6685 if possible. Handle bitfields with gaps.
6686 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6687 Compute bitregion_{start,end} for the stores and construct
6688 store_immediate_info with that. Formatting fixes.
6690 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6693 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6694 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6696 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6698 * gimplify.c: Include tm_p.h.
6700 * common.opt (gcoff): Re-add as ignored option.
6701 (gcoff1, gcoff2, gcoff3): Likewise.
6703 * Makefile.in (OBJS): Delete sdbout.o.
6704 (GTFILES): Delete $(srcdir)/sdbout.c.
6705 * debug.h: Delete sdb_debug_hooks.
6706 * final.c: Delete sdbout.h include.
6707 (final_scan_insn): Delete SDB_DEBUG check.
6708 (rest_of_clean_state): Likewise.
6709 * output.h: Delete sdb_begin_function_line.
6712 * toplev.c: Delete sdbout.h include.
6713 (process_options): Delete SDB_DEBUG check.
6714 * tree-core.h (tree_type_common): Delete pointer field of
6716 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6717 TYPE_SYMTAB_POINTER.
6718 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6719 (TYPE_SYMTAB_IS_POINTER): Delete.
6720 (TYPE_SYMTAB_IS_DIE): Renumber.
6721 * xcoffout.c: Refer to former sdbout.c file.
6722 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6724 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6725 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6726 (Debugging Options): Delete -gcoff.
6727 (-gstabs): Delete SDB reference.
6729 (-gcoff@var{level}): Delete.
6730 * doc/passes.texi (Debugging information output): Delete SDB and
6731 sdbout.c references.
6732 * doc/tm.texi: Regenerate.
6733 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6734 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6736 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6737 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6738 (SDB_DEBUGGING_INFO): Delete.
6739 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6740 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6741 * target.def (output_source_filename): Delete COFF reference.
6743 * common.opt (gcoff): Delete.
6744 (gxcoff+): Update Negative chain.
6745 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6747 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6748 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6749 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6751 (expand_function_start): Change sdb reference to past tense.
6752 (expand_function_end): Change sdb reference to past tense.
6753 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6754 * opts.c (debug_type_names): Delete coff entry.
6755 (common_handle_option): Delete OPT_gcoff case.
6756 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6758 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6759 * config/cris/cris.h: Delete SDB reference in comment.
6760 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6761 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6762 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6763 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6765 (ix86_expand_prologue): Likewise.
6766 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6767 * config/ia64/ia64.h: Likewise.
6768 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6769 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6771 * config/mmix/mmix.h: Likewise.
6772 * config/nds32/nds32.c: Likewise.
6773 * config/stormy/storym16.h: Likewise.
6774 * config/visium/visium.h: Likewise.
6775 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6777 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6779 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6780 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6781 FRAME_POINTER_REGNUM point at high end of local var area.
6783 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6785 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6786 Move comment around. Do not reset best_edge for a copiable
6787 destination if the copy would cause a partition change.
6788 (better_edge_p): Remove redundant check.
6790 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6792 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6794 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6796 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6797 for math functions that have _Float<N> and _Float<N>X variants.
6798 (mathfn_built_in_2): Add support for math functions that have
6799 _Float<N> and _Float<N>X variants.
6800 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6801 (expand_builtin_mathfn_ternary): Add support for fma with
6802 _Float<N> and _Float<N>X variants.
6803 (expand_builtin): Likewise.
6804 (fold_builtin_3): Likewise.
6805 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6806 create math function _Float<N> and _Float<N>X variants as external
6808 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6809 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6810 the __builtin_ prefix and if not strict ansi, without the prefix.
6811 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6812 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6813 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6814 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6815 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6816 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6817 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6818 function signatures for fma _Float<N> and _Float<N>X variants.
6819 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6820 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6821 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6822 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6823 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6824 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6825 * gencfn-macros.c (print_case_cfn): Add support for math functions
6826 that have _Float<N> and _Float<N>X variants.
6827 (print_define_operator_list): Likewise.
6828 (fltfn_suffixes): Likewise.
6830 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6831 for math functions that have _Float<N> and _Float<N>X variants.
6832 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6833 and _Float<N>X variants.
6834 (COPYSIGN): Likewise.
6837 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6838 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6840 (integer_valued_read_call_p): Likewise.
6841 * fold-const-call.c (fold_const_call_ss): Likewise.
6842 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6843 _Float<N> and _Float<N>X variants.
6844 (fold_const_call_ssss): Add support for fma _Float<N> and
6845 _Float<N>X variants.
6846 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6847 support for copysign and fma _Float<N> and _Float<N>X variants.
6848 (backprop::process_builtin_call_use): Likewise.
6849 * tree-call-cdce.c (can_test_argument_range); Add support for
6850 sqrt _Float<N> and _Float<N>X variants.
6851 (edom_only_function): Likewise.
6852 (get_no_error_domain): Likewise.
6853 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6854 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6855 copysign _Float<N> and _Float<N>X variants.
6856 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6857 handled by machine independent code.
6858 (FMAF128): Likewise.
6859 * doc/cpp.texi (Common Predefined Macros): Document defining
6860 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6861 fma _Float<N> and _Float<N>X variants.
6863 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6866 * config/i386/i386-modes.def (CCFPU): Remove definition.
6867 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6868 (ix86_cc_modes_compatible): Ditto.
6869 (ix86_expand_carry_flag_compare): Ditto.
6870 (ix86_expand_int_movcc): Ditto.
6871 (ix86_expand_int_addcc): Ditto.
6872 (ix86_reverse_condition): Ditto.
6873 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6874 Return true/false for unordered/ordered fp comparisons.
6875 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6876 (ix86_prepare_fp_compare_args): Update for rename.
6877 (ix86_expand_fp_compare): Update for rename. Generate unordered
6878 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6879 (ix86_expand_sse_compare_and_jump): Ditto.
6880 * config/i386/predicates.md (fcmov_comparison_operator):
6881 Remove CCFPU mode handling.
6882 (ix86_comparison_operator): Ditto.
6883 (ix86_carry_flag_operator): Ditto.
6884 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6885 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6886 (*cmpu<mode>_cc_i387): Ditto.
6887 (FPCMP): Remove mode iterator.
6888 (unord): Remove mode attribute.
6889 (unord_subst): New define_subst transformation
6890 (unord): New define_subst attribute.
6892 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6893 (*cmpi<unord>xf_i387): Ditto.
6894 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6895 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6896 using unord_subst transformation.
6897 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6898 (round_saeonly): Also handle CCFP mode.
6899 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6900 Remove UNSPEC_SAHF unspec handling.
6902 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6904 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6906 2017-10-27 Jeff Law <law@redhat.com>
6908 * gimple-ssa-sprintf.c: Include domwalk.h.
6909 (class sprintf_dom_walker): New class, derived from dom_walker.
6910 (sprintf_dom_walker::before_dom_children): New function.
6911 (struct call_info): Moved into sprintf_dom_walker class
6912 (compute_formath_length, handle_gimple_call): Likewise.
6913 (sprintf_length::execute): Call the dominator walker rather
6914 than walking the statements.
6916 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6917 gimple statement locations.
6918 (check_array_bounds): Corresponding changes. Get the statement's
6919 location directly from wi->stmt.
6921 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6924 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6926 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6928 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6929 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6931 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6934 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6935 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6936 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6937 maybe_get_pool_constant.
6938 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6941 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6943 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6944 2.26 caveat. Update gas and gld versions.
6945 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6948 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6950 * cgraph.h (set_malloc_flag): Declare.
6951 * cgraph.c (set_malloc_flag_1): New function.
6952 (set_malloc_flag): Likewise.
6953 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6954 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6956 (read_ipa_call_summary): Add support for reading is_return_callee.
6957 (write_ipa_call_summary): Stream is_return_callee.
6958 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6959 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6960 ipa-prop.h, ipa-fnsummary.h.
6961 (pure_const_names): Change to static.
6962 (malloc_state_e): Define.
6963 (malloc_state_names): Define.
6964 (funct_state_d): Add field malloc_state.
6965 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6966 (check_retval_uses): New function.
6967 (malloc_candidate_p): Likewise.
6968 (analyze_function): Add support for malloc attribute.
6969 (pure_const_write_summary): Stream malloc_state.
6970 (pure_const_read_summary): Add support for reading malloc_state.
6971 (dump_malloc_lattice): New function.
6972 (propagate_malloc): New function.
6973 (warn_function_malloc): New function.
6974 (ipa_pure_const::execute): Call propagate_malloc and
6975 ipa_free_fn_summary.
6976 (pass_local_pure_const::execute): Add support for malloc attribute.
6977 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6978 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6980 2017-10-27 Martin Liska <mliska@suse.cz>
6982 PR gcov-profile/82457
6983 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6984 for fork-like functions to be properly instrumented.
6986 2017-10-27 Richard Biener <rguenther@suse.de>
6989 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6990 info when we redirected EH.
6992 2017-10-26 Michael Collison <michael.collison@arm.com>
6994 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6996 (<optab>_trunchf<GPI:mode>2: New pattern.
6997 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6998 * config/aarch64/iterators.md (wv): New mode attribute.
6999 (vf, VF): New mode attributes.
7000 (vgp, VGP): New mode attributes.
7001 (s): Update attribute with SImode and DImode prefixes.
7003 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7005 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7006 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7007 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7008 (nios2_option_overide): Initialize it. Don't allow R0-relative
7009 addressing with PIC.
7010 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7011 (nios2_symbolic_constant_p): Likewise.
7012 (nios2_legitimate_address_p): Likewise.
7013 (nios2_r0rel_section_name_p): New.
7014 (nios2_symbol_ref_in_r0rel_data_p): New.
7015 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7016 (r0rel_constant_p): New.
7017 (nios2_print_operand_address): Handle r0rel_constant_p.
7018 (nios2_cdx_narrow_form_p): Likewise.
7019 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7020 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7021 (Nios II Options): Document -mr0rel-sec.
7023 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7025 * config/nios2/nios2.c: Include xregex.h.
7026 (nios2_gprel_sec_regex): New.
7027 (nios2_option_overide): Initialize it. Don't allow GP-relative
7028 addressing with PIC.
7029 (nios2_small_section_name_p): Check for regex match.
7030 * config/nios2/nios2.opt (mgprel-sec=): New option.
7031 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7032 (Nios II Options): Document -mgprel-sec.
7034 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7036 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7038 2017-10-26 Tom de Vries <tom@codesourcery.com>
7040 PR tree-optimization/82707
7041 * gimple.c (gimple_copy): Fix unsharing of
7042 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7044 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7046 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7047 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7048 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7049 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7050 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7051 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7052 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7053 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7054 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7055 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7056 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7057 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7058 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7059 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7060 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7061 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7062 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7064 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7066 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7068 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7069 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7070 warning to rs6000.c. Remove the Undocumented flag, since it has
7072 (-mabi=ibmlongdouble): Likewise.
7073 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7074 already set, set the default format for long double.
7075 (rs6000_debug_reg_global): Print whether long double is IBM or
7077 (rs6000_option_override_internal): Rework setting long double
7078 format. Only warn if the user is changing the long double default
7079 and they did not use -Wno-psabi.
7080 * doc/invoke.texi (PowerPC options): Update the documentation for
7081 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7083 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7084 Alan Hayward <alan.hayward@arm.com>
7085 David Sherwood <david.sherwood@arm.com>
7087 * rtl.h (wider_subreg_mode): New function.
7088 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7089 rather than an unsigned int *.
7090 * ira-color.c (regno_max_ref_width): Replace with...
7091 (regno_max_ref_mode): ...this new variable.
7092 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7093 Use wider_subreg_mode.
7094 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7095 rather than an unsigned int *.
7096 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7097 (process_alt_operands): Likewise.
7098 (invariant_p): Likewise.
7099 * lra-spills.c (assign_mem_slot): Likewise.
7100 (add_pseudo_to_slot): Likewise.
7101 * lra.c (collect_non_operand_hard_regs): Likewise.
7102 (add_regs_to_insn_regno_info): Likewise.
7103 * reload1.c (regno_max_ref_width): Replace with...
7104 (regno_max_ref_mode): ...this new variable.
7105 (reload): Update accordingly. Update call to
7106 ira_sort_regnos_for_alter_reg.
7107 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7108 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7109 (scan_paradoxical_subregs): Likewise.
7111 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7113 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7114 (aarch64_frame): Add emit_frame_chain boolean.
7115 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7116 Move eh_return case to aarch64_layout_frame.
7117 (aarch64_layout_frame): Initialize emit_frame_chain.
7118 (aarch64_expand_prologue): Use emit_frame_chain.
7120 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7122 * config/aarch64/aarch64.c (aarch64_layout_frame):
7123 Ensure LR is always stored at the bottom of the callee-saves.
7124 Remove rarely used frame layout which saves callee-saves at top of
7125 frame, so the store of LR can be used as a valid probe in all cases.
7127 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7129 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7130 Improve unaligned TImode/TFmode base/offset split.
7132 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7133 Alan Hayward <alan.hayward@arm.com>
7134 David Sherwood <david.sherwood@arm.com>
7136 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7137 * combine.c (find_single_use_1): Likewise.
7138 (expand_field_assignment): Likewise.
7139 (move_deaths): Likewise.
7140 * lra-constraints.c (simplify_operand_subreg): Likewise.
7141 (curr_insn_transform): Likewise.
7142 * lra.c (collect_non_operand_hard_regs): Likewise.
7143 (add_regs_to_insn_regno_info): Likewise.
7144 * rtlanal.c (reg_referenced_p): Likewise.
7145 (covers_regno_no_parallel_p): Likewise.
7147 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7149 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7150 Don't print any bits outside the precision of the value.
7151 * wide-int.cc (test_printing): Add some new tests.
7153 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7155 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7156 supports -xbrace_comment option.
7157 * configure: Regenerate.
7158 * config.in: Regenerate.
7159 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7160 (ASM_CPU_SPEC): Use it.
7162 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7164 * target.def (static_rtx_alignment): New hook.
7165 * targhooks.h (default_static_rtx_alignment): Declare.
7166 * targhooks.c (default_static_rtx_alignment): New function.
7167 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7168 * doc/tm.texi: Regenerate.
7169 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7170 instead of targetm.constant_alignment. Remove call to
7172 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7173 (cris_preferred_mininum_alignment): New function, split out from...
7174 (cris_constant_alignment): ...here.
7175 (cris_static_rtx_alignment): New function.
7176 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7178 (ix86_constant_alignment): ...here.
7179 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7180 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7181 (mmix_static_rtx_alignment): New function.
7182 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7183 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7185 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7188 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7189 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7191 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7193 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7194 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7195 mark as a sign-extending load.
7196 (local_pic_load_u): Define.
7198 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7201 * fold-const.c (operand_equal_for_comparison_p): Also return true
7202 if ARG0 is a simple variant of ARG1 with narrower precision.
7203 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7205 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7207 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7209 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7210 scatter_static, scatter_per_elt.
7211 * x86-tune-costs.h: Add new cost entries.
7213 2017-10-25 Richard Biener <rguenther@suse.de>
7215 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7216 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7217 class pass_fre): Move to ...
7218 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7219 class pass_fre): ... here and adjust for statistics.
7221 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7224 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7225 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7226 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7227 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7230 2017-10-25 Richard Biener <rguenther@suse.de>
7232 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7233 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7234 eliminate_push_avail, eliminate_insert): Move inside...
7235 (class eliminate_dom_walker): ... this class in preparation
7237 (fini_eliminate): Remove by merging with ...
7238 (eliminate): ... this function. Adjust for class changes.
7239 (pass_pre::execute): Remove fini_eliminate call.
7240 (pass_fre::execute): Likewise.
7242 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7245 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7246 (VPERMI2, VPERMI2I): New mode iterators.
7247 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7248 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7250 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7251 mode iterator. Remove 3 old define_insn patterns.
7252 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7253 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7254 VPERMI2 mode iterator, remove the other two expanders.
7255 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7256 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7257 instructions, remove the other two patterns.
7258 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7259 mode iterator, remove the other two patterns.
7260 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7261 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7262 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7263 and adjust argument order accordingly.
7264 (ix86_expand_vec_perm): Adjust caller.
7265 (expand_vec_perm_1): Likewise.
7266 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7267 (expand_vec_perm_vpermt2_vpshub2): ... this.
7268 (ix86_expand_vec_perm_const_1): Adjust caller.
7269 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7272 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7273 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7274 (vec_shl_<mode>): Remove unused expander.
7275 (avx512bw_<shift_insn><mode>3): New define_insn.
7276 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7277 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7279 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7282 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7285 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7287 PR rtl-optimization/82396
7288 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7289 (autopref_multipass_init): Simplify initialization.
7290 (autopref_rank_data): Simplify sort order.
7291 * gcc/sched-int.h (autopref_multipass_data_): Remove
7292 multi_mem_insn_p, min_offset and max_offset.
7294 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7297 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7298 Check special value of flag_omit_frame_pointer.
7299 (aarch64_can_eliminate): Likewise.
7300 (aarch64_override_options_after_change_1): Simplify handling of
7301 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7303 2017-10-24 Richard Biener <rguenther@suse.de>
7305 PR tree-optimization/82697
7306 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7307 zero for conditional load and unconditional store.
7309 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7311 * doc/install.texi: Document bootstrap-cet.
7313 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7316 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7317 ENDBR instruction at function entrance if function is only
7320 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7323 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7324 patterns to better describe from which operation the CF is computed.
7325 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7326 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7327 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7328 is 0, use _0 suffixed expanders instead of emitting a comparison
7331 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7333 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7334 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7336 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7339 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7340 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7341 * loop-iv.c (iv_get_reaching_def): Likewise.
7342 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7343 variable is promoted and the partition contains undefined values.
7345 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7347 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7349 (nios2_address_cost): Define.
7350 (nios2_legitimize_address): Recognize (exp + constant) directly.
7351 (TARGET_ADDRESS_COST): Define.
7353 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7355 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7356 (nios2_symbolic_memory_operand_p): Declare.
7357 (nios2_split_large_constant): Declare.
7358 (nios2_split_symbolic_memory_operand): Declare.
7359 * config/nios2/nios2.c: Adjust includes.
7360 (nios2_symbolic_constant_allowed): New.
7361 (nios2_symbolic_constant_p): New.
7362 (nios2_plus_symbolic_constant_p): New.
7363 (nios2_valid_addr_expr_p): Recognize addresses involving
7365 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7366 (nios2_symbolic_memory_operand_p): New.
7367 (nios2_large_constant_p): New.
7368 (nios2_split_large_constant): New.
7369 (nios2_split_plus_large_constant): New.
7370 (nios2_split_symbolic_memory_operand): New.
7371 (nios2_legitimize_address): Code refactoring. Handle addresses
7372 involving symbolic constants.
7373 (nios2_emit_move_sequence): Likewise.
7374 (nios2_print_operand): Improve error output.
7375 (nios2_print_operand_address): Handle LO_SUM.
7376 (nios2_cdx_narrow_form_p): Likewise.
7377 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7378 operands involving symbolic constants.
7379 (movhi_internal, movsi_internal): Likewise.
7380 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7381 (extendhisi2, extendqi<mode>2): Likewise.
7383 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7385 * tree-pass.h (PROP_rtl_split_insns): Define.
7386 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7388 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7390 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7392 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7395 * target.def (const_not_ok_for_debug_p): Default to
7396 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7397 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7398 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7399 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7400 which targetm.const_not_ok_for_debug_p returned true.
7401 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7403 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7405 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7406 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7407 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7409 (ix86_const_not_ok_for_debug_p): New function.
7410 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7411 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7413 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7416 * system.h: Conditionally include "unique-ptr.h" if
7417 INCLUDE_UNIQUE_PTR is defined.
7418 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7419 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7421 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7423 * config/rl78/rl78.md: New define_expand "subdi3".
7425 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7428 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7429 DF_REF_INSN if DF_REF_INSN_INFO is false.
7431 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7433 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7434 xmm_move instead of sse_move.
7435 (sse_store_index): New function.
7436 (ix86_register_move_cost): Be more sensible about mismatch stall;
7437 model AVX moves correctly; make difference between sse->integer and
7439 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7440 moves; make difference between SSE and AVX.
7441 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7442 and zmm_move. Increase size of sse load and store tables;
7443 add unaligned load and store tables; add ssemmx_to_integer.
7444 * x86-tune-costs.h: Update all entries according to real
7445 move latencies from Agner Fog's manual and chip documentation.
7447 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7450 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7451 * config/i386/constraints.md (Wf): New constraint.
7452 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7453 (cmp<dwi>_doubleword): Removed.
7454 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7455 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7456 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7457 expand with cmp<dwi>_doubleword. For LTU and GEU use
7458 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7460 * common.opt (gcolumn-info): Enable by default.
7461 * doc/invoke.texi (gcolumn-info): Document new default.
7463 2017-10-23 Richard Biener <rguenther@suse.de>
7465 PR tree-optimization/82672
7466 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7467 Fold the stmt if we propagated into it.
7469 2017-10-23 Richard Biener <rguenther@suse.de>
7471 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7472 (bitmap_remove_expr_from_set): ... this. All callers call this
7473 for non-constant values.
7474 (bitmap_set_subtract): Rename to...
7475 (bitmap_set_subtract_expressions): ... this. Adjust and
7477 (bitmap_set_contains_value): Remove superfluous check.
7478 (bitmap_set_replace_value): Inline into single caller ...
7479 (bitmap_value_replace_in_set): ... here and simplify.
7480 (dependent_clean): Merge into ...
7481 (clean): ... this using an overload. Adjust.
7482 (prune_clobbered_mems): Adjust.
7483 (compute_antic_aux): Likewise.
7484 (compute_partial_antic_aux): Likewise.
7486 2017-10-23 Richard Biener <rguenther@suse.de>
7488 PR tree-optimization/82129
7490 2017-08-01 Richard Biener <rguenther@suse.de>
7492 PR tree-optimization/81181
7493 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7494 (compute_antic): ... end of iteration here.
7496 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7498 * target.def (starting_frame_offset): New hook.
7499 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7500 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7501 * doc/tm.texi.in: Regenerate.
7502 * hooks.h (hook_hwi_void_0): Declare.
7503 * hooks.c (hook_hwi_void_0): New function.
7504 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7505 STARTING_FRAME_OFFSET.
7506 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7507 * reload1.c (reload): Likewise.
7508 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7509 instead of STARTING_FRAME_OFFSET.
7510 * function.c (try_fit_stack_local): Likewise.
7511 (assign_stack_local_1): Likewise
7512 (instantiate_virtual_regs): Likewise.
7513 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7514 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7515 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7516 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7517 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7518 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7519 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7520 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7521 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7522 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7523 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7524 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7525 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7526 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7527 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7528 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7529 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7530 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7531 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7532 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7533 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7534 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7535 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7536 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7537 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7538 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7539 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7540 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7541 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7542 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7543 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7544 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7545 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7546 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7547 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7548 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7549 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7550 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7551 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7552 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7553 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7554 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7555 return a HOST_WIDE_INT.
7556 (avr_builtin_setjmp_frame_value): Use it instead of
7557 STARTING_FRAME_OFFSET.
7558 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7559 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7560 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7562 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7563 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7564 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7565 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7566 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7567 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7568 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7569 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7570 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7571 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7572 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7573 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7575 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7576 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7577 * config/mips/mips.c (mips_compute_frame_info): Refer to
7578 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7579 (mips_starting_frame_offset): New function.
7580 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7581 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7582 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7583 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7584 and return a HOST_WIDE_INT.
7585 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7586 (mmix_initial_elimination_offset): Refer to
7587 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7588 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7589 * config/pa/pa.c (pa_starting_frame_offset): New function.
7590 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7591 (pa_expand_prologue): Likewise.
7592 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7593 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7594 !FRAME_GROWS_DOWNWARD handling to...
7595 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7596 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7597 !FRAME_GROWS_DOWNWARD handling to...
7598 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7599 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7600 !FRAME_GROWS_DOWNWARD handling to...
7601 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7602 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7604 (rs6000_starting_frame_offset): New function.
7605 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7606 !FRAME_GROWS_DOWNWARD handling to...
7607 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7608 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7609 !FRAME_GROWS_DOWNWARD handling to...
7610 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7611 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7612 !FRAME_GROWS_DOWNWARD handling to...
7613 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7614 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7615 (rs6000_starting_frame_offset): New function.
7616 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7617 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7618 * config/vax/vax.c (vax_starting_frame_offset): New function.
7619 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7620 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7621 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7622 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7623 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7624 * system.h (STARTING_FRAME_OFFSET): Poison.
7626 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7628 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7629 SCALAR_TYPE_MODE instead of TYPE_MODE.
7631 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7632 Alan Hayward <alan.hayward@arm.com>
7633 David Sherwood <david.sherwood@arm.com>
7635 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7637 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7638 Alan Hayward <alan.hayward@arm.com>
7639 David Sherwood <david.sherwood@arm.com>
7641 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7643 2017-10-23 Richard Biener <rguenther@suse.de>
7645 PR tree-optimization/82129
7646 * tree-ssa-pre.c (bitmap_set_and): Remove.
7647 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7648 canonicalizing expressions in the set to those with lowest
7649 ID rather than taking that from the first edge.
7651 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7653 * combine.c (rtx_equal_for_field_assignment_p): Use
7654 byte_lowpart_offset.
7656 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7657 Alan Hayward <alan.hayward@arm.com>
7658 David Sherwood <david.sherwood@arm.com>
7660 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7661 to a SUBREG_PROMOTED_VAR.
7663 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7664 Alan Hayward <alan.hayward@arm.com>
7665 David Sherwood <david.sherwood@arm.com>
7667 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7668 (expand_debug_source_expr): Likewise.
7669 * combine.c (combine_simplify_rtx): Likewise.
7670 * cse.c (fold_rtx): Likewise.
7671 * optabs.c (expand_float): Likewise.
7672 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7673 (simplify_binary_operation_1): Likewise.
7675 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7676 Alan Hayward <alan.hayward@arm.com>
7677 David Sherwood <david.sherwood@arm.com>
7679 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7680 (record_promoted_value): Likewise.
7681 * expr.c (expand_expr_real_2): Likewise.
7682 * ree.c (update_reg_equal_equiv_notes): Likewise.
7683 (combine_set_extension): Likewise.
7684 * rtlanal.c (low_bitmask_len): Likewise.
7685 * simplify-rtx.c (neg_const_int): Likewise.
7686 (simplify_binary_operation_1): Likewise.
7688 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7689 Alan Hayward <alan.hayward@arm.com>
7690 David Sherwood <david.sherwood@arm.com>
7692 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7693 * regcprop.c (maybe_mode_change): Likewise.
7694 * reload1.c (alter_reg): Likewise.
7696 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7698 * inchash.h (inchash::hash::add_wide_int): New function.
7699 * lto-streamer-out.c (hash_tree): Use it.
7701 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7703 * inchash.h (inchash::hash::add_wide_int): Rename to...
7704 (inchash::hash::add_hwi): ...this.
7705 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7706 (polymorphic_call_target_hasher::hash): Likewise.
7707 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7708 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7709 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7710 * lto-streamer-out.c (hash_tree): Likewise.
7711 * optc-save-gen.awk: Likewise.
7712 * tree.c (add_expr): Likewise.
7714 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7717 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7718 for ordered inequality comparisons even with TARGET_IEEE_FP.
7720 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7723 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7724 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7725 Expand with cmp<dwi>_doubleword.
7727 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7729 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7730 List CET intrinsics.
7731 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7732 specific to -fcf-protection option.
7734 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7736 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7737 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7738 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7739 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7740 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7741 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7742 (extra_objs): Add cet.o for Linux/x86 targets.
7743 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7744 * config/i386/cet.c: New file.
7745 * config/i386/cetintrin.h: Likewise.
7746 * config/i386/t-cet: Likewise.
7747 * config/i386/cpuid.h (bit_SHSTK): New.
7748 (bit_IBT): Likewise.
7749 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7750 pass IBT and SHSTK bits.
7751 * config/i386/i386-builtin-types.def
7752 (VOID_FTYPE_UNSIGNED_PVOID): New.
7753 (VOID_FTYPE_UINT64_PVOID): Likewise.
7754 * config/i386/i386-builtin.def: Add CET intrinsics.
7755 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7756 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7757 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7758 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7760 * config/i386/i386.c (rest_of_insert_endbranch): New.
7761 (pass_data_insert_endbranch): Likewise.
7762 (pass_insert_endbranch): Likewise.
7763 (make_pass_insert_endbranch): Likewise.
7764 (ix86_notrack_prefixed_insn_p): Likewise.
7765 (ix86_target_string): Add -mibt, -mshstk flags.
7766 (ix86_option_override_internal): Add flag_cf_protection
7768 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7769 (ix86_print_operand): Add 'notrack' prefix output.
7770 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7771 (ix86_expand_builtin): Expand CET intrinsics.
7772 (x86_output_mi_thunk): Add 'endbranch' instruction.
7773 * config/i386/i386.h (TARGET_IBT): New.
7774 (TARGET_IBT_P): Likewise.
7775 (TARGET_SHSTK): Likewise.
7776 (TARGET_SHSTK_P): Likewise.
7777 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7778 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7779 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7780 (builtin_setjmp_setup): New pattern.
7781 (builtin_longjmp): Likewise.
7782 (rdssp<mode>): Likewise.
7783 (incssp<mode>): Likewise.
7784 (saveprevssp): Likewise.
7785 (rstorssp): Likewise.
7786 (wrss<mode>): Likewise.
7787 (wruss<mode>): Likewise.
7788 (setssbsy): Likewise.
7789 (clrssbsy): Likewise.
7790 (nop_endbr): Likewise.
7791 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7793 * config/i386/immintrin.h: Include <cetintrin.h>.
7794 * config/i386/linux-common.h
7795 (file_end_indicate_exec_stack_and_cet): New prototype.
7796 (TARGET_ASM_FILE_END): New.
7798 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7800 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7801 latencies instead of having separate table; make difference between
7802 integer and float costs.
7803 * i386.h (processor_costs): Remove scalar_stmt_cost,
7804 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7805 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7807 * x86-tune-costs.h: Remove entries which has been removed in
7808 procesor_costs from all tables; make cond_taken_branch_cost
7809 and cond_not_taken_branch_cost COST_N_INSNS based.
7811 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7813 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7815 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7817 * config/i386/i386.md (isa): Remove fma_avx512f.
7818 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7819 <avx512>_fmadd_<mode>_mask3<round_name>,
7820 <avx512>_fmsub_<mode>_mask<round_name>,
7821 <avx512>_fmsub_<mode>_mask3<round_name>,
7822 <avx512>_fnmadd_<mode>_mask<round_name>,
7823 <avx512>_fnmadd_<mode>_mask3<round_name>,
7824 <avx512>_fnmsub_<mode>_mask<round_name>,
7825 <avx512>_fnmsub_<mode>_mask3<round_name>,
7826 <avx512>_fmaddsub_<mode>_mask<round_name>,
7827 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7828 <avx512>_fmsubadd_<mode>_mask<round_name>,
7829 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7830 (*vec_widen_umult_even_v16si<mask_name>,
7831 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7832 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7834 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7836 * extend.texi: Add 'nocf_check' documentation.
7837 * gimple.texi: Add second parameter to
7838 gimple_build_call_from_tree.
7839 * invoke.texi: Add -fcf-protection documentation.
7840 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7842 2017-10-20 Richard Biener <rguenther@suse.de>
7844 PR tree-optimization/82473
7845 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7846 the largest input type.
7848 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7850 * c-attribs.c (handle_nocf_check_attribute): New function.
7851 (c_common_attribute_table): Add 'nocf_check' handling.
7852 * gimple-parser.c: Add second argument NULL to
7853 gimple_build_call_from_tree.
7854 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7855 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7857 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7858 * common.opt: Add fcf-protection flag.
7859 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7860 * flag-types.h: Add enum cf_protection_level.
7861 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7862 Add 'nocf_check' attribute propagation to gimple call.
7863 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7864 (gimple_build_call_from_tree): Update prototype.
7865 (gimple_call_nocf_check_p): New function.
7866 (gimple_call_set_nocf_check): Likewise.
7867 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7868 * ipa-icf.c: Add nocf_check attribute in statement hash.
7869 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7870 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7871 * toplev.c (process_options): Add flag_cf_protection handling.
7873 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7875 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7877 2017-10-20 Richard Biener <rguenther@suse.de>
7879 PR tree-optimization/82603
7880 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7881 remove false predicated stores.
7883 2017-10-20 Richard Biener <rguenther@suse.de>
7885 * graphite-isl-ast-to-gimple.c
7886 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7887 Remove return value and simplify, dump copied stmt after lhs
7889 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7890 Reduce dump verbosity.
7891 (gsi_insert_earliest): Likewise.
7892 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7893 * graphite.c (print_global_statistics): Adjust dumping.
7894 (print_graphite_scop_statistics): Likewise.
7895 (print_graphite_statistics): Do not dump loops here.
7896 (graphite_transform_loops): But here.
7898 2017-10-20 Nicolas Roche <roche@adacore.com>
7900 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7901 * configure: Regenerate.
7903 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7906 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7907 functions when optimizing replace GIMPLE_RETURN stmts with
7908 calls to __builtin_unreachable ().
7911 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7912 for -fsanitize=thread link of executables.
7913 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7914 link of executables.
7917 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7919 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7920 define_insns for logical vector shifts to use VI248_AVX512BW
7921 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7922 condition, useless isa and prefix attributes. Change the first
7923 2 of these define_insns to ...
7924 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7925 define_insn for avx512vl.
7926 (<shift_insn><mode>3): ... and this, new define_insn without
7927 masking for non-avx512vl.
7930 * config/i386/sse.md (*andnot<mode>3,
7931 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7932 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7933 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7934 not applied use empty suffix even for TARGET_AVX512VL.
7935 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7936 is applied, supply evex,evex or evex,evex,evex instead of just
7939 2017-10-20 Julia Koval <julia.koval@intel.com>
7941 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7942 (OPTION_MASK_ISA_GFNI_UNSET): New.
7943 (ix86_handle_option): Handle OPT_mgfni.
7944 * config/i386/cpuid.h (bit_GFNI): New.
7945 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7946 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7947 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7948 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7949 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7950 * config/i386/i386.opt: Add mgfni.
7952 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7954 * config/msp430/msp430.c (msp430_option_override): Disable
7955 -fdelete-null-pointer-checks.
7956 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7958 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7960 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7961 of x87 and SSE instructions.
7963 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7965 * asan.c (create_cond_insert_point): Do not update edge count.
7966 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7967 (afdo_propagate_circuit): Likewise.
7968 (afdo_calculate_branch_prob): Likewise.
7969 (afdo_annotate_cfg): Likewise.
7970 * basic-block.h (struct edge_def): Remove count.
7971 (edge_def::count): New accessor.
7972 * bb-reorder.c (rotate_loop): Update.
7973 (find_traces_1_round): Update.
7974 (connect_traces): Update.
7975 (sanitize_hot_paths): Update.
7976 * cfg.c (unchecked_make_edge): Update.
7977 (make_single_succ_edge): Update.
7978 (check_bb_profile): Update.
7979 (dump_edge_info): Update.
7980 (update_bb_profile_for_threading): Update.
7981 (scale_bbs_frequencies_int): Update.
7982 (scale_bbs_frequencies_gcov_type): Update.
7983 (scale_bbs_frequencies_profile_count): Update.
7984 (scale_bbs_frequencies): Update.
7985 * cfganal.c (connect_infinite_loops_to_exit): Update.
7986 * cfgbuild.c (compute_outgoing_frequencies): Update.
7987 (find_many_sub_basic_blocks): Update.
7988 * cfgcleanup.c (try_forward_edges): Update.
7989 (try_crossjump_to_edge): Update
7990 * cfgexpand.c (expand_gimple_cond): Update
7991 (expand_gimple_tailcall): Update
7992 (construct_exit_block): Update
7993 * cfghooks.c (verify_flow_info): Update
7994 (redirect_edge_succ_nodup): Update
7995 (split_edge): Update
7996 (make_forwarder_block): Update
7997 (duplicate_block): Update
7998 (account_profile_record): Update
7999 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8000 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8001 * cfgloopmanip.c (scale_loop_profile): Update.
8003 (lv_adjust_loop_entry_edge): Update.
8004 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8005 (force_nonfallthru_and_redirect): Update.
8006 (purge_dead_edges): Update.
8007 (rtl_flow_call_edges_add): Update.
8008 * cgraphunit.c (init_lowered_empty_function): Update.
8009 (cgraph_node::expand_thunk): Update.
8010 * gimple-pretty-print.c (dump_probability): Update.
8011 (dump_edge_probability): Update.
8012 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8013 * haifa-sched.c (sched_create_recovery_edges): Update.
8014 * hsa-gen.c (convert_switch_statements): Update.
8015 * ifcvt.c (dead_or_predicable): Update.
8016 * ipa-inline-transform.c (inline_transform): Update.
8017 * ipa-split.c (split_function): Update.
8018 * ipa-utils.c (ipa_merge_profiles): Update.
8019 * loop-doloop.c (add_test): Update.
8020 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8021 * lto-streamer-in.c (input_cfg): Update.
8022 (input_function): Update.
8023 * lto-streamer-out.c (output_cfg): Update.
8024 * modulo-sched.c (sms_schedule): Update.
8025 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8026 * predict.c (maybe_hot_edge_p): Update.
8027 (unlikely_executed_edge_p): Update.
8028 (probably_never_executed_edge_p): Update.
8029 (dump_prediction): Update.
8030 (drop_profile): Update.
8031 (propagate_unlikely_bbs_forward): Update.
8032 (determine_unlikely_bbs): Update.
8033 (force_edge_cold): Update.
8034 * profile.c (compute_branch_probabilities): Update.
8035 * reg-stack.c (better_edge): Update.
8036 * shrink-wrap.c (handle_simple_exit): Update.
8037 * tracer.c (better_p): Update.
8038 * trans-mem.c (expand_transaction): Update.
8039 (split_bb_make_tm_edge): Update.
8040 * tree-call-cdce.c: Update.
8041 * tree-cfg.c (gimple_find_sub_bbs): Update.
8042 (gimple_split_edge): Update.
8043 (gimple_duplicate_sese_region): Update.
8044 (gimple_duplicate_sese_tail): Update.
8045 (gimple_flow_call_edges_add): Update.
8046 (insert_cond_bb): Update.
8047 (execute_fixup_cfg): Update.
8048 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8049 * tree-complex.c (expand_complex_div_wide): Update.
8050 * tree-eh.c (lower_resx): Update.
8051 (unsplit_eh): Update.
8052 (cleanup_empty_eh_move_lp): Update.
8053 * tree-inline.c (copy_edges_for_bb): Update.
8054 (freqs_to_counts): Update.
8055 (copy_cfg_body): Update.
8056 * tree-ssa-dce.c (remove_dead_stmt): Update.
8057 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8058 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8059 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8060 (unloop_loops): Update.
8061 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8062 * tree-ssa-loop-split.c (connect_loops): Update.
8063 (split_loop): Update.
8064 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8065 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8066 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8067 * tree-ssa-reassoc.c (branch_fixup): Update.
8068 * tree-ssa-tail-merge.c (replace_block_by): Update.
8069 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8070 (compute_path_counts): Update.
8071 (update_profile): Update.
8072 (recompute_probabilities): Update.
8073 (update_joiner_offpath_counts): Update.
8074 (estimated_freqs_path): Update.
8075 (freqs_to_counts_path): Update.
8076 (clear_counts_path): Update.
8077 (ssa_fix_duplicate_block_edges): Update.
8078 (duplicate_thread_path): Update.
8079 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8080 (case_bit_test_cmp): Update.
8081 (collect_switch_conv_info): Update.
8082 (gen_inbound_check): Update.
8083 (do_jump_if_equal): Update.
8084 (emit_cmp_and_jump_insns): Update.
8085 * tree-tailcall.c (decrease_profile): Update.
8086 (eliminate_tail_call): Update.
8087 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8088 (vect_do_peeling): Update.
8089 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8090 * ubsan.c (ubsan_expand_null_ifn): Update.
8091 (ubsan_expand_ptr_ifn): Update.
8092 * value-prof.c (gimple_divmod_fixed_value): Update.
8093 (gimple_mod_pow2): Update.
8094 (gimple_mod_subtract): Update.
8095 (gimple_ic): Update.
8096 (gimple_stringop_fixed_value): Update.
8098 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8101 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8103 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8105 PR rtl-optimization/82395
8106 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8107 based on non_spilled_static_chain_regno_p.
8109 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8111 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8112 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8113 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8115 2017-10-19 Martin Sebor <msebor@redhat.com>
8117 PR tree-optimization/82596
8118 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8120 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8122 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8123 (get_mem_refs_of_builtin_call): Likewise.
8124 * builtins.c (expand_builtin_apply): Adjust call to
8125 allocate_dynamic_stack_space.
8126 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8127 the third argument to allocate_dynamic_stack_space, otherwise -1.
8128 (expand_builtin): Deal with all alloca variants.
8129 (is_inexpensive_builtin): Likewise.
8130 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8131 * calls.c (special_function_p): Deal with all alloca variants.
8132 (initialize_argument_information): Adjust call to
8133 allocate_dynamic_stack_space.
8134 (expand_call): Likewise.
8135 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8136 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8137 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8138 use it for the stack usage computation.
8139 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8140 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8141 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8142 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8143 (in_loop_p): Remove first argument and useless check.
8144 (pass_walloca::execute): Remove useless test and adjust call to above.
8145 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8146 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8147 (gimplify_call_expr): Deal with all alloca variants.
8148 * hsa-gen.c (gen_hsa_alloca): Likewise.
8149 (gen_hsa_insns_for_call): Likewise.
8150 * ipa-pure-const.c (special_builtin_state): Likewise.
8151 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8152 * tree-object-size.c (alloc_object_size): Likewise.
8153 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8154 (call_may_clobber_ref_p_1): Likewise.
8155 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8156 (ccp_fold_stmt): Likewise.
8157 (optimize_stack_restore): Likewise.
8158 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8159 (mark_all_reaching_defs_necessary_1): Likewise.
8160 (propagate_necessity): Likewise.
8161 (eliminate_unnecessary_stmts): Likewise.
8162 * tree.c (build_common_builtin_nodes): Build
8163 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8164 (build_alloca_call_expr): New function.
8165 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8166 (CASE_BUILT_IN_ALLOCA): Likewise.
8167 (build_alloca_call_expr): Declare.
8168 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8170 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8173 * dwarf2out.c (new_die_raw): New static inline function.
8174 (new_die): Use it to create the DIE.
8175 (add_AT_external_die_ref): Likewise.
8176 (clone_die): Likewise.
8177 (clone_as_declaration): Likewise.
8178 (dwarf2out_vms_debug_main_pointer): Likewise.
8179 (base_type_die): Likewise. Remove early return for corner cases.
8180 Do not call add_pubtype on the DIE here.
8181 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8182 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8183 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8184 native order exists for base types, attach the DIE manually and call
8185 add_pubtype on it. Do not equate a reverse order DIE to the type.
8187 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8189 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8190 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8192 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8194 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8196 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8197 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8198 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8199 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8200 * opts.c (sanitizer_opts): Add builtin.
8201 * ubsan.c (instrument_builtin): New function.
8202 (pass_ubsan::execute): Call it.
8203 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8204 * doc/invoke.texi: Document -fsanitize=builtin.
8206 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8207 builtins, store max (log2 (align), 0) into uchar field instead of
8208 align into uptr field.
8209 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8210 store uchar 0 field instead of uptr 0 field.
8211 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8212 instead of passing one address of struct with 2 locations pass
8213 two addresses of structs with 1 location each.
8214 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8215 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8216 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8217 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8218 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8219 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8220 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8221 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8223 2017-10-19 Martin Liska <mliska@suse.cz>
8226 * file-find.c (remove_prefix): Remove.
8227 * file-find.h (remove_prefix): Likewise.
8228 * gcc-ar.c: Remove smartness of lookup.
8230 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8232 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8233 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8234 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8236 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8239 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8240 (setcc + and to xor + setcc): New peephole2.
8242 2017-10-19 Tom de Vries <tom@codesourcery.com>
8244 * doc/sourcebuild.texi (Test Directives, Variants of
8245 dg-require-support): Add dg-require-stack-size.
8247 2017-10-19 Martin Liska <mliska@suse.cz>
8250 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8251 that have a large alignment.
8252 (gimplify_target_expr): Likewise.
8254 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8256 PR rtl-optimization/82602
8257 * ira.c (rtx_moveable_p): Return false for volatile asm.
8259 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8262 * config/i386/i386-modes.def (CCGZ): New CC mode.
8263 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8264 * config/i386/predicates.md (ix86_comparison_operator):
8266 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8267 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8268 with double-word subtraction.
8269 (put_condition_code): Handle CCGZmode.
8271 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8273 * wide-int.cc (debug (const wide_int &)): New.
8274 (debug (const wide_int *)): New.
8275 (debug (const widest_int &)): New.
8276 (debug (const widest_int *)): New.
8278 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8281 * lra-constraints.c (curr_insn_transform): Use non-input operand
8282 instead of output one for matched reload.
8284 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8286 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8287 (tree-ssa-loop-ivopts.h): New header file.
8288 (struct builtin_info): New fields.
8289 (classify_builtin_1): Compute and record base and offset parts for
8290 memset builtin partition by calling strip_offset.
8291 (offset_cmp, fuse_memset_builtins): New functions.
8292 (finalize_partitions): Fuse adjacent memset partitions by calling
8294 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8295 Expose the interface.
8296 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8298 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8300 PR tree-optimization/82574
8301 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8302 that data reference must be executed exactly once per iteration
8303 against the outermost loop in nest.
8304 (classify_partition): Update call to above function.
8306 2017-10-18 Richard Biener <rguenther@suse.de>
8308 PR tree-optimization/82591
8309 * graphite.c (graphite_transform_loops): Move code gen message
8311 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8312 Here. Handle scop_to_isl_ast failing.
8313 (scop_to_isl_ast): Limit the number of ISL operations.
8315 2017-10-18 Richard Biener <rguenther@suse.de>
8317 * graphite-isl-ast-to-gimple.c
8318 (translate_isl_ast_to_gimple::set_rename): Simplify.
8319 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8320 (graphite_copy_stmts_from_block): ... here.
8321 (copy_bb_and_scalar_dependences): Simplify.
8322 (add_parameters_to_ivs_params): Canonicalize.
8323 (generate_entry_out_of_ssa_copies): Simplify.
8324 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8325 by passing in ISL dimension.
8326 (parameter_index_in_region_1): Rename to ...
8327 (parameter_index_in_region): ... this.
8328 (extract_affine): Adjust assert, pass down parameter index.
8329 (add_param_constraints): Use range-info when available.
8330 (build_scop_context): Adjust.
8331 * sese.c (new_sese_info): Adjust.
8332 (free_sese_info): Likewise.
8333 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8334 Remove unused typedefs.
8335 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8337 2017-10-18 Martin Liska <mliska@suse.cz>
8339 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8341 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8343 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8344 (s390_sched_init): Do not reset s390_sched_state if we entered the
8345 current basic block via a fallthru edge and all others are unlikely.
8347 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8349 * config/s390/s390.c (NUM_SIDES): New variable.
8350 (LONGRUNNING_THRESHOLD): New variable.
8351 (LATENCY_FACTOR): New variable.
8352 (s390_sched_score): Decrease score for long-running instructions on
8354 (s390_sched_variable_issue): Perform bookkeeping for long-running
8357 2017-10-18 Richard Biener <rguenther@suse.de>
8359 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8360 Simplify with removal of the parameter rename map.
8361 (set_rename): Likewise.
8362 (should_copy_to_new_region): Likewise.
8363 (graphite_copy_stmts_from_block): Likewise.
8364 (copy_bb_and_scalar_dependences): Remove initialization of
8365 unused copied_bb_map.
8367 (copy_internal_parameters): Likewise.
8368 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8369 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8370 Use INTEGRAL_TYPE_P.
8371 (parameter_index_in_region_1): Rename to ...
8372 (assign_parameter_index_in_region): ... this. Assert we have
8373 a parameter we handle.
8374 (scan_tree_for_params): Adjust.
8375 * sese.h (parameter_rename_map_t): Remove.
8376 (struct sese_info_t): Remove unused parameter_rename_map and
8377 copied_bb_map members.
8378 * sese.c (new_sese_info): Adjust.
8379 (free_sese_info): Likewise.
8381 2017-10-18 Martin Liska <mliska@suse.cz>
8384 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8385 on an abnormal edge.
8387 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8389 * doc/invoke.texi (ffunction-sections and fdata-sections):
8392 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8394 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8395 the use statement can throw internally.
8397 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8399 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8400 any RTX present on the RHS of a SET.
8401 * compare-elim.c (try_eliminate_compare): Restore comment.
8403 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8405 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8406 may be also called on tcc_type nodes.
8407 * langhooks.c (lhd_tree_size): Likewise.
8409 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8411 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8412 format_warning_at_substring.
8413 (maybe_warn): Convert source_range * param to a location_t. Pass
8414 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8415 (format_directive): Remove code to extract source_ranges and
8416 source_range * in favor of just a location_t.
8417 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8419 * substring-locations.c (format_warning_va): Convert
8420 source_range * param to a location_t.
8421 (format_warning_at_substring): Likewise.
8422 * substring-locations.h (format_warning_va): Likewise.
8423 (format_warning_at_substring): Likewise.
8425 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8427 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8429 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8430 and scatter/gather ops.
8432 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8433 vec_gather_load and vec_scatter_store.
8434 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8435 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8436 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8437 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8438 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8439 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8441 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8443 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8444 Detect FTST instruction and handle its register pops. Only pop
8445 second operand if can_pop_second_op is true.
8446 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8447 set can_pop_second_op to false in the compare_for_stack_reg call.
8449 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8450 output_fp_compare for stack register operands.
8451 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8452 instructions here. Do not emit stack register pops here. Assert
8453 that FCOMPP pops next to top stack register. Rewrite function.
8455 2017-10-17 Nathan Sidwell <nathan@acm.org>
8458 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8459 use DECL_ASSEMBLER_NAME_RAW.
8462 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8465 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8466 Wilco Dijkstra <wilco.dijkstra@arm.com>
8468 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8469 converstion to Pmode from the buf_addr.
8471 2017-10-17 Richard Biener <rguenther@suse.de>
8473 * graphite-dependences.c (scop_get_reads_and_writes): Change
8474 output parameters to references.
8476 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8478 PR 71026/tree-optimization
8479 * fold-const.c (distribute_real_division): Removed.
8480 (fold_binary_loc): Remove calls to distribute_real_divison.
8482 2017-10-17 Richard Biener <rguenther@suse.de>
8484 * graphite-scop-detection.c
8485 (scop_detection::stmt_has_simple_data_refs_p): Always use
8486 the full nest as region.
8487 (try_generate_gimple_bb): Likewise.
8488 * sese.c (scalar_evolution_in_region): Simplify now that
8489 SCEV can handle instantiation in regions.
8490 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8491 in the non-loop part of a function if requested.
8493 2017-10-17 Richard Biener <rguenther@suse.de>
8495 PR tree-optimization/82563
8496 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8498 (graphite_regenerate_ast_isl): Call it.
8499 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8501 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8503 PR tree-optimization/82549
8504 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8505 Formatting fixes. Instead of calling make_bit_field_ref with negative
8508 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8510 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8511 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8512 _mm_maskz_reduce_ss): New.
8513 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8514 __builtin_ia32_reducess_mask): Ditto..
8515 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8516 * config/i386/sse.md (reduces<mode>): Renamed to ...
8517 (reduces<mode><mask_scalar_name>): ... this.
8518 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8520 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8521 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8523 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8525 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8526 * selftest-run-tests.c (selftest::run_tests): Call
8527 selftest::unique_ptr_tests_cc_tests.
8528 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8529 * unique-ptr-tests.cc: New file.
8531 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8534 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8536 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8537 (make_hard_regno_born, make_hard_regno_dead): Update
8538 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8540 2017-10-16 Jeff Law <law@redhat.com>
8542 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8544 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8546 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8548 2017-10-16 Olivier Hainque <hainque@adacore.com>
8550 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8551 with_cpu if we were configured for an e500v2 target cpu name.
8553 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8555 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8556 * doc/invoke.texi: Document +nodsp as a valid extension for
8559 2017-10-16 Martin Liska <mliska@suse.cz>
8561 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8562 (test_set_range): Likewise.
8563 (test_range_functions): Rename to ...
8564 (test_bit_in_range): ... this.
8565 (sbitmap_c_tests): Add new test.
8567 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8569 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8571 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8572 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8574 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8576 * config/aarch64/aarch64-builtins.c
8577 (aarch64_types_quadopu_lane_qualifiers): New.
8578 (TYPES_QUADOPU_LANE): New.
8579 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8580 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8581 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8582 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8583 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8584 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8585 (Vdottype, DOTPROD): New.
8586 (sur): Add SDOT and UDOT.
8588 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8590 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8591 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8592 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8594 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8595 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8596 Enable TARGET_DOTPROD.
8597 (cortex-a75.cortex-a55): Likewise.
8598 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8600 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8602 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8603 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8605 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8607 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8608 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8609 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8610 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8611 * config/arm/types.md (neon_dot, neon_dot_q): New.
8612 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8614 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8616 * config/arm/arm.h (TARGET_DOTPROD): New.
8617 * config/arm/arm.c (arm_arch_dotprod): New.
8618 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8619 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8620 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8621 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8622 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8623 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8624 * doc/invoke.texi (armv8.2-a): Document dotprod
8626 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8628 * i386.c (ix86_vec_cost): New function.
8629 (ix86_rtx_costs): Handle vector operations better.
8630 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8631 * x86-tune-costs.h: Add new costs to all tables.
8633 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8635 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8637 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8638 divsd, sqrtss and sqrtsd
8639 * x86-tune-costs.h: Add new entries to all costs.
8640 (znver1_cost): Fix to match real instruction latencies.
8642 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8643 Michael Collison <michael.collison@arm.com>
8645 * compare-elim.c: Include emit-rtl.h.
8646 (can_merge_compare_into_arith): New function.
8647 (try_validate_parallel): Likewise.
8648 (try_merge_compare): Likewise.
8649 (try_eliminate_compare): Call the above when no previous clobber
8651 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8654 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8658 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8659 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8663 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8664 to be any operand_equal_p operands. For & (B - 1) require
8665 B to be power of 2. Recognize
8666 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8668 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8671 * optabs.c (expand_memory_blockage): Fix call of
8672 targetm.have_memory_blockage.
8674 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8677 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8678 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8679 objects to extra_objs instead of overwriting it.
8681 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8683 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8684 Use any_fp_register_operand as operand[3] predicate. Simplify
8685 equality test for operands[2] and operands[4] memory location.
8686 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8687 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8688 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8689 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8690 any_fp_register_operand as operand[1] predicate. Simplify
8691 equality test for operands[0] and operands[3] memory location.
8692 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8693 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8694 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8696 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8698 * target-insns.def: Add memory_blockage.
8699 * optabs.c (expand_memory_blockage): New function.
8700 (expand_asm_memory_barrier): Rename ...
8701 (expand_asm_memory_blockage): ... to this.
8702 (expand_mem_thread_fence): Call expand_memory_blockage
8703 instead of expand_asm_memory_barrier.
8704 (expand_mem_singnal_fence): Ditto.
8705 (expand_atomic_load): Ditto.
8706 (expand_atomic_store): Ditto.
8707 * doc/md.texi (Standard Pattern Names For Generation):
8708 Document memory_blockage instruction pattern.
8710 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8712 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8713 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8714 * config/rl78/rl78.md: New define_expand "adddi3".
8716 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8718 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8721 2017-10-13 Jeff Law <law@redhat.com>
8723 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8725 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8728 * internal-fn.c (expand_mul_overflow): If both operands have
8729 the same highpart of -1 or 0 and the topmost bit of lowpart
8730 is different, overflow is if res <= 0 rather than res < 0.
8732 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8734 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8735 TARGET_P9_VECTOR code for unaligned_load case.
8737 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8739 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8741 2017-10-13 Nathan Sidwell <nathan@acm.org>
8743 * tree-core.h (tree_contains_struct): Make bool.
8744 * tree.c (tree_contains_struct): Likewise.
8745 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8746 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8747 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8748 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8749 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8751 2017-10-13 Richard Biener <rguenther@suse.de>
8753 * graphite-isl-ast-to-gimple.c
8754 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8755 parameters and dominance check.
8756 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8757 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8758 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8759 Do not update SSA form here or do intermediate IL verification.
8760 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8761 (graphite_initialize): Remove check on the number of loops in
8762 the function and inline into graphite_transform_loops.
8763 (graphite_finalize): Inline into graphite_transform_loops.
8764 (graphite_transform_loops): Perform SSA update and IL verification
8766 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8768 2017-10-13 Richard Biener <rguenther@suse.de>
8770 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8771 graphite_expression_type_precision): Avoid global constructor
8773 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8774 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8775 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8776 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8777 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8778 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8780 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8783 * config/i386/i386.h (ix86_red_zone_size): New.
8784 * config/i386/i386.md (push peephole2s): Replace
8785 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8787 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8788 Alan Hayward <alan.hayward@arm.com>
8789 David Sherwood <david.sherwood@arm.com>
8791 * combine.c (can_change_dest_mode): Reject changes in
8792 REGMODE_NATURAL_SIZE.
8794 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8795 Alan Hayward <alan.hayward@arm.com>
8796 David Sherwood <david.sherwood@arm.com>
8798 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8799 (expand_debug_source_expr): Likewise.
8800 * combine.c (combine_simplify_rtx): Likewise.
8801 * cse.c (fold_rtx): Likewise.
8802 * fwprop.c (canonicalize_address): Likewise.
8803 * targhooks.c (default_shift_truncation_mask): Likewise.
8805 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8806 Alan Hayward <alan.hayward@arm.com>
8807 David Sherwood <david.sherwood@arm.com>
8809 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8810 (widened_mode): Likewise.
8811 (expand_unop): Likewise.
8812 * ree.c (transform_ifelse): Likewise.
8813 (merge_def_and_ext): Likewise.
8814 (combine_reaching_defs): Likewise.
8815 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8817 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8818 Alan Hayward <alan.hayward@arm.com>
8819 David Sherwood <david.sherwood@arm.com>
8821 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8822 * combine.c (gen_lowpart_for_combine): Likewise.
8823 * dwarf2out.c (rtl_for_decl_location): Likewise.
8824 * final.c (alter_subreg): Likewise.
8825 * rtlhooks.c (gen_lowpart_general): Likewise.
8826 (gen_lowpart_if_possible): Likewise.
8828 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8829 Alan Hayward <alan.hayward@arm.com>
8830 David Sherwood <david.sherwood@arm.com>
8832 * calls.c (expand_call): Use subreg_lowpart_offset.
8833 * cse.c (cse_insn): Likewise.
8834 * regcprop.c (copy_value): Likewise.
8835 (copyprop_hardreg_forward_1): Likewise.
8837 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8840 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8841 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8842 =Q constraints to +Q and into insn condition add check
8843 that operands[0] and operands[1] are equal.
8844 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8845 =Q constraints to +Q and into insn condition add check
8846 that operands[0] is equal to either operands[1] or operands[2].
8849 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8850 instead of handling MINUS_EXPR twice (once for each argument),
8851 canonicalize operand order and handle just once, use rtype where
8852 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8855 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8856 any values of __C while still being pattern recognizable as a simple
8859 2017-10-13 Richard Biener <rguenther@suse.de>
8861 PR tree-optimization/82451
8863 2017-10-02 Richard Biener <rguenther@suse.de>
8865 PR tree-optimization/82355
8866 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8867 a mapping for the enclosing loop but avoid generating one for
8869 (copy_bb_and_scalar_dependences): Remove premature codegen
8870 error on PHIs in blocks duplicated into multiple places.
8871 * graphite-scop-detection.c
8872 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8873 in the region use it as loop and nest to analyze the DR in.
8874 (try_generate_gimple_bb): Likewise.
8875 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8876 (add_loop_constraints): For blocks in a loop not in the region
8877 create a dimension with a single iteration.
8878 * sese.h (gbb_loop_at_index): Remove assert.
8880 * cfgloop.c (loop_preheader_edge): For the loop tree root
8881 return the single successor of the entry block.
8882 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8883 Reset the SCEV hashtable and niters.
8884 * graphite-scop-detection.c
8885 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8886 assert that we only have POLYNOMIAL_CHREC that vary in loops
8887 contained in the region.
8888 (scop_detection::graphite_can_represent_expr): Adjust.
8889 (scop_detection::stmt_has_simple_data_refs_p): For loops
8890 not in the region set loop to NULL. The nest is now the
8891 entry edge to the region.
8892 (try_generate_gimple_bb): Likewise.
8893 * sese.c (scalar_evolution_in_region): Adjust for
8894 instantiate_scev change.
8895 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8896 Make nest parameter the edge into the region.
8897 (create_data_ref): Likewise.
8898 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8899 entry edge into a region and adjust instantiate_scev calls.
8900 (create_data_ref): Likewise.
8901 (graphite_find_data_references_in_stmt): Likewise.
8902 (find_data_references_in_stmt): Pass the loop preheader edge
8903 from the nest argument.
8904 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8905 parameter the edge into the region.
8906 (instantiate_parameters): Use the loop preheader edge as entry.
8907 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8909 (get_instantiated_value_entry): Make instantiate_below parameter
8910 the edge into the region.
8911 (instantiate_scev_name): Likewise. Adjust dominance checks,
8912 when we cannot use loop-based instantiation instantiate by
8913 walking use-def chains.
8914 (instantiate_scev_poly): Adjust.
8915 (instantiate_scev_binary): Likewise.
8916 (instantiate_scev_convert): Likewise.
8917 (instantiate_scev_not): Likewise.
8918 (instantiate_array_ref): Remove.
8919 (instantiate_scev_3): Likewise.
8920 (instantiate_scev_2): Likewise.
8921 (instantiate_scev_1): Likewise.
8922 (instantiate_scev_r): Do not blindly handle N-operand trees.
8923 Do not instantiate array-refs. Handle all constants and invariants.
8924 (instantiate_scev): Make instantiate_below parameter
8925 the edge into the region.
8926 (resolve_mixers): Use the loop preheader edge for the region
8927 parameter to instantiate_scev_r.
8928 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8930 2017-10-13 Richard Biener <rguenther@suse.de>
8932 PR tree-optimization/82525
8933 * graphite-isl-ast-to-gimple.c
8934 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8936 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8937 Fail code generation when we cannot represent the isl integer.
8938 (binary_op_to_tree): Elide modulo operations that are no-ops
8939 in the type we code generate. Remove now superfluous code
8942 2017-10-13 Richard Biener <rguenther@suse.de>
8944 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8945 (scop_detection::harmful_loop_in_region): Remove premature
8946 IV type restriction.
8947 (scop_detection::graphite_can_represent_scev): We can handle
8948 pointer IVs just fine.
8950 2017-10-13 Alan Modra <amodra@gmail.com>
8952 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8953 "Clobbers and Scratch Registers". Add paragraph on
8954 alternative to clobbers for scratch registers and OpenBLAS
8957 2017-10-13 Alan Modra <amodra@gmail.com>
8959 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8960 example of a memory input for a string of known length. Move
8961 commentary out of table. Add a number of new examples
8962 covering array memory inputs.
8964 2017-10-12 Martin Liska <mliska@suse.cz>
8966 PR tree-optimization/82493
8967 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8968 (test_range_functions): New function.
8969 (sbitmap_c_tests): Likewise.
8970 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8971 * selftest.h (sbitmap_c_tests): New function.
8973 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8975 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8977 * config/rs6000/amo.h: Fix spacing issue.
8979 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8982 * config/i386/i386.md (*ashl<mode>3_mask_1,
8983 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8984 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8987 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8989 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8990 (profile_probability): Set max_probability
8991 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8993 * profile-count.c (profile_probability::differs_from_p): Do not
8994 rely on max_probaiblity == 10000
8996 2017-10-12 Jeff Law <law@redhat.com>
8998 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9001 2017-10-12 Martin Sebor <msebor@redhat.com>
9005 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9006 (handle_alias_pairs): Call it.
9007 * common.opt (-Wattribute-alias): New option.
9008 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9009 * doc/invoke.texi (-Wattribute-alias): Document.
9011 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9014 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9016 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9018 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9019 (make_hard_regno_born, make_hard_regno_dead): Update
9020 bb_killed_pseudos and bb_gen_pseudos.
9022 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9024 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9026 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9028 * config/alpha/alpha.c (alpha_split_conditional_move):
9029 Use std::swap instead of manually swapping.
9030 (alpha_stdarg_optimize_hook): Ditto.
9031 (alpha_canonicalize_comparison): Ditto.
9033 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9035 * tree-loop-distribution.c (struct builtin_info): New struct.
9036 (struct partition): Refactor fields into struct builtin_info.
9037 (partition_free): Free struct builtin_info.
9038 (build_size_arg_loc, build_addr_arg_loc): Delete.
9039 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9040 information from struct builtin_info.
9041 (find_single_drs): New function refactored from classify_partition.
9042 Also moved builtin validity checks to this function.
9043 (compute_access_range, alloc_builtin): New functions.
9044 (classify_builtin_st, classify_builtin_ldst): New functions.
9045 (classify_partition): Refactor code into functions find_single_drs,
9046 classify_builtin_st and classify_builtin_ldst.
9047 (distribute_loop): Don't do runtime alias check when distributing
9049 (find_seed_stmts_for_distribution): New function.
9050 (pass_loop_distribution::execute): Refactor code finding seed
9051 stmts into above function. Support distribution for the innermost
9052 two-level loop nest. Adjust dump information.
9054 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9056 * tree-loop-distribution.c: Adjust the general comment.
9057 (NUM_PARTITION_THRESHOLD): New macro.
9058 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9059 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9060 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9061 in call to build_partition_graph.
9062 (finalize_partitions): New parameter. Make loop distribution more
9063 conservative by fusing more partitions.
9064 (distribute_loop): Don't do runtime alias check in case of loop nest
9066 (find_seed_stmts_for_distribution): New function.
9067 (prepare_perfect_loop_nest): New function.
9068 (pass_loop_distribution::execute): Refactor code finding seed stmts
9069 and loop nest into above functions. Support loop nest distribution.
9070 Adjust dump information accordingly.
9072 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9074 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9075 and set PTYPE_SEQUENTIAL for merged partition.
9077 2017-10-12 Richard Biener <rguenther@suse.de>
9079 PR tree-optimization/69728
9081 2017-09-19 Richard Biener <rguenther@suse.de>
9083 PR tree-optimization/69728
9084 * graphite-sese-to-poly.c (schedule_error): New global.
9085 (add_loop_schedule): Handle empty domain by failing the
9087 (build_original_schedule): Handle schedule_error.
9089 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9090 domain by returning an unchanged schedule.
9092 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9094 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9095 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9097 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9099 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9102 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9105 * expr.c (store_field): Don't optimize away bitsize == 0 store
9106 from CALL_EXPR with addressable return type.
9108 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9110 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9111 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9112 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9113 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9114 TARGET_ISEL instead of TARGET_ISEL<sel>.
9116 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9118 * config/rs6000/rs6000.c
9119 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9121 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9123 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9125 (reg_or_cint_operand): Fix comment.
9126 (reg_or_zero_operand): New predicate.
9127 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9128 * config/rs6000/rs6000.c (output_isel): Delete.
9129 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9130 instead of reg_or_cint_operand. Output instruction directly (not via
9132 (isel_unsigned_<mode>): Ditto.
9133 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9134 gpc_reg_operand. Add an instruction alternative for this. Output
9135 instruction directly.
9136 (*isel_reversed_unsigned_<mode>): Ditto.
9138 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9140 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9141 (TARGET_CANONICALIZE_COMPARISON): Define.
9143 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9146 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9147 Check whether the dest is REG before adding REG_EQUIV note.
9149 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9152 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9154 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9155 (make_hard_regno_born, make_hard_regno_dead): Update
9156 bb_killed_pseudos and bb_gen_pseudos.
9158 2017-10-11 Nathan Sidwell <nathan@acm.org>
9160 * incpath.h (enum incpath_kind): Name enum, prefix values.
9161 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9162 * incpath.c (heads, tails): Use INC_MAX.
9163 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9164 (merge_include_chains, split_quote_chain,
9165 register_include_chains): Update incpath_kind names.
9166 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9167 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9169 (add_framework_path, darwin_register_objc_includes): Likewise.
9170 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9172 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9174 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9175 Do not use float_operator operator predicate.
9176 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9177 * config/i386/predicates.md (float_operator): Remove predicate.
9179 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9181 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9182 (*jccxf_i387): Ditto.
9183 (*jcc<mode>_i387): Ditto.
9184 (*jccu<mode>_i387): Ditto.
9185 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9186 (*jcc_*_i387 splitters): Remove.
9187 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9188 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9189 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9192 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9194 * profile-count.h (slow_safe_scale_64bit): New function.
9195 (safe_scale_64bit): New inline.
9196 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9197 * profile-count.c: Include wide-int.h
9198 (slow_safe_scale_64bit): New.
9200 2017-10-11 Nathan Sidwell <nathan@acm.org>
9202 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9203 HAS_DECL_ASSEMBLER_NAME_P.
9204 * gimple-expr.c (gimple_decl_printable_name: Check
9205 HAS_DECL_ASSEMBLER_NAME_P too.
9206 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9207 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9208 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9209 * passes.c (rest_of_decl_compilation): Check
9210 HAS_DECL_ASSEMBLER_NAME_P too.
9211 * recog.c (verify_changes): Likewise.
9212 * tree-pretty-print.c (dump_decl_name): Likewise.
9213 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9215 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9216 (SET_DECL_ASSEMBLER_NAME): Use it.
9217 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9218 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9219 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9221 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9223 * config.gcc (i386, x86_64): Add extra objects.
9224 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9225 (ix86_min_insn_size): Declare.
9226 (ix86_issue_rate): Declare.
9227 (ix86_adjust_cost): Declare.
9228 (ia32_multipass_dfa_lookahead): Declare.
9229 (ix86_macro_fusion_p): Declare.
9230 (ix86_macro_fusion_pair_p): Declare.
9231 (ix86_bd_has_dispatch): Declare.
9232 (ix86_bd_do_dispatch): Declare.
9233 (ix86_core2i7_init_hooks): Declare.
9234 (ix86_atom_sched_reorder): Declare.
9235 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9236 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9237 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9238 (rip_relative_addr_p): Rename to ...
9239 (ix86_rip_relative_addr_p): ... this one; export.
9240 (memory_address_length): Update.
9241 (ix86_issue_rate): Move to x86-tune-sched.c.
9242 (ix86_flags_dependent): Move to x86-tune-sched.c.
9243 (ix86_agi_dependent): Move to x86-tune-sched.c.
9244 (exact_dependency_1): Move to x86-tune-sched.c.
9245 (exact_store_load_dependency): Move to x86-tune-sched.c.
9246 (ix86_adjust_cost): Move to x86-tune-sched.c.
9247 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9248 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9249 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9250 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9251 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9252 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9253 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9254 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9255 (min_insn_size): Rename to ...
9256 (ix86_min_insn_size): ... this one; export.
9257 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9258 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9259 (core2i7_first_cycle_multipass_backtrack): Move to
9260 x86-tune-sched-core.c.
9261 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9262 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9263 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9264 (ix86_avoid_jump_mispredicts): Update.
9265 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9266 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9267 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9268 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9269 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9270 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9271 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9272 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9273 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9274 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9275 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9276 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9277 (BIG): Move to ix86-tune-sched-bd.c.
9278 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9279 (enum insn_path): Move to ix86-tune-sched-bd.c.
9280 (get_mem_group): Move to ix86-tune-sched-bd.c.
9281 (is_cmp): Move to ix86-tune-sched-bd.c.
9282 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9283 (is_branch): Move to ix86-tune-sched-bd.c.
9284 (is_prefetch): Move to ix86-tune-sched-bd.c.
9285 (init_window): Move to ix86-tune-sched-bd.c.
9286 (allocate_window): Move to ix86-tune-sched-bd.c.
9287 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9288 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9289 (process_end_window): Move to ix86-tune-sched-bd.c.
9290 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9291 (find_constant): Move to ix86-tune-sched-bd.c.
9292 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9293 (has_immediate): Move to ix86-tune-sched-bd.c.
9294 (get_insn_path): Move to ix86-tune-sched-bd.c.
9295 (get_insn_group): Move to ix86-tune-sched-bd.c.
9296 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9297 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9298 (add_insn_window): Move to ix86-tune-sched-bd.c.
9299 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9300 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9301 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9302 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9303 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9304 (do_dispatch): Move to ix86-tune-sched-bd.c.
9305 (has_dispatch): Move to ix86-tune-sched-bd.c.
9306 * config/i386/t-i386: Add new object files.
9307 * config/i386/x86-tune-costs.h: New file.
9308 * config/i386/x86-tune-sched-atom.c: New file.
9309 * config/i386/x86-tune-sched-bd.c: New file.
9310 * config/i386/x86-tune-sched-core.c: New file.
9311 * config/i386/x86-tune-sched.c: New file.
9313 2017-10-11 Liu Hao <lh_mouse@126.com>
9315 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9316 the generic version below instead.
9317 (should_colorize): Recognize Windows consoles as terminals
9319 * pretty-print.c [__MINGW32__] (write_all): New function.
9320 [__MINGW32__] (find_esc_head): Likewise.
9321 [__MINGW32__] (find_esc_terminator): Likewise.
9322 [__MINGW32__] (eat_esc_sequence): Likewise.
9323 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9325 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9328 2017-10-11 Richard Biener <rguenther@suse.de>
9330 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9331 Properly call analyze_scalar_evolution with the loop of the stmt.
9333 2017-10-11 Richard Biener <rguenther@suse.de>
9335 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9336 * tree-core.h (tree_base): Add chrec_var union member.
9337 * tree.h (CHREC_VAR): Remove.
9338 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9339 * tree-chrec.h (build_polynomial_chrec): Adjust.
9340 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9341 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9343 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9345 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9346 * match.pd: ... here.
9347 ((T) X == (T) Y): Relax condition.
9349 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9351 PR tree-optimization/82472
9352 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9354 (break_alias_scc_partitions): Update postorder number.
9356 2017-10-11 Martin Liska <mliska@suse.cz>
9359 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9361 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9363 2017-10-11 Martin Liska <mliska@suse.cz>
9365 * print-rtl.c (print_insn): Move declaration of idbuf
9366 to same scope as name.
9368 2017-10-11 Martin Liska <mliska@suse.cz>
9373 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9376 2017-10-11 Martin Liska <mliska@suse.cz>
9379 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9382 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9384 * config.gcc (default_gnu_indirect_function): Default to yes for
9385 arm*-*-linux* with glibc.
9387 2017-10-11 Richard Biener <rguenther@suse.de>
9389 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9390 default-defs and types we do not want to analyze.
9391 (interpret_loop_phi): Replace unreachable code with an assert.
9392 (compute_scalar_evolution_in_loop): Remove and inline ...
9393 (analyze_scalar_evolution_1): ... here, replacing condition with
9394 what makes the intent clearer. Remove handling of cases
9395 get_scalar_evolution now handles.
9397 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9399 PR rtl-optimization/81434
9400 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9401 comment for main loop. In sched_group_found if, also add checks for
9402 pass and min_cost_group.
9404 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9406 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9407 (rs6000_insn_cost): New function.
9408 * config/rs6000/rs6000.md (cost): New attribute.
9410 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9411 H.J. Lu <hongjiu.lu@intel.com>
9415 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9416 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9417 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9418 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9419 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9420 OPTION_MASK_ISA_MMX and some other bit set require both
9421 mmx and the ISAs without the mmx bit.
9422 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9423 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9424 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9425 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9426 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9427 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9428 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9429 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9430 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9431 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9432 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9433 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9434 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9435 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9436 Add OPTION_MASK_ISA_MMX.
9438 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9440 * config.gcc (armv7*-*-freebsd*): New target.
9441 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9443 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9445 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9446 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9449 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9451 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9452 a template parameter.
9453 (WIDE_INT_REF_FOR): Update accordingly.
9454 * tree.h (wi::int_traits <const_tree>): Delete.
9455 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9456 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9457 (wi::tree_to_wide_ref): New typedef.
9458 (wi::to_wide): New function.
9459 * calls.c (get_size_range): Use wi::to_wide when operating on
9461 * cgraph.c (cgraph_node::create_thunk): Likewise.
9462 * config/i386/i386.c (ix86_data_alignment): Likewise.
9463 (ix86_local_alignment): Likewise.
9464 * dbxout.c (stabstr_O): Likewise.
9465 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9466 * expr.c (const_vector_from_tree): Likewise.
9467 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9468 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9469 (fold_negate_expr_1, int_const_binop_1, const_binop)
9470 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9471 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9472 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9473 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9474 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9475 (fold_not_const, round_up_loc): Likewise.
9476 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9477 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9478 (alloca_call_type): Likewise.
9479 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9480 * godump.c (go_output_typedef): Likewise.
9481 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9482 * internal-fn.c (get_min_precision): Likewise.
9483 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9484 * ipa-polymorphic-call.c
9485 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9486 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9487 (ipa_modify_call_arguments): Likewise.
9488 * match.pd: Likewise.
9489 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9490 * print-tree.c (print_node_brief, print_node): Likewise.
9491 * stmt.c (expand_case): Likewise.
9492 * stor-layout.c (layout_type): Likewise.
9493 * tree-affine.c (tree_to_aff_combination): Likewise.
9494 * tree-cfg.c (group_case_labels_stmt): Likewise.
9495 * tree-data-ref.c (dr_analyze_indices): Likewise.
9496 (prune_runtime_alias_test_list): Likewise.
9497 * tree-dump.c (dequeue_and_dump): Likewise.
9498 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9499 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9500 * tree-pretty-print.c (dump_generic_node): Likewise.
9501 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9502 (simple_iv_with_niters): Likewise.
9503 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9504 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9505 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9506 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9507 (refine_value_range_using_guard, number_of_iterations_ne_max)
9508 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9509 (get_cst_init_from_scev, record_nonwrapping_iv)
9510 (scev_var_range_cant_overflow): Likewise.
9511 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9512 * tree-ssa-pre.c (compute_avail): Likewise.
9513 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9514 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9515 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9516 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9517 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9518 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9519 (dump_case_nodes, try_switch_expansion): Likewise.
9520 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9521 (vect_do_peeling): Likewise.
9522 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9523 * tree-vect-stmts.c (vectorizable_load): Likewise.
9524 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9525 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9526 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9527 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9528 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9529 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9530 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9531 (evrp_dom_walker::before_dom_children): Likewise.
9532 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9533 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9534 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9535 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9536 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9537 (get_range_pos_neg): Likewise.
9538 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9539 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9540 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9541 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9542 * config/avr/avr.c (avr_fold_builtin): Likewise.
9543 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9544 * config/msp430/msp430.c (msp430_attr): Likewise.
9545 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9546 * config/powerpcspe/powerpcspe-c.c
9547 (altivec_resolve_overloaded_builtin): Likewise.
9548 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9549 (rs6000_expand_ternop_builtin): Likewise.
9550 * config/rs6000/rs6000-c.c
9551 (altivec_resolve_overloaded_builtin): Likewise.
9552 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9553 (rs6000_expand_ternop_builtin): Likewise.
9554 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9556 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9558 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9559 when copying loop nest with only one inner loop.
9561 2017-10-10 Richard Biener <rguenther@suse.de>
9563 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9564 blocks if SCEV is active.
9565 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9567 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9568 (scev_initialize): Assert we are not yet initialized.
9570 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9572 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9573 inner loop's exit stmt by making it always exit the loop, otherwise
9574 we would generate an infinite empty loop.
9576 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9578 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9579 renaming variables in new preheader if it's deleted.
9581 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9583 * tree-loop-distribution.c (struct partition): Remove unused field
9584 loops of the structure.
9585 (partition_alloc, partition_free): Ditto.
9586 (build_rdg_partition_for_vertex): Ditto.
9588 2017-10-09 Jeff Law <law@redhat.com>
9590 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9591 return type to match prototype and documentation.
9593 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9595 * config/rs6000/rs6000.c (processor_costs): Move to ...
9596 * config/rs6000/rs6000.h: ... here.
9597 (rs6000_cost): Declare.
9599 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9601 * except.c (setjmp_fn): New global variable.
9602 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9603 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9604 if DONT_USE_BUILTIN_SETJMP is defined.
9606 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9608 * target.def (insn_cost): New hook.
9609 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9610 * doc/tm.texi: Regenerate.
9611 * rtlanal.c (insn_cost): Use the new hook.
9613 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9615 * combine.c (combine_validate_cost): Compute the new insn_cost,
9616 not just pattern_cost.
9617 (try_combine): Adjust comment.
9619 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9621 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9623 * combine.c (uid_insn_cost): Adjust comment.
9624 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9626 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9627 * dse.c (find_shift_sequence): Ditto.
9628 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9629 (bb_valid_for_noce_process_p): Use pattern_cost.
9630 * rtl.h (insn_rtx_cost): Delete.
9631 (pattern_cost): New prototype.
9632 (insn_cost): New prototype.
9633 * rtlanal.c (insn_rtx_cost): Rename to...
9634 (pattern_cost): ... this.
9637 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9639 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9640 (*jcc<mode>_0_r_i387): Ditto.
9641 (*jccxf_r_i387): Ditto.
9642 (*jcc<mode>_r_i387): Ditto.
9643 (*jccu<mode>_r_i387): Ditto.
9644 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9645 (*jcc): Rename from *jcc_1.
9647 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9649 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9650 deferred rescans after the lvx/stvx recombination pre-pass.
9652 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9654 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9655 memory operation instruction support.
9656 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9657 (rs6000-ibm-aix[789]*): Likewise.
9658 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9659 Document new functions.
9661 2017-10-09 Richard Biener <rguenther@suse.de>
9663 PR tree-optimization/82397
9664 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9665 equality only for semantically equal trees.
9667 2017-10-09 Richard Biener <rguenther@suse.de>
9669 PR tree-optimization/82449
9670 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9671 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9672 allow constant addresses.
9673 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9676 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9678 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9681 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9684 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9687 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9690 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9692 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9695 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9696 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9698 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9700 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9701 (WI_BINARY_PREDICATE_RESULT): Likewise.
9702 (wi::binary_traits::operator_result): New type.
9703 (wi::binary_traits::predicate_result): Likewise.
9704 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9705 (generic_wide_int::operator==, generic_wide_int::operator!=)
9706 (generic_wide_int::operator&, generic_wide_int::and_not)
9707 (generic_wide_int::operator|, generic_wide_int::or_not)
9708 (generic_wide_int::operator^, generic_wide_int::operator+
9709 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9711 (operator~, unary operator-, operator==, operator!=, operator&)
9712 (operator|, operator^, operator+, binary operator-, operator*): New
9714 * expr.c (get_inner_reference): Use wi::bit_and_not.
9715 * fold-const.c (fold_binary_loc): Likewise.
9716 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9717 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9718 (bit_value_binop): Likewise.
9719 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9720 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9721 (extract_range_from_binary_expr_1): Likewise.
9722 (masked_increment): Likewise.
9723 (simplify_bit_ops_using_ranges): Likewise.
9725 2017-10-09 Martin Jambor <mjambor@suse.cz>
9728 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9729 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9730 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9732 (hsa_fixup_mov_insn_type): New function.
9733 (hsa_op_with_type::get_in_type): Use it.
9734 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9735 immediates in an assert.
9736 (hsa_op_with_type::extend_int_to_32bit): New method.
9737 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9738 types. Convert to dest type if necessary.
9739 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9740 (reg_for_gimple_ssa): Pass false as min32int to
9741 hsa_type_for_scalar_tree_type.
9742 (gen_hsa_addr): Fixup type when creating addresable temporary.
9743 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9744 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9745 necessary. Call hsa_fixup_mov_insn_type.
9746 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9747 extend operands and convert to dest type if necessary.
9748 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9749 to dest type if necessary.
9750 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9751 if conversion nt necessary and size matches.
9752 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9753 to dest type if necessary.
9754 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9755 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9757 (gen_hsa_clrsb): Likewise.
9758 (gen_hsa_ffs): Likewise.
9759 (gen_hsa_divmod): Extend operands and convert to dest type if
9761 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9763 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9765 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9766 Remove empty default arguments. Use a brace block as output
9768 (conditional return): Ditto.
9770 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9771 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9773 (group_ending_nop): Ditto.
9774 (doloop_end): Ditto.
9775 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9776 (splitters for those): Ditto.
9778 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9780 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9781 a conditional jump (and the compare for it) so that pc_rtx is the
9783 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9784 for the deleted and renamed ctr<mode>_internal[234] patterns.
9785 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9786 Delete second conditional return pattern.
9787 (ctr<mode>_internal2): Delete this second bdnz pattern.
9788 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9789 (ctr<mode>_internal4): Delete this second bdz pattern.
9791 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9793 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9794 (always_initialized_rtx_for_ssa_name_p): New predicate.
9795 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9796 (finish_out_of_ssa): Free new field of SA.
9797 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9798 * tree-ssa-coalesce.c: Include tree-ssa.h.
9799 (get_parm_default_def_partitions): Remove extern keyword.
9800 (get_undefined_value_partitions): New function.
9801 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9802 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9804 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9806 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9808 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9810 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9812 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9813 for targets that preffer 128bit.
9815 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9817 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9819 2017-10-08 Olivier Hainque <hainque@adacore.com>
9821 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9822 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9824 (thumb_set_return_address): Likewise.
9826 2017-10-08 Olivier Hainque <hainque@adacore.com>
9828 * common/config/arm/arm-common.c (arm_except_unwind_info):
9829 Handle DWARF2_UNWIND_INFO.
9831 2017-10-07 Michael Collison <michael.collison@arm.com>
9833 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9836 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9838 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9839 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9840 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9841 defined, force the creation of a new block for a dispatch label.
9843 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9845 * invoke.texi (Wsuggest-attribute=cold): Document.
9846 * common.opt (Wsuggest-attribute=cold): New
9847 * ipa-pure-const.c (warn_function_cold): New function.
9848 * predict.c (compute_function_frequency): Use it.
9849 * predict.h (warn_function_cold): Declare.
9851 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9853 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9856 2017-10-06 Martin Liska <mliska@suse.cz>
9858 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9859 keyword for member functions.
9860 (struct sanopt_tree_couple): New struct.
9861 (struct sanopt_tree_couple_hash): New function.
9862 (struct sanopt_ctx): Add new hash_map.
9863 (has_dominating_ubsan_ptr_check): New function.
9864 (record_ubsan_ptr_check_stmt): Likewise.
9865 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9866 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9867 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9869 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9872 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9873 aarch64_simd_valid_immediate on CONST_VECTORs.
9874 (aarch64_reg_or_bic_imm): Likewise.
9876 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9878 PR rtl-optimization/82396
9879 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9881 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9883 * graphite-dependences.c (scop_get_reads): Move code to...
9884 (scop_get_must_writes): Move code to...
9885 (scop_get_may_writes): Move code to...
9886 (scop_get_reads_and_writes): ... here.
9887 (scop_get_dependences): Call scop_get_reads_and_writes.
9889 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9891 PR tree-optimization/82434
9892 * fold-const.h (can_native_encode_type_p,
9893 can_native_encode_string_p): Remove.
9894 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9895 don't encode anything, just return what would be otherwise returned.
9896 (native_encode_fixed, native_encode_complex, native_encode_vector):
9898 (native_encode_string): Likewise. Inline by hand
9899 can_native_encode_string_p.
9900 (can_native_encode_type_p): Remove.
9901 (can_native_encode_string_p): Remove.
9902 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9903 STRING_CSTs using can_native_encode_string_p, test all
9904 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9905 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9906 argument from native_encode_expr.
9907 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9908 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9911 2017-10-06 Richard Biener <rguenther@suse.de>
9913 PR tree-optimization/82397
9914 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9915 operand_equal_p but rely on data_ref_compare_tree for detecting
9917 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9918 to match up with dr_group_sort_cmp.
9920 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9923 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9925 * config/s390/s390-builtin-types.def: Regenerate.
9927 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9930 * config/s390/s390-builtin-types.def: Regenerate.
9931 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9932 Change flag from B_VXE to B_VX.
9933 (s390_vec_min_dbl): Remove B_VXE flag.
9935 2017-10-06 Richard Biener <rguenther@suse.de>
9937 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9938 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9939 translate_isl_ast_to_gimple::is_valid_rename,
9940 translate_isl_ast_to_gimple::get_rename,
9941 translate_isl_ast_to_gimple::get_def_bb_for_const,
9942 translate_isl_ast_to_gimple::get_new_name,
9943 translate_isl_ast_to_gimple::collect_all_ssa_names,
9944 translate_isl_ast_to_gimple::copy_loop_phi_args,
9945 translate_isl_ast_to_gimple::collect_all_ssa_names,
9946 translate_isl_ast_to_gimple::copy_loop_phi_args,
9947 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9948 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9949 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9950 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9951 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9952 translate_isl_ast_to_gimple::copy_cond_phi_args,
9953 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9954 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9955 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9956 translate_isl_ast_to_gimple::rename_uses,
9957 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9958 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9959 (set_rename_for_each_def): Likewise.
9960 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9961 here. Handle rewriting SCEV analyzable uses here.
9962 (copy_bb_and_scalar_dependences): Generate code for PHI
9964 (graphite_regenerate_ast_isl): Adjust.
9965 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9966 (add_write, add_read): New functions.
9967 (build_cross_bb_scalars_def): Use it and simplify.
9968 (build_cross_bb_scalars_use): Likewise.
9969 (graphite_find_cross_bb_scalar_vars): Inline into...
9970 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9971 simulating out-of-SSA. Compute liveout and add dependencies.
9972 (build_scops): Force an empty entry block.
9973 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9975 (sese_build_liveouts): Declare.
9976 (sese_trivially_empty_bb_p): Likewise.
9977 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9978 compute liveout and debug_liveout.
9979 (sese_bad_liveouts_use): Remove.
9980 (sese_reset_debug_liveouts_bb): Likewise.
9981 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9982 (sese_build_liveouts): Build liveout and debug_liveout and store
9984 (new_sese_info): Adjust.
9985 (free_sese_info): Likewise.
9986 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9987 do not build liveout here.
9988 (move_sese_in_condition): Adjust region entry.
9989 (scev_analyzable_p): Match up with chrec_apply requirements.
9990 (sese_trivially_empty_bb_p): New.
9991 * tree-into-ssa.c (get_reaching_def): Properly support generating
9992 default-defs for incremental rewrite of anonymous names.
9994 2017-10-06 Richard Biener <rguenther@suse.de>
9996 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9997 precision do not perform modulo reduction.
9999 2017-10-06 Richard Biener <rguenther@suse.de>
10001 PR tree-optimization/82436
10002 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10003 conservatively choose the vectorization factor when checking
10004 whether we can perform the required load permutation.
10005 (vect_transform_slp_perm_load): Assert when we may not fail.
10007 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10009 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10010 message for incompatible -msdata=* and -mcall-* options.
10012 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10014 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10015 rate for post-reload scheduling.
10017 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10019 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10021 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10023 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10024 to improve monte carlo in scimark.
10026 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10028 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10029 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10030 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10031 pentium4_cost, nocona_cost): Set reassociation width to 1.
10032 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10033 width to 2 for fp operations and 1 otherwise.
10034 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10036 (atom_cost): Set reassociation width to 2.
10037 (slm_cost, generic_cost): Set fp reassociation width
10038 to 2 and 1 otherwise.
10039 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10040 (core_cost): Set fp reassociation width to 4 and vector to 2.
10041 (ix86_reassociation_width): Rewrite using cost table; special case
10042 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10043 and TARGET_AVX128_OPTIMAL.
10044 * config/i386/i386.h (processor_costs): Add
10045 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10046 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10047 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10048 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10049 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10050 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10052 2017-10-05 Nathan Sidwell <nathan@acm.org>
10054 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10056 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10058 * config/arm/arm.c (arm_test_fpu_data): New.
10059 (arm_run_selftests): Call arm_test_fpu_data.
10061 2017-10-04 Nathan Sidwell <nathan@acm.org>
10063 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10065 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10067 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10069 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10070 check type for aarch64_simd_valid_immediate.
10071 (aarch64_output_simd_mov_immediate): Update prototype.
10072 (aarch64_simd_valid_immediate): Update prototype.
10073 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10074 support for ORR-immediate.
10075 (and<mode>3): modified pattern to add support for BIC-immediate.
10076 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10077 now checks for valid immediate for BIC and ORR based on new enum
10079 (aarch64_output_simd_mov_immediate): Function now used to output
10080 BIC/ORR imm as well based on new enum argument.
10081 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10083 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10084 (aarch64_reg_or_bic_imm): Likewise.
10086 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10088 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10089 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10091 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10095 PR rtl-optimization/82396
10096 * haifa-sched.c (autopref_multipass_init): Simplify
10098 (autopref_rank_data): Simplify sort order.
10099 * sched-int.h (autopref_multipass_data_): Remove
10100 multi_mem_insn_p, min_offset and max_offset.
10102 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10104 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10106 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10108 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10109 vect_doubleint_cvt.
10111 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10113 * doc/sourcebuild.texi: Document vect_long_mult.
10115 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10117 PR tree-optimization/82413
10118 * fold-const.c (build_range_check): Use widest_int when comparing
10119 the maximum ETYPE value with HIGH.
10121 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10123 PR rtl-optimization/82396
10124 * haifa-sched.c (autopref_multipass_init): Simplify
10126 (autopref_rank_data): Simplify sort order.
10127 * sched-int.h (autopref_multipass_data_): Remove
10128 multi_mem_insn_p, min_offset and max_offset.
10130 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10132 PR tree-optimization/82381
10133 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10134 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10137 PR tree-optimization/82374
10138 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10139 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10140 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10141 current_function_decl to the new decl.
10143 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10145 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10146 helper macro for IEEE float128 hardware built-in functions.
10147 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10149 (TRUNCF128_ODD): Likewise.
10150 (ADDF128_ODD): Likewise.
10151 (SUBF128_ODD): Likewise.
10152 (MULF128_ODD): Likewise.
10153 (DIVF128_ODD): Likewise.
10154 (FMAF128_ODD): Likewise.
10155 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10156 UNSPEC_TRUNC_ROUND_TO_ODD.
10157 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10158 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10159 floating point round to odd instructions.
10160 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10161 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10162 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10163 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10164 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10165 (trunc<mode>sf2_hw): Change the truncate with round to odd
10166 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10167 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10168 to odd hardware instructions.
10169 (sub<mode>3_odd): Likewise.
10170 (mul<mode>3_odd): Likewise.
10171 (div<mode>3_odd): Likewise.
10172 (sqrt<mode>2_odd): Likewise.
10173 (fma<mode>4_odd): Likewise.
10174 (fms<mode>4_odd): Likewise.
10175 (nfma<mode>4_odd): Likewise.
10176 (nfms<mode>4_odd): Likewise.
10177 (trunc<mode>df2_odd): Change the truncate with round to odd
10178 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10180 * doc/extend.texi (PowerPC built-in functions): Update documentation
10181 for existing IEEE float128-bit built-in functions. Add built-in
10182 functions that generate the IEEE 128-bit floating point round to
10185 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10187 PR rtl-optimization/77729
10188 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10189 to (X&(C1&~C2))|C2 transformations.
10191 2017-10-03 Martin Jambor <mjambor@suse.cz>
10193 PR tree-optimization/82363
10194 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10195 mismatch, mark lacc written regardless of racc.
10197 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10199 PR tree-optimization/82381
10200 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10201 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10202 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10203 fallthrough into reassoc_stmt_dominates_stmt_p.
10206 * combine.c (combine_instructions): Don't combine in unreachable
10209 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10213 function to not use the have_cpu variable. Do not set cpu_index,
10214 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10215 or the default cpu.
10216 (rs6000_valid_attribute_p): Remove duplicate initializations of
10217 old_optimize and func_optimize.
10218 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10219 (rs6000_activate_target_options): Make global.
10220 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10223 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10225 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10226 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10227 Set *poffset to 0 and *psize and *pmax_size to -1 if
10228 *poffset + *psize overflows in HOST_WIDE_INT.
10230 PR tree-optimization/82387
10231 PR tree-optimization/82388
10232 PR tree-optimization/82389
10233 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10234 instead of live_bytes non-NULL.
10236 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10239 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10242 2017-10-02 Richard Biener <rguenther@suse.de>
10244 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10245 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10246 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10248 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10250 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10251 requested precision matches the type's.
10252 * calls.c (alloc_max_size): Calculate the new candidate size as
10253 a widest_int and use wi::to_widest when comparing it with the
10254 current candidate size.
10255 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10256 zero rather than integer_zero_node.
10257 * match.pd: Check for a no-op conversion before using wi::add
10258 rather than after. Use tree_to_uhwi when summing small shift
10259 counts into an unsigned int.
10261 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10262 Alan Hayward <alan.hayward@arm.com>
10263 David Sherwood <david.sherwood@arm.com>
10266 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10267 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10268 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10269 POINTER_AND_FP_REGS.
10271 2017-10-02 Richard Biener <rguenther@suse.de>
10273 PR tree-optimization/82355
10274 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10275 a mapping for the enclosing loop but avoid generating one for
10276 the loop tree root.
10277 (copy_bb_and_scalar_dependences): Remove premature codegen
10278 error on PHIs in blocks duplicated into multiple places.
10279 * graphite-scop-detection.c
10280 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10281 in the region use it as loop and nest to analyze the DR in.
10282 (try_generate_gimple_bb): Likewise.
10283 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10284 (add_loop_constraints): For blocks in a loop not in the region
10285 create a dimension with a single iteration.
10286 * sese.h (gbb_loop_at_index): Remove assert.
10288 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10290 * omp-expand.c (adjust_context_scope): New function.
10291 (expand_parallel_call): Call adjust_context_scope.
10293 2017-10-01 Jeff Law <law@redhat.com>
10295 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10296 dom_opt_dom_walker class with direct access to private members.
10297 Add comments. Call test_for_singularity.
10298 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10299 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10300 m_dummy_cond anymore.
10301 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10303 (pass_dominator:execute): Build the dummy_cond here and pass it
10304 to the dom_opt_dom_walker ctor.
10305 (test_for_singularity): New function.
10307 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10308 Maya Rashish <coypu@sdf.org>
10310 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10311 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10312 (alpha*-*-netbsd*) Use nbsd_tm_file.
10313 (arm*-*-netbsdelf*) Likewise.
10314 (i[34567]86-*-netbsdelf*) Likewise.
10315 (x86_64-*-netbsd*) Likewise.
10316 (mips*-*-netbsd*) Likewise.
10317 (powerpc-*-netbsd*) Likewise.
10318 (sh*-*-netbsd*) Likewise.
10319 (sparc-*-netbsdelf*) Likewise.
10320 (sparc64-*-netbsd*) Likewise.
10321 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10323 (vax-*-netbsdelf*) Likewise.
10324 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10325 (UINT_FAST8_TYPE) Likewise.
10326 (INT_FAST16_TYPE) Check CHAR_FAST16.
10327 (UINT_FAST16_TYPE) Likewise.
10329 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10332 * config/i386/i386.md
10333 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10334 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10335 *divmodsi4_zext_2): New define_insn_and_split.
10336 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10337 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10338 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10339 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10340 New define_insn_and_split.
10341 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10342 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10343 operands[1] having DImode when mode is SImode.
10345 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10346 always SImode for DIV and MOD in REG_EQUAL notes.
10348 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10350 PR middle-end/82319
10351 * match.pd: Fix handling of NaNs in pattern.
10353 2017-09-29 Jeff Law <law@redhat.com>
10355 * sbitmap.c (bitmap_bit_in_range_p): New function.
10356 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10357 * tree-ssa-dse.c (live_bytes_read): New function.
10358 (dse_classify_store): Ignore reads of dead bytes.
10360 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10361 typos and whitespace errors.
10362 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10363 * config/s390/s390.c (s390_emit_prologue): Likewise.
10365 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10368 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10369 with a symbol for LRA.
10371 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10373 PR rtl-optimization/82338
10374 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10376 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10378 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10379 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10380 (qsort_chk): Declare.
10381 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10382 (qsort_chk): New function.
10384 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10386 PR tree-optimization/82337
10387 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10388 phi definition if the PHI result appears in an abnormal PHI.
10389 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10390 basis appears in an abnormal PHI.
10392 2017-09-29 Richard Biener <rguenther@suse.de>
10394 * graphite-isl-ast-to-gimple.c
10395 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10396 (binary_op_to_tree): Use it.
10397 (get_rename_from_scev): Likewise.
10398 (copy_loop_phi_nodes): Likewise.
10399 (copy_bb_and_scalar_dependences): Likewise.
10400 (translate_pending_phi_nodes): Likewise.
10402 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10405 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10406 for movabsq $(i32 << shift), r64.
10408 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10410 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10411 index when encoding %esp as %rsp to avoid 0x67 prefix.
10413 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10415 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10416 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10418 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10420 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10421 Extensions with more than 16 double VFP registers.
10422 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10423 to_clear_mask and all code related to it. Replace the remaining
10424 entry by a sbitmap and adapt code accordingly.
10426 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10428 * brig-builtins.def: Change pure attributes to const.
10430 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10432 * config.gcc (default_gnu_indirect_function): Default to yes for
10433 sparc*-*-linux* with glibc.
10435 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10437 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10438 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10439 when creating .init_array and .fini_array sections with priority
10442 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10445 * config/aarch64/aarch64.c
10446 (aarch64_builtin_support_vector_misalignment): Always return false
10447 when misalignment is unknown.
10449 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10451 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10452 this function to return false if the definition used by the swap
10453 instruction is artificial, or if the memory address from which the
10454 constant value is loaded is not represented by a base address held
10455 in a register or if the base address register is a frame or stack
10456 pointer. Additionally, return false if the base address of the
10457 loaded constant is a SYMBOL_REF but is not considered to be a
10459 (replace_swapped_load_constant): New function.
10460 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10461 loaded constant vector with a load of a swapped constant vector.
10463 2017-09-27 Carl Love <cel@us.ibm.com>
10465 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10466 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10467 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10470 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10472 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10473 first, always call autopref_rank_data otherwise.
10475 2017-09-27 Richard Biener <rguenther@suse.de>
10477 * graphite-scop-detection.c (find_scop_parameters): Move
10478 loop bound handling ...
10479 (gather_bbs::before_dom_children): ... here, avoiding the need
10480 to build scop_info->loop_nest.
10481 (record_loop_in_sese): Remove.
10482 * sese.h (sese_info_t::loop_nest): Remove.
10483 * sese.c (new_sese_info): Do not allocate loop_nest.
10484 (free_sese_info): Do not free loop_nest.
10486 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10489 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10490 lhs from calls if the lhs has addressable type.
10492 2017-09-27 Richard Biener <rguenther@suse.de>
10494 * graphite.h (scop::max_alias_set): New member.
10495 * graphite-scop-detection.c: Remove references to non-existing
10496 --param in comments.
10497 (build_alias_sets): Record the maximum alias set used for drs.
10498 (build_scops): Support zero as unlimited for
10499 --param graphite-max-arrays-per-scop.
10500 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10501 and inline into ...
10502 (build_poly_sr_1): ... here. Compute alias set based on the
10503 maximum alias set used for drs rather than
10504 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10506 2017-09-27 Richard Biener <rguenther@suse.de>
10508 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10509 --param loop-block-tile-size=0 to disable tiling.
10511 2017-09-27 Richard Biener <rguenther@suse.de>
10513 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10514 (graphite-max-nb-scop-params): Document special value zero.
10515 * domwalk.h (dom_walker::STOP): New symbolical constant.
10516 (dom_walker::dom_walker): Add optional parameter for bb to
10518 (dom_walker::~dom_walker): Declare.
10519 (dom_walker::before_dom_children): Document STOP return value.
10520 (dom_walker::m_user_bb_to_rpo): New member.
10521 (dom_walker::m_bb_to_rpo): Likewise.
10522 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10523 mapping here if not provided by the user.
10524 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10525 provided by the user.
10526 (dom_walker::STOP): Define.
10527 (dom_walker::walk): Do not compute bb to RPO mapping here.
10528 Support STOP return value from before_dom_children to stop
10530 * graphite-optimize-isl.c (optimize_isl): If the schedule
10531 is the same still generate code if -fgraphite-identity
10532 or -floop-parallelize-all are given.
10533 * graphite-scop-detection.c: Include cfganal.h.
10534 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10536 (gather_bbs::before_dom_children): Return STOP for BBs
10538 (build_scops): Compute bb to RPO mapping and pass it to
10539 the domwalk. Treat --param graphite-max-nb-scop-params=0
10540 as not limiting the number of params.
10541 * graphite.c (graphite_initialize): Remove limit on the
10542 number of basic-blocks in a function.
10543 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10544 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10545 default value of 10.
10547 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10549 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10550 Adjust code to eliminate needing to do the shift right 32-bits
10551 operation after XSCVDPSPN.
10553 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10555 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10556 ((X / Y) != 0 -> X >= Y): Likewise.
10558 2017-09-26 Carl Love <cel@us.ibm.com>
10560 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10561 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10562 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10563 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10564 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10565 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10566 definitions and overloading.
10567 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10568 statement for P9V_BUILTIN_XST_LEN_R.
10569 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10570 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10571 define_expand and define_insn for the instructions and builtins.
10572 * doc/extend.texi: Update the built-in documentation file for the new
10573 built-in functions.
10574 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10575 define_insn for the instructions
10577 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10580 * gcc/config/netbsd-protos.h: New file.
10581 * gcc/config/netbsd.c: New file.
10582 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10583 * gcc/config/t-netbsd: New file.
10584 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10585 (tmake_file) Add t-netbsd.
10586 (extra_objs) Add netbsd.o.
10588 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10592 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10594 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10596 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10597 sign extension from a vector register to a GPR by doing a 32-bit
10598 direct move and then an EXTSW.
10599 (extendsi<mode>2 splitter): Likewise.
10600 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10601 right or vector extract after doing XSCVDPSPN. Use
10602 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10604 (movdi_from_sf_zero_ext): Likewise.
10605 (reload_gpr_from_vsxsf): Likewise.
10606 (p8_mfvsrd_4_disf): Delete, no longer used.
10607 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10608 then needing to move the SFmode to a GPR to use the XSCVDPSP
10609 instruction instead of FRSP and XSCVDPSPN.
10610 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10611 it is adjacent to the other XSCVSPDP insns.
10612 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10613 SFmode to be in traditional Altivec registers.
10614 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10615 (vsx_xscvspdpn): Likewise.
10616 (vsx_xscvspdpn_scalar): Likewise.
10618 2017-09-26 Martin Jambor <mjambor@suse.cz>
10620 * tree-sra.c (compare_access_positions): Put integral types first,
10621 stabilize sorting of integral types, remove conditions putting
10622 non-full-precision integers last.
10623 (sort_and_splice_var_accesses): Disable scalarization if a
10624 non-integert would be represented by a non-full-precision integer.
10626 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10628 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10629 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10630 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10631 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10632 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10633 conditionals inside the function instead of around it. Call
10634 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10635 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10637 2017-09-26 Richard Biener <rguenther@suse.de>
10639 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10641 (scop_detection::build_scop_breadth): ... this. Removed.
10642 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10643 (scop_detection::harmful_stmt_in_bb): Likewise.
10644 (scop_detection::graphite_can_represent_stmt): Likewise.
10645 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10646 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10647 (scop_detection::can_represent_loop_1): ... this. Removed.
10648 (scop_detection::harmful_loop_in_region): Simplify after inlining
10649 the above and remove more quadraticness.
10650 (build_scops): Adjust.
10651 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10654 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10657 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10658 REGNO (base) == SP_REG if base is a REG.
10660 PR middle-end/35691
10661 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10662 if it is different SSA_NAME.
10663 (optimize_range_tests_cmp_bitwise): New function.
10664 (optimize_range_tests): Call it.
10666 2017-09-26 Richard Biener <rguenther@suse.de>
10668 PR tree-optimization/82321
10669 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10670 for the def being inside the loop.
10672 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10674 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10676 * config/s390/s390-builtins.def: Fix constraint on op4.
10678 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10680 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10681 independent expanders.
10682 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10683 ("vec_ordered", "vec_unordered"): New expanders.
10685 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10687 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10690 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10692 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10693 vec_unpacks_lo_v16qi.
10694 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10696 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10698 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10699 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10700 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10702 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10704 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10706 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10708 ("*vec_slb<mode>"): New insn pattern.
10709 ("vec_shr_<mode>"): New expander.
10710 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10711 and force the shift count operand to V16QImode.
10712 ("vec_srb<mode>"): Set shift count mode to V16QI.
10714 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10716 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10717 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10718 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10720 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10723 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10725 2017-09-26 Richard Biener <rguenther@suse.de>
10727 PR tree-optimization/82320
10728 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10731 2017-09-25 Jeff Law <law@redhat.com>
10733 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10734 prototype for new argument.
10735 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10736 mostly extracted from rs6000_emit_allocate_stack.
10737 (rs6000_emit_probe_stack_range_stack_clash): New function.
10738 (rs6000_emit_allocate_stack): Call
10739 rs6000_emit_probe_stack_range_stack_clash as needed.
10740 (rs6000_emit_probe_stack_range): Add additional argument
10741 to call to gen_probe_stack_range{si,di}.
10742 (output_probe_stack_range): New.
10743 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10744 (output_probe_stack_range_stack_clash): New.
10745 (rs6000_emit_prologue): Emit notes into dump file as requested.
10746 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10747 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10748 Add additional operand and pass it to output_probe_stack_range.
10750 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10752 PR tree-optimization/82163
10753 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10754 (checking_verify_loop_closed_ssa): New parameter.
10755 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10756 (check_loop_closed_ssa_stmt): Delete.
10757 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10758 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10759 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10762 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10764 * brig-builtins.def: Treat HSAIL barrier builtins as
10765 setjmp/longjump style functions.
10767 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10769 * target.def (constant_alignment): New hook.
10770 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10771 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10772 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10773 * doc/tm.texi: Regenerate.
10774 * targhooks.h (default_constant_alignment): Declare.
10775 (constant_alignment_word_strings): Likewise.
10776 * targhooks.c (default_constant_alignment): New function.
10777 (constant_alignment_word_strings): Likewise.
10778 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10779 instead of CONSTANT_ALIGNMENT.
10780 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10781 (force_const_mem): Likewise.
10782 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10783 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10784 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10785 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10786 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10788 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10789 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10790 constant_alignment_word_strings.
10791 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10792 (CONSTANT_ALIGNMENT): Likewise.
10793 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10794 (arm_constant_alignment): New function.
10795 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10796 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10797 constant_alignment_word_strings.
10798 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10799 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10800 constant_alignment_word_strings.
10801 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10802 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10803 (cris_constant_alignment): New function.
10804 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10805 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10806 (epiphany_constant_alignment): New function.
10807 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10808 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10809 constant_alignment_word_strings.
10810 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10811 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10812 constant_alignment_word_strings.
10813 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10814 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10815 constant_alignment_word_strings.
10816 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10817 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10818 * config/i386/i386.c (ix86_constant_alignment): Make static.
10819 Use the same interface as the target hook.
10820 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10821 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10822 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10823 constant_alignment_word_strings.
10824 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10825 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10826 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10827 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10828 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10829 constant_alignment_word_strings.
10830 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10831 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10832 constant_alignment_word_strings.
10833 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10834 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10835 constant_alignment_word_strings.
10836 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10837 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10839 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10840 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10841 * config/mips/mips.c (mips_constant_alignment): New function.
10842 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10843 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10844 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10845 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10846 (mmix_constant_alignment): Make static. Use the same interface
10847 as the target hook.
10848 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10849 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10850 constant_alignment_word_strings.
10851 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10852 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10853 constant_alignment_word_strings.
10854 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10855 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10856 constant_alignment_word_strings.
10857 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10858 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10859 (rs6000_constant_alignment): New function.
10860 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10861 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10862 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10863 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10864 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10865 (rs6000_constant_alignment): New function.
10866 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10867 * config/s390/s390.c (s390_constant_alignment): New function.
10868 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10869 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10870 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10871 constant_alignment_word_strings.
10872 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10873 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10874 (sparc_constant_alignment): New function.
10875 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10876 * config/spu/spu.c (spu_constant_alignment): New function.
10877 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10878 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10879 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10880 constant_alignment_word_strings.
10881 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10882 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10883 constant_alignment_word_strings.
10884 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10885 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10886 constant_alignment_word_strings.
10887 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10888 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10889 (visium_constant_alignment): New function.
10890 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10891 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10892 (xtensa_constant_alignment): New function.
10893 * system.h (CONSTANT_ALIGNMENT): Poison.
10895 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10897 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10898 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10899 (rs6000_builtin_valid_without_lhs): New helper function.
10900 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10901 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10903 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10905 * target.h (vec_perm_indices): Use unsigned short rather than
10907 (auto_vec_perm_indices): Likewise.
10908 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10909 Use unsigned int rather than unsigned char.
10910 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10912 2017-09-25 Richard Biener <rguenther@suse.de>
10914 * cfgloop.h (sort_sibling_loops): Declare.
10915 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10916 (sort_sibling_loops): New function sorting the sibling loop list
10918 * graphite.c (graphite_transform_loops): Sort sibling loops.
10920 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10922 * target.def (vec_perm_const_ok): Change sel parameter to
10924 * optabs-query.c (can_vec_perm_p): Update accordingly.
10925 * doc/tm.texi: Regenerate.
10926 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10927 auto_vec_perm_indices and remove separate nelt field.
10928 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10929 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10930 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10931 (aarch64_expand_vec_perm_const): Update accordingly.
10932 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10933 to vec_perm_indices.
10934 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10935 auto_vec_perm_indices and remove separate nelt field.
10936 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10937 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10938 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10940 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10941 to vec_perm_indices.
10942 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10943 sel to vec_perm_indices.
10944 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10945 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10946 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10948 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10951 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10954 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10955 on the FUNCTION_DECL function context if it has a DIE that is a
10958 2017-09-25 Richard Biener <rguenther@suse.de>
10960 PR tree-optimization/82285
10961 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10964 2017-09-25 Tom de Vries <tom@codesourcery.com>
10968 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10971 2017-09-25 Richard Biener <rguenther@suse.de>
10973 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10974 ISL errors other than isl_error_quota happen. Dump if the
10975 schedule is the same.
10976 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10977 errors instead of aborting inside ISL.
10979 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10982 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10983 of libgcc_eh for m64.
10984 * config/i386/darwin64.h: Likewise.
10986 2017-09-25 Richard Biener <rguenther@suse.de>
10988 PR middle-end/82144
10989 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10990 attribute for incomplete types nor twice for complete ones.
10992 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10995 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10996 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10999 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11002 * config/i386/i386.opt (mprefer-avx256): Use
11003 ix86_target_flags variable.
11004 * config/i386/i386.c (ix86_target_string): Move
11005 -mprefer-avx256 to flag2_opts.
11007 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11009 PR middle-end/35691
11010 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11011 and x != -1 | y != -1 into (x & y) != -1.
11013 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11015 * config.gcc: Add new case statement to set
11016 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11017 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11018 s390x-*-linux* case statements. Added aarch64 to the list of
11019 supported architectures.
11021 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11023 PR tree-optimization/82289
11024 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11025 STMT_VINFO_RELEVANT_P.
11027 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11028 Alan Hayward <alan.hayward@arm.com>
11029 David Sherwood <david.sherwood@arm.com>
11031 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11032 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11033 (extract_range_from_binary_expr_1): Don't call
11034 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11036 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11037 Alan Hayward <alan.hayward@arm.com>
11038 David Sherwood <david.sherwood@arm.com>
11040 * target.def (preferred_vector_alignment): New hook.
11041 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11043 * doc/tm.texi: Regenerate.
11044 * targhooks.h (default_preferred_vector_alignment): Declare.
11045 * targhooks.c (default_preferred_vector_alignment): New function.
11046 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11048 (DR_TARGET_ALIGNMENT): New macro.
11049 (aligned_access_p): Update commentary.
11050 (vect_known_alignment_in_bytes): New function.
11051 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11053 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11054 Calculate the misalignment based on the target alignment rather than
11056 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11057 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11058 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11059 the target alignment, rather than masking the element misalignment
11060 with the number of elements in a vector. Also use the target
11061 alignment when calculating the maximum number of peels.
11062 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11063 instead of TYPE_ALIGN_UNIT.
11064 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11065 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11066 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11067 (vect_create_data_ref_ptr): Likewise.
11068 (vect_setup_realignment): Realign by ANDing with
11069 -DR_TARGET_MISALIGNMENT.
11070 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11071 the number of peels based on DR_TARGET_ALIGNMENT.
11072 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11073 with the guaranteed alignment boundary when deciding whether
11075 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11076 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11077 (ensure_base_align): Remove stmt_info parameter. Get the
11078 target base alignment from DR_TARGET_ALIGNMENT.
11079 (vectorizable_store): Update call accordingly. Interpret
11080 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11082 (vectorizable_load): Likewise.
11084 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11085 Alan Hayward <alan.hayward@arm.com>
11086 David Sherwood <david.sherwood@arm.com>
11088 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11089 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11090 (vect_enhance_data_refs_alignment): Likewise.
11092 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11094 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11095 error. Only quit immediately if parsing is complete.
11096 (BEGIN): Initialize fatal_err and parse_done.
11097 (begin fpu, end fpu): Check number of arguments.
11098 (begin arch, end arch): Likewise.
11099 (begin cpu, end cpu): Likewise.
11100 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11101 (optalias): Likewise.
11103 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11105 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11106 * config/arm/arm-isa.h: Delete. Move definitions to ...
11107 * arm-cpus.in: ... here. Use new feature and fgroup values.
11108 * config/arm/arm.c (arm_option_override): Use lower case for feature
11110 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11111 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11112 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11114 (print_isa_bits_for): New function.
11115 (gen_isa): New function.
11116 (gen_comm_data): Use print_isa_bits_for.
11117 (define feature): New keyword.
11118 (define fgroup): New keyword.
11119 * config/arm/t-arm (TM_H): Remove.
11120 (GTM_H): Add arm-isa.h.
11121 (arm-isa.h): Add rule to generate file.
11122 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11123 case for feature bit names.
11125 2017-09-22 Richard Biener <rguenther@suse.de>
11127 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11129 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11130 print of no dependency loops ...
11131 * graphite.c (graphite_transform_loops): ... here.
11132 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11134 (same_close_phi_node, remove_duplicate_close_phi,
11135 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11136 (canonicalize_loop_closed_ssa): ... here and simplify.
11137 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11138 (optimize_isl): Use dump_printf_loc to tell when we stopped
11139 optimizing because of an ISL timeout.
11141 2017-09-22 Richard Biener <rguenther@suse.de>
11143 PR tree-optimization/82291
11144 * tree-if-conv.c (predicate_mem_writes): Make sure to
11145 remove writes in blocks predicated with false.
11147 2017-09-22 Richard Biener <rguenther@suse.de>
11149 * sese.c: Include cfganal.h.
11150 (if_region_set_false_region): Remove.
11151 (create_if_region_on_edge): Likewise.
11152 (move_sese_in_condition): Re-implement without destroying
11155 2017-09-22 Richard Biener <rguenther@suse.de>
11157 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11158 Verify both BBs contain loop PHI nodes before dispatching to
11159 copy_loop_phi_args.
11160 (graphite_regenerate_ast_isl): Do not recompute dominators,
11161 do not verify three times. Restructure for clarity.
11162 * graphite-scop-detection.c (same_close_phi_node,
11163 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11164 defined_in_loop_p, canonicalize_loop_closed_ssa,
11165 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11166 checking and SSA rewrite, move to ...
11167 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11168 (graphite_initialize): Do not pass in ctx, do not reset the
11169 SCEV cache, compute only dominators.
11170 (graphite_transform_loops): Allocate ISL ctx after
11171 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11172 Maintain post-dominators only around build_scops.
11173 * sese.c (if_region_set_false_region): Make static. Free
11174 and recompute dominators.
11175 (move_sese_in_condition): Assert we don't get called with
11176 post-dominators computed.
11177 * sese.h (if_region_set_false_region): Remove.
11179 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11181 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11182 mode attribute for TARGET_AVX512VL.
11184 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11186 * config/i386/i386.opt (mprefer-avx256): New option.
11187 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11189 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11190 for TARGET_PREFER_AVX256.
11191 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11193 2017-09-21 Jeff Law <law@redhat.com>
11195 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11196 Fix dump output if the only stack space is for pushed registers.
11198 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11200 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11203 2017-09-21 Martin Sebor <msebor@redhat.com>
11206 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11207 code (in C++) or code that triggers warnings.
11209 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11211 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11213 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11215 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11216 * sched-rgn.c: Ditto.
11217 * sel-sched-ir.c: Ditto.
11219 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11221 * toplev.h (set_random_seed): Adjust return type.
11222 * toplev.c (init_local_tick): Move eager initialization of random_seed
11223 to get_random_seed. Adjust comment.
11224 (init_random_seed): Inline to get_random_seed, delete.
11225 (get_random_seed): Initialize random_seed lazily.
11226 (set_random_seed): Do not return previous value.
11227 (print_switch_value): Do not call get_random_seed.
11229 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11231 * cgraph.c (delete_function_version): New, broken out from...
11232 (cgraph_node::delete_function_version): ...here. Rename to
11233 cgraph_node::delete_function_version_by_decl. Update all uses.
11234 (cgraph_node::remove): Call delete_function_version.
11236 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11239 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11240 VAR_DECLs to which addressable non-volatile parameters are mapped
11241 and for id->retvar after the return value assignment. Clear
11242 id->retval and id->retbnd after inlining.
11244 2017-09-21 Richard Biener <rguenther@suse.de>
11246 PR tree-optimization/82276
11247 PR tree-optimization/82244
11248 * tree-vrp.c (build_assert_expr_for): Set
11249 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11251 (remove_range_assertions): Revert earlier change.
11253 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11256 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11258 2017-09-21 Richard Biener <rguenther@suse.de>
11260 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11261 Restore valid IL after code generation errors.
11262 * graphite.c (graphite_transform_loops): Diagnose code
11263 generation issues as MSG_MISSED_OPTIMIZATION and continue
11264 with processing SCOPs.
11266 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11267 Alan Hayward <alan.hayward@arm.com>
11268 David Sherwood <david.sherwood@arm.com>
11270 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11271 rather than choosing between plus_constant and gen_rtx_<CODE>.
11272 * expr.c (emit_push_insn): Likewise.
11273 (expand_expr_real_2): Likewise.
11275 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11276 Alan Hayward <alan.hayward@arm.com>
11277 David Sherwood <david.sherwood@arm.com>
11279 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11281 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11282 Alan Hayward <alan.hayward@arm.com>
11283 David Sherwood <david.sherwood@arm.com>
11285 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11286 calling tree_to_uhwi.
11288 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11289 Alan Hayward <alan.hayward@arm.com>
11290 David Sherwood <david.sherwood@arm.com>
11292 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11293 INTEGER_CST rather than a negative test for ADDR_EXPR.
11295 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11296 Alan Hayward <alan.hayward@arm.com>
11297 David Sherwood <david.sherwood@arm.com>
11299 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11300 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11302 2017-09-21 Richard Biener <rguenther@suse.de>
11304 PR tree-optimization/71351
11305 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11306 graphite_create_new_loop_guard): Remove, fold remaining parts
11308 (translate_isl_ast_node_for): ... here and simplify.
11310 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11313 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11314 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11315 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11316 alternative always use QI mode, for -Os imov (=R,R) alternative
11317 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11320 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11321 Jeff Law <law@redhat.com>
11323 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11324 (allocate_stack_space): New function, partially extracted from
11325 s390_emit_prologue.
11326 (s390_emit_prologue): Track offset to most recent stack probe.
11327 Code to allocate space moved into allocate_stack_space.
11328 Dump actions when no stack is allocated.
11329 (s390_prologue_plus_offset): New function.
11330 (s390_emit_stack_probe): Likewise.
11332 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11334 * common.opt (Wa, Wl, Wp, g, gz=): Add
11336 (gno-column-info): Remove.
11337 (gcolumn-info): Drop RejectNegative.
11338 (gno-): New prefix.
11339 (gno-record-gcc-switches): Remove.
11340 (grecord-gcc-switches): Drop RejectNegative.
11341 (gno-split-dwarf): Remove.
11342 (gsplit-dwarf): Drop RejectNegative.
11343 (gno-strict-dwarf): Remove.
11344 (gstrict-dwarf): Drop RejectNegative.
11345 * config/darwin.opt (gfull, gused): Add RejectNegative.
11346 * dwarf2out.c (gen_producer_string): Drop
11347 gno-record-gcc-switches handler.
11348 * optc-gen.awk: Add g to prefixes with negative forms.
11349 * opts-common.c (remapping_prefix_p): New.
11350 (find_opt): Check it.
11351 (generate_canonical_option): Test g prefix.
11352 (option_map): Add -gno- mapping.
11353 (add_misspelling_candidates): Check remapping_prefix_p.
11355 2017-09-20 Jeff Law <law@redhat.com>
11357 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11358 thinko in stack clash protection support.
11360 * explow.c (compute_stack_clash_protection_loop_data): Use
11361 CONST_INT_P instead of explicit test. Verify object is a
11362 CONST_INT_P before looking at INTVAL.
11363 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11364 instead of explicit test.
11366 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11369 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11370 address instead of to r1 and r11.
11372 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11374 * config.gcc: Support "knm".
11375 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11376 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11378 * config/i386/i386.c (m_KNM): Define.
11379 (processor_target_table): Add "knm".
11381 (ix86_option_override_internal): Add "knm".
11382 (ix86_issue_rate): Add PROCESSOR_KNM.
11383 (ix86_adjust_cost): Ditto.
11384 (ia32_multipass_dfa_lookahead): Ditto.
11385 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11386 (fold_builtin_cpu): Add M_INTEL_KNM.
11387 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11388 (processor_type): Add PROCESSOR_KNM.
11389 * config/i386/x86-tune.def: Add m_KNM.
11390 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11392 2017-09-20 Richard Biener <rguenther@suse.de>
11394 PR tree-optimization/80213
11395 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11396 are allowed in empty BBs as well.
11397 (canonicalize_loop_closed_ssa): Also look for other complex
11399 (scop_detection::get_sese): Include the loop-closed PHI block
11401 (scop_detection::merge_sese): Remove code adding extra blocks.
11402 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11403 (build_scops): Assert the final returned scop is invalid.
11405 2017-09-20 Richard Biener <rguenther@suse.de>
11407 PR tree-optimization/82264
11408 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11410 (vn_phi_lookup): Likewise.
11411 (vn_phi_insert): Likewise.
11413 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11415 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11416 that fits into uhwi or shwi, add DW_AT_const_value regardless
11417 of early_dwarf without going through RTL, using add_AT_unsigned
11420 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11421 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11422 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11423 (DEBUG_MACRO_SECTION): Likewise.
11424 (DEBUG_LTO_MACRO_SECTION): Likewise.
11425 (DEBUG_STR_DWO_SECTION): Likewise.
11426 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11427 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11428 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11429 (DEBUG_LTO_LINE_STR_SECTION): Define.
11430 (init_sections_and_labels): Initialize debug_line_str_section
11431 variable. Initialize debug_loc_section for -gdwarf-5 to
11432 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11434 2017-09-20 Richard Biener <rguenther@suse.de>
11436 * graphite-sese-to-poly.c (extract_affine): Properly handle
11437 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11439 2017-09-20 Richard Biener <rguenther@suse.de>
11441 PR tree-optimization/81373
11442 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11443 Force SESE live-out defs to be handled even if they are
11446 2017-09-19 Jeff Law <law@redhat.com>
11448 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11449 nothing for stack adjustments with REG_STACK_CHECK.
11450 * sched-deps.c (parse_add_or_inc): Reject insns with
11451 REG_STACK_CHECK from dependency breaking.
11452 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11453 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11454 * reg-notes.def (STACK_CHECK): New note.
11456 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11457 (ix86_expand_prologue): Dump stack clash info as needed.
11458 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11460 * function.c (dump_stack_clash_frame_info): New function.
11461 * function.h (dump_stack_clash_frame_info): Prototype.
11462 (enum stack_clash_probes): New enum.
11464 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11465 flag_stack_clash_protection.
11466 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11467 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11468 (arm_frame_pointer_required): Likewise.
11469 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11470 (ia64_expand_prologue): Likewise.
11471 * config/mips/mips.c (mips_expand_prologue): Likewise.
11472 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11473 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11474 (sparc_flat_expand_prologue): Likewise.
11475 * config/spu/spu.c (spu_expand_prologue): Likewise.
11477 * explow.c: Include "params.h".
11478 (anti_adjust_stack_and_probe_stack_clash): New function.
11479 (get_stack_check_protect): Likewise.
11480 (compute_stack_clash_protection_loop_data): Likewise.
11481 (emit_stack_clash_protection_loop_start): Likewise.
11482 (emit_stack_clash_protection_loop_end): Likewise.
11483 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11484 Use anti_adjust_stack_and_probe_stack_clash.
11485 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11486 (emit_stack_clash_protection_loop_start): Likewise.
11487 (emit_stack_clash_protection_loop_end): Likewise.
11488 * rtl.h (get_stack_check_protect): Prototype.
11489 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11490 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11491 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11493 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11495 * doc/tm.texi: Rebuilt.
11496 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11497 get_stack_check_protect.
11498 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11499 * config/arm/arm.c (arm_expand_prologue): Likewise.
11500 (arm_frame_pointer_required): Likewise.
11501 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11502 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11503 * config/mips/mips.c (mips_expand_prologue): Likewise.
11504 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11505 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11506 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11507 (sparc_flat_expand_prologue): Likewise.
11509 * common.opt (-fstack-clash-protection): New option.
11510 * flag-types.h (enum stack_check_type): Note difference between
11511 -fstack-check= and -fstack-clash-protection.
11512 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11513 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11514 * toplev.c (process_options): Issue warnings/errors for cases
11515 not handled with -fstack-clash-protection.
11516 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11517 (-fstack-check): Note additional problem with -fstack-check=generic.
11518 Note that -fstack-check is primarily for Ada and refer users
11519 to -fstack-clash-protection for stack-clash-protection.
11520 Document new params for stack clash protection.
11522 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11524 * config/i386/i386.c (ix86_split_long_move): Do not handle
11525 address used for LEA in a special way.
11527 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11529 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11531 2017-09-19 Martin Sebor <msebor@redhat.com>
11534 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11535 of incompatible types.
11537 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11539 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11540 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11541 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11542 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11544 2017-09-19 Richard Biener <rguenther@suse.de>
11546 PR tree-optimization/82244
11547 * tree-vrp.c (remove_range_assertions): Do not propagate
11548 a constant to abnormals but replace the assert with a copy.
11550 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11552 PR rtl-optimization/57878
11553 PR rtl-optimization/68988
11554 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11555 avoidance test involving non_reload_pseudos. Move frequency test
11556 below the general fragmentation avoidance test.
11558 2017-09-19 Richard Biener <rguenther@suse.de>
11560 PR tree-optimization/69728
11561 * graphite-sese-to-poly.c (schedule_error): New global.
11562 (add_loop_schedule): Handle empty domain by failing the
11564 (build_original_schedule): Handle schedule_error.
11566 2017-09-19 Richard Biener <rguenther@suse.de>
11568 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11569 Do not iterate to sibling loops but only to siblings of inner
11572 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11575 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11576 that the registers are different.
11578 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11580 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11581 to processor_model and "amdfam17h" to arch_names_table.
11582 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11584 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11587 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11589 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11590 Alan Hayward <alan.hayward@arm.com>
11591 David Sherwood <david.sherwood@arm.com>
11593 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11595 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11597 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11598 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11599 vect_schedule_slp_instance.
11600 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11601 Update call to vect_slp_analyze_node_operations. Simplify return
11603 (vect_slp_analyze_bb_1): Update call accordingly.
11604 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11605 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11606 (vect_schedule_slp): Update call accordingly.
11608 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11609 Alan Hayward <alan.hayward@arm.com>
11610 David Sherwood <david.sherwood@arm.com>
11612 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11613 with types that aren't in fact scalar.
11615 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11617 * tree-vect-slp.c (vect_record_max_nunits): New function,
11619 (vect_build_slp_tree_1): ...here.
11620 (vect_build_slp_tree_2): Call it for phis too.
11622 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11624 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11625 to vect_get_vec_def_for_operand when getting the mask operand.
11627 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11628 Alan Hayward <alan.hayward@arm.com>
11629 David Sherwood <david.sherwood@arm.com>
11631 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11634 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11635 Alan Hayward <alan.hayward@arm.com>
11636 David Sherwood <david.sherwood@arm.com>
11638 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11639 calculation for vector booleans.
11641 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11642 Alan Hayward <alan.hayward@arm.com>
11643 David Sherwood <david.sherwood@arm.com>
11645 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11647 (vect_transform_stmt): ...here.
11648 (vect_analyze_stmt): Use it instead of calling
11649 vectorizable_live_operation directly.
11651 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11653 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11654 non-SIMT targets in acc vector loops.
11656 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11658 * configure.ac: Add arc and check if assembler supports gdwarf2.
11659 * configure: Regenerate.
11661 2017-09-18 Richard Biener <rguenther@suse.de>
11663 PR tree-optimization/82220
11664 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11665 epilogue niters from the min_profitable_iters compute.
11667 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11670 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11671 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11672 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11674 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11677 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11678 switching to a new text section.
11680 2017-09-18 Richard Biener <rguenther@suse.de>
11682 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11684 (build_alias_set): Reject aliases with no access function.
11686 2017-09-18 Richard Biener <rguenther@suse.de>
11688 PR tree-optimization/79622
11689 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11691 (build_cross_bb_scalars_use): Likewise.
11693 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11695 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11697 2017-09-18 Alan Modra <amodra@gmail.com>
11700 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11701 stack_pointer_rtx for count 0. Update comments. Break up
11702 large rtl expression.
11704 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11706 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11707 Increase to 20 bytes.
11708 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11709 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11710 or avx version of the stub.
11712 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11715 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11716 compute the minimum stack alignment. Also update preferred stack
11717 boundary for leaf functions.
11719 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11721 PR tree-optimization/82228
11722 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11725 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11727 * common/config/nds32/nds32-common.c
11728 (nds32_option_optimization_table): Refine formatting.
11729 (nds32_option_optimization_table): Use -fsched-pressure and
11730 -fomit-frame-pointer for specific optimization level.
11732 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11734 * config/nds32/nds32.c: Refine formatting and comments.
11735 * config/nds32/nds32.h: Likewise.
11736 * config/nds32/nds32.md: Likewise.
11737 * config/nds32/nds32-cost.c: Likewise.
11738 * config/nds32/nds32-isr.c: Likewise.
11739 * config/nds32/nds32-md-auxiliary.c: Likewise.
11740 * config/nds32/nds32-multiple.md: Likewise.
11741 * config/nds32/nds32-predicates.c: Likewise.
11743 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11744 Jakub Jelinek <jakub@redhat.com>
11746 Add support for -std=c++2a.
11747 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11749 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11751 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11754 * doc/extend.texi (Common Function Attributes): Add
11755 references to ARM, AArch64, and S/390 specific attributes.
11756 (Function Specific Option Pragmas): Add AArch64 and S/390
11757 to list of back ends that support the target pragma.
11759 2017-09-15 Nathan Sidwell <nathan@acm.org>
11761 * doc/standards.texi: Fix C++17 description. Update URLs for
11764 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11766 * common.opt (Wcast-align=strict): New warning option.
11767 * doc/invoke.texi: Document -Wcast-align=strict.
11769 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11771 * cgraph.h (cgraph_thunk_info): Add comments.
11772 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11773 assert for VIRTUAL_* arguments stricter.
11775 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11777 PR tree-optimization/71026
11778 * match.pd: Move RDIV patterns from fold-const.c
11779 * fold-const.c (distribute_real_division): Removed.
11780 (fold_binary_loc): Remove calls to distribute_real_divison.
11782 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11784 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11785 c++1z and gnu++1z as deprecated. Change other references to
11786 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11787 Change -Wc++1z-compat to -Wc++17-compat.
11788 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11789 * dwarf2out.c (highest_c_language): Handle C++17.
11790 (gen_compile_unit_die): Likewise.
11792 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11794 PR rtl-optimization/82192
11795 * combine.c (make_extraction): Don't look through non-paradoxical
11796 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11799 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11800 Alan Hayward <alan.hayward@arm.com>
11801 David Sherwood <david.sherwood@arm.com>
11803 * target.def (function_arg_offset): New hook.
11804 * targhooks.h (default_function_arg_offset): Declare.
11805 * targhooks.c (default_function_arg_offset): New function.
11806 * function.c (locate_and_pad_parm): Use
11807 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11808 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11809 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11810 * doc/tm.texi: Regenerate.
11811 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11812 * config/spu/spu.c (spu_function_arg_offset): New function.
11813 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11814 * system.h (FUNCTION_ARG_OFFSET): Poison.
11816 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11817 Alan Hayard <alan.hayward@arm.com>
11818 David Sherwood <david.sherwood@arm.com>
11820 * target.def (truly_noop_truncation): New hook.
11821 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11822 than TRULY_NOOP_TRUNCATION.
11823 * hooks.h (hook_bool_uint_uint_true): Declare.
11824 * hooks.c (hook_bool_uint_uint_true): New function.
11825 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11826 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11827 * doc/tm.texi: Regenerate.
11828 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11829 rather than TRULY_NOOP_TRUNCATION in comments.
11830 (simplify_comparison): Likewise.
11831 (record_truncated_value): Likewise.
11832 * expmed.c (extract_bit_field_1): Likewise.
11833 (extract_split_bit_field): Likewise.
11834 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11835 instead of TRULY_NOOP_TRUNCATION.
11836 * function.c (assign_parm_setup_block): Likewise.
11837 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11838 * rtlhooks.c: Include target.h.
11839 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11840 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11841 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11842 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11843 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11844 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11845 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11846 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11847 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11848 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11849 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11850 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11851 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11852 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11853 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11854 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11855 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11856 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11857 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11858 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11859 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11860 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11861 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11862 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11863 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11864 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11865 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11866 rather than TRULY_NOOP_TRUNCATION in comments.
11867 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11868 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11869 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11870 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11871 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11872 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11873 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11874 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11875 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11876 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11877 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11878 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11879 rather than TRULY_NOOP_TRUNCATION in comments.
11880 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11881 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11882 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11883 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11884 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11885 TRULY_NOOP_TRUNCATION condition.
11886 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11887 (TRULY_NOOP_TRUNCATION): Delete.
11888 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11889 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11890 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11891 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11892 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11893 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11894 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11895 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11896 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11897 rather than TRULY_NOOP_TRUNCATION in comments.
11898 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11899 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11900 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11901 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11902 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11903 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11905 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11908 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11909 (*cmp_ior): Likewise.
11910 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11911 (*ior_scc_scc_cmp): Likewise.
11912 (*and_scc_scc): Likewise.
11913 (*and_scc_scc_cmp): Likewise.
11915 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11916 Alan Hayard <alan.hayward@arm.com>
11917 David Sherwood <david.sherwood@arm.com>
11919 * target.def (can_change_mode_class): New hook.
11920 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11921 (hard_regno_nregs): Likewise.
11922 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11923 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11924 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11925 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11926 (LOAD_EXTEND_OP): Update accordingly.
11927 * doc/tm.texi: Regenerate.
11928 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11929 CANNOT_CHANGE_MODE_CLASS.
11930 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11931 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11932 * combine.c (simplify_set): Update accordingly.
11933 * emit-rtl.c (validate_subreg): Likewise.
11934 * recog.c (general_operand): Likewise.
11935 * regcprop.c (mode_change_ok): Likewise.
11936 * reload1.c (choose_reload_regs): Likewise.
11937 (inherit_piecemeal_p): Likewise.
11938 * rtlanal.c (simplify_subreg_regno): Likewise.
11939 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11940 instead of CANNOT_CHANGE_MODE_CLASS.
11941 (reload_cse_simplify_operands): Likewise.
11942 * reload.c (push_reload): Use targetm.can_change_mode_class
11943 instead of CANNOT_CHANGE_MODE_CLASS.
11944 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11945 REG_CANNOT_CHANGE_MODE_P.
11946 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11947 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11948 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11949 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11950 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11951 (arm_can_change_mode_class): New function.
11952 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11953 than CANNOT_CHANGE_MODE_CLASS in comments.
11954 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11955 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11956 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11957 (ix86_can_change_mode_class): ...this new function, inverting the
11958 sense of the return value.
11959 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11960 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11961 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11962 (ia64_can_change_mode_class): New function.
11963 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11964 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11965 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11966 (m32c_can_change_mode_class): ...this new function, inverting the
11967 sense of the return value.
11968 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11969 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11970 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11971 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11972 (mips_can_change_mode_class): ...this new function, inverting the
11973 sense of the return value.
11974 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11975 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11976 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11977 (msp430_can_change_mode_class): New function.
11978 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11979 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11980 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11981 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11982 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11983 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11984 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11985 (pa_cannot_change_mode_class): Replace with...
11986 (pa_can_change_mode_class): ...this new function, inverting the
11987 sense of the return value.
11988 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11989 than CANNOT_CHANGE_MODE_CLASS in comments.
11990 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11991 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11992 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11993 (pdp11_cannot_change_mode_class): Replace with...
11994 (pdp11_can_change_mode_class): ...this new function, inverting the
11995 sense of the return value.
11996 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11997 * config/powerpcspe/powerpcspe-protos.h
11998 (rs6000_cannot_change_mode_class_ptr): Delete.
11999 * config/powerpcspe/powerpcspe.c
12000 (rs6000_cannot_change_mode_class_ptr): Delete.
12001 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12002 (rs6000_option_override_internal): Assign to
12003 targetm.can_change_mode_class instead of
12004 rs6000_cannot_change_mode_class_ptr.
12005 (rs6000_cannot_change_mode_class): Replace with...
12006 (rs6000_can_change_mode_class): ...this new function, inverting the
12007 sense of the return value.
12008 (rs6000_debug_cannot_change_mode_class): Replace with...
12009 (rs6000_debug_can_change_mode_class): ...this new function.
12010 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12011 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12012 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12013 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12014 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12016 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12017 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12018 (rs6000_option_override_internal): Assign to
12019 targetm.can_change_mode_class instead of
12020 rs6000_cannot_change_mode_class_ptr.
12021 (rs6000_cannot_change_mode_class): Replace with...
12022 (rs6000_can_change_mode_class): ...this new function, inverting the
12023 sense of the return value.
12024 (rs6000_debug_cannot_change_mode_class): Replace with...
12025 (rs6000_debug_can_change_mode_class): ...this new function.
12026 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12027 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12028 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12029 (s390_can_change_mode_class): ...this new function, inverting the
12030 sense of the return value.
12031 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12032 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12033 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12034 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12035 (sh_cannot_change_mode_class): Replace with...
12036 (sh_can_change_mode_class): ...this new function, inverting the
12037 sense of the return value.
12038 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12039 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12040 (sparc_can_change_mode_class): New function.
12041 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12042 * config/spu/spu.c (spu_can_change_mode_class): New function.
12043 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12044 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12045 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12046 (visium_can_change_mode_class): New function.
12047 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12049 2017-09-15 Richard Biener <rguenther@suse.de>
12051 PR tree-optimization/82217
12052 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12053 but not undefined case.
12055 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12058 * postreload.c (reload_cse_simplify_operands): Skip
12059 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12061 2017-09-15 Richard Biener <rguenther@suse.de>
12063 PR tree-optimization/68823
12064 * graphite-scop-detection.c (build_alias_set): If we have a
12065 possible dependence check whether we can handle them by just
12066 looking at the DRs DR_ACCESS_FNs.
12067 (build_scops): If build_alias_set fails, fail the SCOP.
12069 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12071 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12072 to support float128 built-in functions that require the ISA 3.0
12074 (BU_FLOAT128_3_HW): Likewise.
12075 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12076 built-in functions.
12077 (FMAF128): Likewise.
12079 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12080 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12081 floating point instructions.
12082 (rs6000_invalid_builtin): Likewise.
12083 (rs6000_builtin_mask_names): Likewise.
12084 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12085 (RS6000_BTM_FLOAT128_HW): Likewise.
12086 (RS6000_BTM_COMMON): Likewise.
12087 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12089 * doc/extend.texi (RS/6000 built-in functions): Document the
12090 IEEE 128-bit floating point square root and fused multiply-add
12091 built-in functions.
12093 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12095 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12096 reg (r2) isn't in the set of registers defined in the prologue.
12098 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12099 Alan Hayward <alan.hayward@arm.com>
12100 David Sherwood <david.sherwood@arm.com>
12102 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12103 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12104 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12105 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12106 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12108 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12109 max_vectorization_factor.
12110 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12112 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12113 Alan Hayward <alan.hayward@arm.com>
12114 David Sherwood <david.sherwood@arm.com>
12116 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12117 (vect_worthwhile_without_simd_p): Declare.
12118 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12119 (vectorizable_reduction): Use it.
12120 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12121 (vectorizable_operation): Likewise.
12123 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12124 Alan Hayward <alan.hayward@arm.com>
12125 David Sherwood <david.sherwood@arm.com>
12127 * tree-vectorizer.h (vect_get_num_copies): New function.
12128 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12129 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12130 (vectorizable_induction): Likewise.
12131 (vectorizable_live_operation): Likewise.
12132 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12133 (vectorizable_bswap): Likewise.
12134 (vectorizable_call): Likewise.
12135 (vectorizable_conversion): Likewise.
12136 (vectorizable_assignment): Likewise.
12137 (vectorizable_shift): Likewise.
12138 (vectorizable_operation): Likewise.
12139 (vectorizable_store): Likewise.
12140 (vectorizable_load): Likewise.
12141 (vectorizable_condition): Likewise.
12142 (vectorizable_comparison): Likewise.
12143 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12145 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12146 Alan Hayward <alan.hayward@arm.com>
12147 David Sherwood <david.sherwood@arm.com>
12149 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12150 of vect_init_vector.
12152 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12153 Alan Hayward <alan.hayward@arm.com>
12154 David Sherwood <david.sherwood@arm.com>
12156 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12157 an inline wrapper that provides a location.
12158 (gimple_build_vector): Likewise.
12159 * gimple-fold.c (gimple_build_vector_from_val): New function.
12160 (gimple_build_vector): Likewise.
12161 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12162 functions to build the initial value. Always return a gimple value.
12163 (get_initial_defs_for_reduction): Likewise. Only compute
12165 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12166 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12167 (vectorizable_induction): Use gimple_build_vector rather than
12170 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12171 Alan Hayward <alan.hayward@arm.com>
12172 David Sherwood <david.sherwood@arm.com>
12174 * target.h (vec_perm_indices): New typedef.
12175 (auto_vec_perm_indices): Likewise.
12176 * optabs-query.h: Include target.h
12177 (can_vec_perm_p): Take a vec_perm_indices *.
12178 * optabs-query.c (can_vec_perm_p): Likewise.
12179 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12180 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12181 * tree-vect-generic.c (lower_vec_perm): Likewise.
12182 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12183 (vect_grouped_load_supported): Likewise.
12184 (vect_shift_permute_load_chain): Likewise.
12185 (vect_permute_store_chain): Use auto_vec_perm_indices.
12186 (vect_permute_load_chain): Likewise.
12187 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12188 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12189 Update uses of can_vec_perm_p.
12190 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12191 mode with a number of elements. Take a vec_perm_indices *.
12192 (vect_create_epilog_for_reduction): Update accordingly.
12193 Use auto_vec_perm_indices.
12194 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12195 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12196 (vect_transform_slp_perm_load): Likewise.
12197 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12198 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12199 (vect_gen_perm_mask_checked): Likewise.
12200 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12201 (vect_gen_perm_mask_checked): Likewise.
12202 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12203 (vectorizable_store): Likewise.
12204 (vectorizable_load): Likewise.
12205 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12206 (vectorizable_bswap): Likewise.
12208 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12209 Alan Hayward <alan.hayward@arm.com>
12210 David Sherwood <david.sherwood@arm.com>
12212 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12213 * tree.c (build_vector): Likewise.
12214 (build_vector_from_ctor): Update accordingly.
12215 (build_vector_from_val): Likewise.
12216 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12217 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12218 * tree-vect-generic.c (add_rshift): Likewise.
12219 (expand_vector_divmod): Likewise.
12220 (optimize_vector_constructor): Likewise.
12221 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12222 (vect_transform_slp_perm_load): Likewise.
12223 (vect_schedule_slp_instance): Likewise.
12224 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12225 (vectorizable_call): Likewise.
12226 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12227 * expmed.c (make_tree): Likewise.
12228 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12229 a vector passed to build_vector.
12230 (fold_convert_const): Likewise.
12231 (exact_inverse): Likewise.
12232 (fold_ternary_loc): Likewise.
12233 (fold_relational_const): Likewise.
12234 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12235 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12236 (const_unop): Likewise. Store the reduction accumulator in a
12237 variable rather than an array.
12238 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12239 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12240 the new vector, rather than constructing it after the input arrays.
12241 (native_interpret_vector): Use auto_vec<tree> when building
12242 a vector passed to build_vector. Add elements in order.
12243 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12244 auto_vec<tree> when building a vector passed to build_vector.
12245 (vect_create_epilog_for_reduction): Likewise.
12246 (vectorizable_induction): Likewise.
12247 (get_initial_def_for_reduction): Likewise. Fix indentation of
12249 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12251 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12252 passed to build_vector.
12254 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12255 Alan Hayward <alan.hayward@arm.com>
12256 David Sherwood <david.sherwood@arm.com>
12258 * tree-core.h (tree_base::u): Add an "nelts" field.
12259 (tree_vector): Use VECTOR_CST_NELTS as the length.
12260 * tree.c (tree_size): Likewise.
12261 (make_vector): Initialize VECTOR_CST_NELTS.
12262 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12263 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12264 TYPE_VECTOR_SUBPARTS.
12265 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12266 as the number of units, setting it from VECTOR_CST_NELTS.
12267 (const_vector_from_tree): Likewise.
12268 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12269 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12270 (fold_negate_expr_1): Likewise.
12271 (fold_convert_const): Likewise.
12272 (const_binop): Likewise. Differentiate the number of output and
12274 (const_unop): Likewise.
12275 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12276 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12277 in cases that did the opposite.
12279 2017-09-14 Richard Biener <rguenther@suse.de>
12281 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12284 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12286 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12288 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12291 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12292 if and where to split a bb, except for splitting before debug insn
12293 sequences followed by non-label real insn. Delete debug insns
12294 in between basic blocks.
12296 * combine.c (make_compound_operation_int): Formatting fixes.
12298 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12299 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12300 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12301 * config/sol2.h (LINK_EH_SPEC): Likewise.
12302 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12303 * config/s390/linux.h (LINK_SPEC): Likewise.
12304 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12305 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12306 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12307 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12308 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12309 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12310 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12311 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12313 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12315 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12317 (ENDFILE_LINUX_SPEC): Likewise.
12318 (LINK_EH_SPEC): Likewise.
12319 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12320 (LINK_OS_LINUX_SPEC32): Likewise.
12321 (LINK_OS_LINUX_SPEC64): Likewise.
12322 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12323 (LINK_OS_LINUX_SPEC): Likewise.
12325 2017-09-13 Martin Liska <mliska@suse.cz>
12327 PR middle-end/82154
12328 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12329 CASE_HIGH is NULL_TREE.
12331 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12332 Alan Hayward <alan.hayward@arm.com>
12333 David Sherwood <david.sherwood@arm.com>
12335 * target.def (secondary_memory_needed): New hook.
12336 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12337 instead of SECONDARY_MEMORY_NEEDED.
12338 (secondary_memory_needed_mode): Likewise.
12339 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12340 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12341 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12342 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12343 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12344 * doc/tm.texi: Regenerate.
12345 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12346 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12347 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12348 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12349 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12350 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12351 mode argument first and change the reg_class arguments to reg_class_t.
12352 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12353 Make static. Update the call to inline_secondary_memory_needed.
12354 (ix86_register_move_cost): Update the call to
12355 inline_secondary_memory_needed.
12356 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12357 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12359 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12360 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12362 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12363 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12364 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12365 and match hook interface. Add comment from mips.h.
12366 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12367 * config/mmix/mmix.md (truncdfsf2): Refer to
12368 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12370 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12371 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12372 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12373 (pa_secondary_memory_needed): New function.
12374 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12375 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12376 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12377 (pdp11_secondary_memory_needed): Make static and match hook interface.
12378 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12379 * config/powerpcspe/powerpcspe-protos.h
12380 (rs6000_secondary_memory_needed_ptr): Delete.
12381 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12383 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12384 (rs6000_option_override_internal): Assign to
12385 targetm.secondary_memory_needed rather than
12386 rs6000_secondary_memory_needed_ptr.
12387 (rs6000_secondary_memory_needed): Match hook interface.
12388 (rs6000_debug_secondary_memory_needed): Likewise.
12389 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12390 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12391 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12392 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12393 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12394 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12396 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12397 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12398 (rs6000_option_override_internal): Assign to
12399 targetm.secondary_memory_needed rather than
12400 rs6000_secondary_memory_needed_ptr.
12401 (rs6000_secondary_memory_needed): Match hook interface.
12402 (rs6000_debug_secondary_memory_needed): Likewise.
12403 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12404 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12405 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12406 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12407 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12408 (sparc_secondary_memory_needed): New function.
12409 * lra-constraints.c (check_and_process_move): Refer to
12410 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12412 (curr_insn_transform): Likewise.
12413 (process_alt_operands): Use targetm.secondary_memory_needed
12414 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12415 (check_secondary_memory_needed_p): Likewise.
12416 (choose_split_class): Likewise.
12417 * reload.c: Unconditionally include code that was previously
12418 conditional on SECONDARY_MEMORY_NEEDED.
12419 (push_secondary_reload): Use targetm.secondary_memory_needed
12420 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12421 (push_reload): Likewise.
12422 * reload1.c: Unconditionally include code that was previously
12423 conditional on SECONDARY_MEMORY_NEEDED.
12424 (choose_reload_regs): Use targetm.secondary_memory_needed
12425 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12426 (gen_reload): Likewise.
12427 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12429 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12430 Alan Hayward <alan.hayward@arm.com>
12431 David Sherwood <david.sherwood@arm.com>
12433 * target.def (secondary_memory_needed_mode): New hook:
12434 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12435 * targhooks.h (default_secondary_memory_needed_mode): New function.
12436 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12437 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12438 * doc/tm.texi: Regenerate.
12439 * lra-constraints.c (check_and_process_move): Use
12440 targetm.secondary_memory_needed_mode instead of
12441 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12442 (curr_insn_transform): Likewise.
12443 * reload.c (get_secondary_mem): Likewise.
12444 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12445 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12447 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12448 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12449 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12450 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12451 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12453 * config/powerpcspe/powerpcspe-protos.h
12454 (rs6000_secondary_memory_needed_mode): Delete.
12455 * config/powerpcspe/powerpcspe.c
12456 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12457 (rs6000_secondary_memory_needed_mode): Make static.
12458 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12459 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12461 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12463 (rs6000_secondary_memory_needed_mode): Make static.
12464 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12465 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12466 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12467 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12468 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12470 (sparc_secondary_memory_needed_mode): New function.
12471 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12473 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12475 * config/aarch64/constraints.md (Umq): New constraint.
12476 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12478 (mov<mode>): Update condition.
12480 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12482 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12483 when bitposition is the same.
12485 2017-09-13 Richard Biener <rguenther@suse.de>
12487 * dwarf2out.c (output_die_symbol): Remove.
12488 (output_die): Do not output a DIEs symbol.
12490 2017-09-13 Richard Biener <rguenther@suse.de>
12492 PR middle-end/82128
12493 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12494 default-def to avoid breaking iterator update with the weird
12495 interaction with cgraph_update_edges_for_call_stmt_node.
12497 2017-09-13 Richard Biener <rguenther@suse.de>
12499 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12500 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12501 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12502 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12504 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12506 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12507 Disable pc relative literal load irrespective of
12508 TARGET_FIX_ERR_A53_84341 for default.
12510 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12512 * config/sparc/sparc.c (output_return): Output the source location of
12513 the insn in the delay slot, if any.
12514 (output_sibcall): Likewise.
12516 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12519 * common.opt (-static-pie): New alias.
12520 (shared): Negate static-pie.
12521 (-no-pie): Update help text.
12523 (static-pie): New option.
12524 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12525 -static-pie support.
12526 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12527 (LINK_EH_SPEC): Likewise.
12528 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12529 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12530 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12531 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12532 (init_gcc_specs): Likewise.
12533 (init_spec): Likewise.
12534 (display_help): Update help message for -pie.
12535 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12538 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12540 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12541 (movdi_aarch64): Likewise.
12542 (movti_aarch64): Likewise.
12544 2017-09-12 Simon Wright <simon@pushface.org>
12547 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12548 calculation of the minor version, always output as 0.
12550 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12553 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12554 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12555 on it early, rather than manual conversion late. For
12556 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12557 instead of performing manual conversion.
12559 2017-09-12 Carl Love <cel@us.ibm.com>
12561 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12562 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12564 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12565 VMULOSW): Add definitions.
12566 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12567 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12568 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12569 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12570 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12572 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12574 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12576 (movti_aarch64): Likewise.
12577 (movdf_aarch64): Likewise.
12578 (movtf_aarch64): Likewise.
12579 (load_pairdi): Likewise.
12580 (store_pairdi): Likewise.
12581 (load_pairdf): Likewise.
12582 (store_pairdf): Likewise.
12583 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12584 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12585 (ldr_got_small_<mode>): Likewise.
12586 (ldr_got_small_28k_<mode>): Likewise.
12587 (ldr_got_tiny): Likewise.
12588 * config/aarch64/iterators.md (ldst_sz): New.
12589 (ldpstp_sz): Likewise.
12590 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12592 (thunderx_load): Split load_8 to load_16.
12593 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12595 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12596 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12597 (xgene1_store_pair): Split store_8 to store_16.
12598 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12599 (falkor_st_0_st_sd): Split store_8 to store_16.
12601 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12603 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12604 and store1/2/3/4 to store_4/8/12/16.
12605 * config/aarch64/aarch64.md: Update for rename.
12606 * config/arm/arm.md: Likewise.: Likewise.
12607 * config/arm/arm.c: Likewise.
12608 * config/arm/thumb1.md: Likewise.
12609 * config/arm/thumb2.md: Likewise.
12610 * config/arm/vfp.md: Likewise.
12611 * config/arm/arm-generic.md: Likewise.
12612 * config/arm/arm1020e.md: Likewise.
12613 * config/arm/arm1026ejs.md: Likewise.
12614 * config/arm/arm1136jfs.md: Likewise.
12615 * config/arm/arm926ejs.md: Likewise.
12616 * config/arm/cortex-a15.md: Likewise.
12617 * config/arm/cortex-a17.md: Likewise.
12618 * config/arm/cortex-a5.md: Likewise.
12619 * config/arm/cortex-a53.md: Likewise.
12620 * config/arm/cortex-a57.md: Likewise.
12621 * config/arm/cortex-a7.md: Likewise.
12622 * config/arm/cortex-a8.md: Likewise.
12623 * config/arm/cortex-a9.md: Likewise.
12624 * config/arm/cortex-m4.md: Likewise.
12625 * config/arm/cortex-m7.md: Likewise.
12626 * config/arm/cortex-r4.md: Likewise.
12627 * config/arm/exynos-m1.md: Likewise.
12628 * config/arm/fa526.md: Likewise.
12629 * config/arm/fa606te.md: Likewise.
12630 * config/arm/fa626te.md: Likewise.
12631 * config/arm/fa726te.md: Likewise.
12632 * config/arm/fmp626.md: Likewise.
12633 * config/arm/iwmmxt.md: Likewise.
12634 * config/arm/ldmstm.md: Likewise.
12635 * config/arm/marvell-pj4.md: Likewise.
12636 * config/arm/xgene1.md: Likewise.
12637 * config/aarch64/thunderx.md: Likewise.
12638 * config/aarch64/thunderx2t99.md: Likewise.
12639 * config/aarch64/falkor.md: Likewise.
12641 2017-09-12 Martin Liska <mliska@suse.cz>
12643 * attribs.c (private_lookup_attribute): New function.
12644 * attribs.h (private_lookup_attribute): Declared here.
12645 (lookup_attribute): Called from this place.
12647 2017-09-12 Richard Biener <rguenther@suse.de>
12649 PR tree-optimization/82157
12650 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12651 stmts with side-effects.
12653 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12654 Alan Hayward <alan.hayward@arm.com>
12655 David Sherwood <david.sherwood@arm.com>
12657 * target.def (hard_regno_nregs): New hook.
12658 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12659 * targhooks.h (default_hard_regno_nregs): Declare.
12660 * targhooks.c (default_hard_regno_nregs): New function.
12661 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12662 (TARGET_HARD_REGNO_NREGS): ...this hook.
12663 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12664 (CLASS_MAX_NREGS): Likewise.
12665 * doc/tm.texi: Regenerate.
12666 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12667 instead of HARD_REGNO_NREGS.
12668 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12669 HARD_REGNO_NREGS in the comment.
12670 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12671 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12672 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12673 Return an unsigned int.
12674 (TARGET_HARD_REGNO_NREGS): Redefine.
12675 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12676 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12677 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12678 (arc_hard_regno_nregs): New function.
12679 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12680 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12681 (arm_hard_regno_nregs): New function.
12682 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12683 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12684 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12685 (TARGET_HARD_REGNO_NREGS): Redefine.
12686 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12687 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12688 (HARD_REGNO_NREGS): Delete.
12689 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12690 (cr16_hard_regno_nregs): New function.
12691 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12692 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12693 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12694 (cris_hard_regno_nregs): New function.
12695 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12696 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12697 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12698 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12699 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12700 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12701 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12702 (frv_hard_regno_nregs): Make static. Take and return an
12704 (frv_class_max_nregs): Remove outdated copy of documentation.
12705 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12706 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12707 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12708 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12709 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12710 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12711 (TARGET_HARD_REGNO_NREGS): Redefine.
12712 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12713 (CLASS_MAX_NREGS): Update comment.
12714 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12715 (ia64_hard_regno_nregs): New function.
12716 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12717 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12718 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12719 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12720 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12722 (m32c_hard_regno_nregs): Likewise. Make static.
12723 (TARGET_HARD_REGNO_NREGS): Redefine.
12724 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12725 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12726 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12727 (m68k_hard_regno_nregs): New function.
12728 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12729 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12730 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12731 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12732 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12733 Take and return an unsigned int.
12734 (TARGET_HARD_REGNO_NREGS): Redefine.
12735 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12736 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12737 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12738 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12739 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12740 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12741 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12742 (msp430_hard_regno_nregs): Make static. Take and return an
12744 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12745 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12746 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12747 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12748 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12749 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12750 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12751 (TARGET_HARD_REGNO_NREGS): Redefine.
12752 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12753 (PA_HARD_REGNO_NREGS): ...this.
12754 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12755 (PA_HARD_REGNO_NREGS): ...this.
12756 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12757 (pa_hard_regno_nregs): New function.
12758 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12759 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12760 (pdp11_hard_regno_nregs): New function.
12761 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12762 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12763 (rs6000_hard_regno_nregs_hook): New function.
12764 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12765 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12766 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12767 Take and return an unsigned int. Move earlier in file.
12768 (TARGET_HARD_REGNO_NREGS): Redefine.
12769 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12770 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12771 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12772 (rl78_hard_regno_nregs): Make static. Take and return an
12774 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12775 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12776 (rs6000_hard_regno_nregs_hook): New function.
12777 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12778 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12779 (TARGET_HARD_REGNO_NREGS): Redefine.
12780 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12781 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12782 instead of HARD_REGNO_NREGS.
12783 (s390_hard_regno_nregs): New function.
12784 (s390_hard_regno_mode_ok): Add comment from s390.h.
12785 (TARGET_HARD_REGNO_NREGS): Redefine.
12786 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12787 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12788 (sh_hard_regno_nregs): New function.
12789 (sh_pass_in_reg_p): Use it.
12790 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12791 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12792 (sparc_hard_regno_nregs): New function.
12793 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12794 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12795 (spu_function_arg_advance): Use it, supplying a valid register number.
12796 (TARGET_HARD_REGNO_NREGS): Redefine.
12797 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12798 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12799 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12800 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12801 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12802 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12803 (CLASS_MAX_NREGS): Remove copy of old documentation.
12804 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12805 (visium_hard_regno_nregs): New function.
12806 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12807 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12808 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12809 xtensa_hard_regno_nregs): New function.
12810 * system.h (HARD_REGNO_NREGS): Poison.
12812 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12814 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12815 hard_regno_nregs instead of HARD_REGNO_NREGS.
12816 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12817 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12818 (c6x_expand_epilogue): Likewise.
12819 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12820 (frv_read_iacc_argument): Likewise.
12821 * config/sh/sh.c: Include regs.h.
12822 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12823 (regs_used): Likewise.
12824 (output_stack_adjust): Likewise.
12825 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12826 * expmed.c: Include regs.h.
12827 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12828 * ree.c: Include regs.h.
12829 (combine_reaching_defs): Use hard_regno_nregs instead of
12831 (add_removable_extension): Likewise.
12833 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12835 * regs.h (hard_regno_nregs): Turn into a function.
12836 (end_hard_regno): Update accordingly.
12837 * caller-save.c (setup_save_areas): Likewise.
12838 (save_call_clobbered_regs): Likewise.
12839 (replace_reg_with_saved_mem): Likewise.
12840 (insert_restore): Likewise.
12841 (insert_save): Likewise.
12842 * combine.c (can_change_dest_mode): Likewise.
12843 (move_deaths): Likewise.
12844 (distribute_notes): Likewise.
12845 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12846 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12847 (rs6000_split_multireg_move): Likewise.
12848 (rs6000_register_move_cost): Likewise.
12849 (rs6000_memory_move_cost): Likewise.
12850 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12851 (rs6000_split_multireg_move): Likewise.
12852 (rs6000_register_move_cost): Likewise.
12853 (rs6000_memory_move_cost): Likewise.
12854 * cselib.c (cselib_reset_table): Likewise.
12855 (cselib_lookup_1): Likewise.
12856 * emit-rtl.c (set_mode_and_regno): Likewise.
12857 * function.c (aggregate_value_p): Likewise.
12858 * ira-color.c (setup_profitable_hard_regs): Likewise.
12859 (check_hard_reg_p): Likewise.
12860 (calculate_saved_nregs): Likewise.
12861 (assign_hard_reg): Likewise.
12862 (improve_allocation): Likewise.
12863 (calculate_spill_cost): Likewise.
12864 * ira-emit.c (modify_move_list): Likewise.
12865 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12866 (ira_hard_reg_in_set_p): Likewise.
12867 * ira.c (setup_reg_mode_hard_regset): Likewise.
12868 (clarify_prohibited_class_mode_regs): Likewise.
12869 (check_allocation): Likewise.
12870 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12871 (lra_setup_reg_renumber): Likewise.
12872 (setup_try_hard_regno_pseudos): Likewise.
12873 (spill_for): Likewise.
12874 (assign_hard_regno): Likewise.
12875 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12876 * lra-constraints.c (in_class_p): Likewise.
12877 (lra_constraint_offset): Likewise.
12878 (simplify_operand_subreg): Likewise.
12879 (lra_constraints): Likewise.
12880 (split_reg): Likewise.
12881 (split_if_necessary): Likewise.
12882 (invariant_p): Likewise.
12883 (inherit_in_ebb): Likewise.
12884 * lra-lives.c (process_bb_lives): Likewise.
12885 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12886 (get_hard_regs): Likewise.
12887 (do_remat): Likewise.
12888 * lra-spills.c (assign_spill_hard_regs): Likewise.
12889 * mode-switching.c (create_pre_exit): Likewise.
12890 * postreload.c (reload_combine_recognize_pattern): Likewise.
12891 * recog.c (peep2_find_free_register): Likewise.
12892 * regcprop.c (kill_value_regno): Likewise.
12893 (set_value_regno): Likewise.
12894 (copy_value): Likewise.
12895 (maybe_mode_change): Likewise.
12896 (find_oldest_value_reg): Likewise.
12897 (copyprop_hardreg_forward_1): Likewise.
12898 * regrename.c (check_new_reg_p): Likewise.
12899 (regrename_do_replace): Likewise.
12900 * reload.c (push_reload): Likewise.
12901 (combine_reloads): Likewise.
12902 (find_dummy_reload): Likewise.
12903 (operands_match_p): Likewise.
12904 (find_reloads): Likewise.
12905 (find_equiv_reg): Likewise.
12906 (reload_adjust_reg_for_mode): Likewise.
12907 * reload1.c (count_pseudo): Likewise.
12908 (count_spilled_pseudo): Likewise.
12909 (find_reg): Likewise.
12910 (clear_reload_reg_in_use): Likewise.
12911 (free_for_value_p): Likewise.
12912 (allocate_reload_reg): Likewise.
12913 (choose_reload_regs): Likewise.
12914 (reload_adjust_reg_for_temp): Likewise.
12915 (emit_reload_insns): Likewise.
12916 (delete_output_reload): Likewise.
12917 * rtlanal.c (subreg_get_info): Likewise.
12918 * sched-deps.c (sched_analyze_reg): Likewise.
12919 * sel-sched.c (init_regs_for_mode): Likewise.
12920 (mark_unavailable_hard_regs): Likewise.
12921 (choose_best_reg_1): Likewise.
12922 (verify_target_availability): Likewise.
12923 * valtrack.c (dead_debug_insert_temp): Likewise.
12924 * var-tracking.c (track_loc_p): Likewise.
12925 (emit_note_insn_var_location): Likewise.
12926 * varasm.c (make_decl_rtl): Likewise.
12927 * reginfo.c (choose_hard_reg_mode): Likewise.
12928 (init_reg_modes_target): Refer directly to
12929 this_target_regs->x_hard_regno_nregs.
12931 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12933 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12934 instead of hard_regno_nregs.
12936 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12938 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12939 end_hard_regno instead of hard_regno_nregs.
12940 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12941 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12942 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12943 * ira-color.c (improve_allocation): Likewise.
12944 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12945 * lra-lives.c (mark_regno_live): Likewise.
12946 (mark_regno_dead): Likewise.
12947 * lra-remat.c (operand_to_remat): Likewise.
12948 * lra.c (collect_non_operand_hard_regs): Likewise.
12949 * postreload.c (reload_combine_note_store): Likewise.
12950 (move2add_valid_value_p): Likewise.
12951 * reload.c (regno_clobbered_p): Likewise.
12953 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12955 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12957 * config/v850/v850.c (v850_reorg): Likewise.
12958 * reload.c (refers_to_regno_for_reload_p): Likewise.
12959 (find_equiv_reg): Likewise.
12960 * reload1.c (reload_reg_reaches_end_p): Likewise.
12962 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12964 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12966 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12967 * config/arm/arm.c (output_move_neon): Likewise.
12968 (arm_attr_length_move_neon): Likewise.
12969 (neon_split_vcombine): Likewise.
12970 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12971 (c6x_mark_reg_written): Likewise.
12972 (c6x_dwarf_register_span): Likewise.
12973 * config/i386/i386.c (ix86_save_reg): Likewise.
12974 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12975 (rws_access_reg): Likewise.
12976 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12977 * mode-switching.c (create_pre_exit): Likewise.
12978 * ree.c (combine_reaching_defs): Likewise.
12979 (add_removable_extension): Likewise.
12980 * regcprop.c (find_oldest_value_reg): Likewise.
12981 (copyprop_hardreg_forward_1): Likewise.
12982 * reload.c (reload_inner_reg_of_subreg): Likewise.
12983 (push_reload): Likewise.
12984 (combine_reloads): Likewise.
12985 (find_dummy_reload): Likewise.
12986 (reload_adjust_reg_for_mode): Likewise.
12987 * reload1.c (find_reload_regs): Likewise.
12988 (forget_old_reloads_1): Likewise.
12989 (reload_reg_free_for_value_p): Likewise.
12990 (reload_adjust_reg_for_temp): Likewise.
12991 (emit_reload_insns): Likewise.
12992 (delete_output_reload): Likewise.
12993 * sel-sched.c (choose_best_reg_1): Likewise.
12994 (choose_best_pseudo_reg): Likewise.
12996 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12997 Alan Hayward <alan.hayward@arm.com>
12998 David Sherwood <david.sherwood@arm.com>
13000 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13001 * target.def (slow_unaligned_access): New hook.
13002 * targhooks.h (default_slow_unaligned_access): Declare.
13003 * targhooks.c (default_slow_unaligned_access): New function.
13004 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13005 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13006 * doc/tm.texi: Regenerate.
13007 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13008 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13009 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13011 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13012 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13014 (rs6000_slow_unaligned_access): New function.
13015 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13016 (expand_block_compare): Likewise.
13017 (expand_strn_compare): Likewise.
13018 (rs6000_rtx_costs): Likewise.
13019 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13020 (riscv_slow_unaligned_access): Likewise.
13021 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13022 (riscv_slow_unaligned_access_p): ...this and make static.
13023 (riscv_option_override): Update accordingly.
13024 (riscv_slow_unaligned_access): New function.
13025 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13026 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13027 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13028 (rs6000_slow_unaligned_access): New function.
13029 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13030 (rs6000_rtx_costs): Likewise.
13031 * config/rs6000/rs6000-string.c (expand_block_compare)
13032 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13033 of SLOW_UNALIGNED_ACCESS.
13034 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13035 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13036 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13037 of SLOW_UNALIGNED_ACCESS.
13038 * expmed.c (simple_mem_bitfield_p): Likewise.
13039 * expr.c (alignment_for_piecewise_move): Likewise.
13040 (emit_group_load_1): Likewise.
13041 (emit_group_store): Likewise.
13042 (copy_blkmode_from_reg): Likewise.
13043 (emit_push_insn): Likewise.
13044 (expand_assignment): Likewise.
13045 (store_field): Likewise.
13046 (expand_expr_real_1): Likewise.
13047 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13048 * lra-constraints.c (simplify_operand_subreg): Likewise.
13049 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13050 * gimple-ssa-store-merging.c: Likewise in block comment at start
13052 * tree-ssa-strlen.c: Include target.h.
13053 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13054 of SLOW_UNALIGNED_ACCESS.
13055 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13057 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13059 PR rtl-optimization/82185
13060 * expmed.c (emit_store_flag_int): Only test tem if it has been
13063 2017-09-12 Richard Biener <rguenther@suse.de>
13065 PR middle-end/82149
13066 * match.pd ((FTYPE) N CMP CST): Fix typo.
13068 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13070 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13072 (mips_near_type_p): Add 'short_call' attribute as a synonym
13074 * doc/extend.texi (short_call): Document new function attribute.
13076 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13079 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13080 assertion check that in the condition.
13081 (get_atomic_generic_size): Likewise. Before testing if parameter
13082 has pointer type, if it has array type, call for C++
13083 default_conversion to perform array-to-pointer conversion.
13085 2017-09-12 Richard Biener <rguenther@suse.de>
13087 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13088 for operations we cannot scalarize.
13090 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13092 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13093 vectors heap vectors. Clean up comments.
13094 Make visited_bbs a reference.
13095 (profitable_jump_thread_path): Make GC
13096 vectors heap vectors. Clean up comments.
13098 (convert_and_register_jump_thread_path): Make GC vectors heap
13100 (check_subpath_and_update_thread_path): Same. Clean up comments.
13101 Make visited_bbs a reference.
13102 (handle_phi): Abstract common code to to
13103 register_jump_thread_path_if_profitable.
13104 Rename VAR_BB to DEF_BB.
13106 Make GC vectors heap vectors.
13107 Make visited_bbs a reference.
13108 (handle_assignment): Same.
13109 (register_jump_thread_path_if_profitable): New.
13110 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13112 Make GC vectors heap vectors. Clean up comments.
13113 Make visited_bbs a reference.
13114 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13115 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13118 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13121 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13122 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13124 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13126 Revert r251800 and r251799.
13128 2017-09-11 Martin Jambor <mjambor@suse.cz>
13131 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13132 arguments in advance.
13133 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13134 use it to find predecessor edges.
13135 (naive_outof_ssa): Collect vector of predecessors.
13137 2017-09-08 Jason Merrill <jason@redhat.com>
13139 PR c++/70029 - ICE with ref-qualifier and -flto
13140 * langhooks.h (struct lang_hooks_for_types): Add
13141 copy_lang_qualifiers.
13142 * attribs.c (build_type_attribute_qual_variant): Use it.
13143 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13145 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13146 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13148 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13151 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13152 (*mulsi3_sp64): New instruction.
13153 (mulsi3): New expander.
13155 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13157 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13159 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13161 * sancov.c: Include memmodel.h.
13163 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13166 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13169 2017-09-07 Carl Love <cel@us.ibm.com>
13171 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13172 the sldi instruction.
13174 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13176 * sancov.c: Include tm_p.h.
13178 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13181 * output.h (switch_to_other_text_partition): New declaration.
13182 * varasm.c (switch_to_other_text_partition): New function.
13183 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13184 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13185 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13186 to the other text partition before emitting LCL label and switch back
13187 after emitting the word after it.
13189 2017-09-07 Richard Biener <rguenther@suse.de>
13191 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13192 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13193 critical edges here, after loop init.
13194 (pass_data_pre): Remove PROP_no_crit_edges flags.
13195 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13196 for valueization of call args to avoid leaking VN_TOP.
13197 (visit_use): Assert we do not visit default defs.
13198 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13199 Use error_mark_node to more easily detect leaking VN_TOP.
13200 All default-defs are varying, not VN_TOP. Mark them visited.
13201 (run_scc_vn): Make code match comment.
13203 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13205 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13206 OPTION_MASK_FLOAT128_KEYWORD.
13207 (POWERPC_MASKS): Likewise.
13208 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13209 support for the -mfloat128-type option, and make -mfloat128
13210 default on PowerPC Linux systems. Define or undefine
13211 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13212 Define __float128 to be __ieee128 if IEEE 128-bit support is
13213 enabled, or undefine it.
13214 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13215 Delete defining __FLOAT128_TYPE__.
13216 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13217 -mfloat128-type option and make -mfloat128 default on PowerPC
13219 (TARGET_FLOAT128_TYPE): Likewise.
13220 (-mfloat128-type): Likewise.
13221 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13222 Delete the -mfloat128-type option and make -mfloat128 default on
13223 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13224 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13225 128-bit floating point is enabled. Change tests from using
13226 -mfloat128-type to -mfloat128.
13227 (rs6000_mangle_type): Use the correct mangling for the __float128
13228 type even if normal long double is restricted to 64-bits.
13229 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13231 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13232 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13233 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13235 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13236 documentation for -mfloat128.
13238 2017-09-06 Olivier Hainque <hainque@adacore.com>
13240 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13242 2017-09-06 Wish Wu <wishwu007@gmail.com>
13243 Jakub Jelinek <jakub@redhat.com>
13245 * asan.c (initialize_sanitizer_builtins): Add
13246 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13247 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13248 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13249 BT_FN_VOID_UINT64_PTR variables.
13250 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13251 (BT_FN_VOID_UINT16_UINT16): Likewise.
13252 (BT_FN_VOID_UINT32_UINT32): Likewise.
13253 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13254 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13255 (BT_FN_VOID_UINT64_PTR): Likewise.
13256 * common.opt (flag_sanitize_coverage): New variable.
13257 (fsanitize-coverage=trace-pc): Remove.
13258 (fsanitize-coverage=): Add.
13259 * flag-types.h (enum sanitize_coverage_code): New enum.
13260 * fold-const.c (fold_range_test): Disable non-short-circuit
13261 optimization if flag_sanitize_coverage.
13262 (fold_truth_andor): Likewise.
13263 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13264 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13265 (coverage_sanitizer_opts): New array.
13266 (get_closest_sanitizer_option): Add OPTS argument, handle also
13267 OPT_fsanitize_coverage_.
13268 (parse_sanitizer_options): Adjusted to also handle
13269 OPT_fsanitize_coverage_.
13270 (common_handle_option): Add OPT_fsanitize_coverage_.
13271 * sancov.c (instrument_comparison, instrument_switch): New function.
13272 (sancov_pass): Add trace-cmp support.
13273 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13274 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13275 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13276 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13277 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13278 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13279 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13280 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13281 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13282 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13284 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13286 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13287 error. Only quit immediately if parsing is complete.
13288 (BEGIN): Initialize fatal_err and parse_done.
13289 (begin fpu, end fpu): Check number of arguments.
13290 (begin arch, end arch): Likewise.
13291 (begin cpu, end cpu): Likewise.
13292 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13293 (optalias): Likewise.
13295 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13297 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13298 * config/arm/arm-isa.h: Delete. Move definitions to ...
13299 * arm-cpus.in: ... here. Use new feature and fgroup values.
13300 * config/arm/arm.c (arm_option_override): Use lower case for feature
13302 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13303 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13304 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13306 (print_isa_bits_for): New function.
13307 (gen_isa): New function.
13308 (gen_comm_data): Use print_isa_bits_for.
13309 (define feature): New keyword.
13310 (define fgroup): New keyword.
13311 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13312 (arm-isa.h): Add rule to generate file.
13313 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13314 case for feature bit names.
13316 2017-09-06 Richard Biener <rguenther@suse.de>
13318 * tree-ssa-pre.c (NECESSARY): Remove.
13319 (create_expression_by_pieces): Do not touch pass-local flags.
13320 (insert_into_preds_of_block): Likewise.
13321 (do_pre_regular_insertion): Likewise.
13322 (eliminate_insert): Likewise.
13323 (eliminate_dom_walker::before_dom_children): Likewise.
13324 (fini_eliminate): Do not look at inserted_exprs.
13325 (mark_operand_necessary): Remove.
13326 (remove_dead_inserted_code): Replace with simple work-list
13327 algorithm based on inserted_exprs and SSA uses.
13328 (pass_pre::execute): Re-order fini_eliminate and
13329 remove_dead_inserted_code.
13331 2017-09-06 Olivier Hainque <hainque@adacore.com>
13333 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13334 for VxWorks 7. Adjust surrounding comments.
13336 2017-09-06 Richard Biener <rguenther@suse.de>
13338 * gimple-ssa-strength-reduction.c
13339 (find_candidates_dom_walker::before_dom_children): Also allow
13342 2017-09-06 Richard Biener <rguenther@suse.de>
13344 PR tree-optimization/82108
13345 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13346 for gap in the non-permutation SLP case.
13348 2017-09-06 Martin Jambor <mjambor@suse.cz>
13350 PR tree-optimization/82078
13351 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13352 add_access_to_work_queue...
13353 (build_accesses_from_assign): ...here.
13354 (propagate_all_subaccesses): Make sure racc is the group
13355 representative, if there is one.
13357 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13359 PR middle-end/82095
13360 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13363 2017-09-06 Richard Biener <rguenther@suse.de>
13365 * gimple-ssa-strength-reduction.c
13366 (find_candidates_dom_walker::before_doom_children): Use a
13367 type and not a mode check.
13369 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13372 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13374 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13376 2017-09-05 Jeff Law <law@redhat.com>
13378 PR tree-optimization/64910
13379 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13380 cases where we have 3 or more operands.
13382 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13384 PR middle-end/81768
13385 * omp-low.c (lower_omp_for): Recompute tree invariant if
13386 gimple_omp_for_initial/final is ADDR_EXPR.
13388 PR middle-end/81768
13389 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13390 into gimple val before gimplification fo the COND_EXPR.
13392 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13394 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13395 REGION_COPY argument.
13396 (thread_through_all_blocks): Remove unused argument to
13397 duplicate_thread_path.
13399 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13400 Alan Hayward <alan.hayward@arm.com>
13401 David Sherwood <david.sherwood@arm.com>
13403 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13404 Take a scalar_mode rather than a machine_mode.
13405 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13406 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13407 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13408 (aarch64_gen_adjusted_ldpstp): Likewise.
13409 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13411 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13412 Alan Hayward <alan.hayward@arm.com>
13413 David Sherwood <david.sherwood@arm.com>
13415 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13416 Take a scalar_int_mode instead of a machine_mode.
13417 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13418 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13419 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13420 (aarch64_simd_attr_length_rglist): Delete.
13421 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13422 a scalar_int_mode instead of a machine_mode.
13423 (aarch64_add_offset): Likewise.
13424 (aarch64_internal_mov_immediate): Likewise
13425 (aarch64_add_constant_internal): Likewise.
13426 (aarch64_add_constant): Likewise.
13427 (aarch64_movw_imm): Likewise.
13428 (aarch64_rtx_arith_op_extract_p): Likewise.
13429 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13430 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13431 Remove assert that the mode isn't a vector.
13432 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13433 (aarch64_expand_mov_immediate): Update calls after above changes.
13434 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13435 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13436 (aarch64_move_imm): Likewise.
13437 (aarch64_can_const_movi_rtx_p): Likewise.
13438 (aarch64_strip_extend): Likewise.
13439 (aarch64_extr_rtx_p): Likewise.
13440 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13441 a CONST_INT when the mode parameter is VOIDmode.
13442 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13444 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13446 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13447 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13448 (bitwise_type_for_mode): Update accordingly.
13450 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13452 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13453 * stor-layout.c (mode_for_size_tree): Likewise.
13454 (mode_for_array): Update accordingly.
13455 (layout_decl): Likewise.
13456 (compute_record_mode): Likewise. Only set the mode once.
13458 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13460 * target.def (get_mask_mode): Change return type to opt_mode.
13462 * doc/tm.texi: Regenerate.
13463 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13464 * targhooks.c (default_get_mask_mode): Likewise.
13465 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13466 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13467 targetm.get_mask_mode.
13468 * tree.c (build_truth_vector_type): Likewise.
13470 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13472 * machmode.h (mode_for_vector): Return an opt_mode.
13473 * stor-layout.c (mode_for_vector): Likewise.
13474 (mode_for_int_vector): Update accordingly.
13475 (layout_type): Likewise.
13476 * config/i386/i386.c (emit_memmov): Likewise.
13477 (ix86_expand_set_or_movmem): Likewise.
13478 (ix86_expand_vector_init): Likewise.
13479 (ix86_get_mask_mode): Likewise.
13480 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13482 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13483 * expmed.c (extract_bit_field_1): Likewise.
13484 * expr.c (expand_expr_real_2): Likewise.
13485 * optabs-query.c (can_vec_perm_p): Likewise.
13486 (can_vec_mask_load_store_p): Likewise.
13487 * optabs.c (expand_vec_perm): Likewise.
13488 * targhooks.c (default_get_mask_mode): Likewise.
13489 * tree-vect-stmts.c (vectorizable_store): Likewise.
13490 (vectorizable_load): Likewise.
13491 (get_vectype_for_scalar_type_and_size): Likewise.
13493 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13495 * machmode.h (mode_for_int_vector): New function.
13496 * stor-layout.c (mode_for_int_vector): Likewise.
13497 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13498 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13499 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13500 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13501 (s390_expand_vcond): Likewise.
13503 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13505 * machmode.h (opt_machine_mode): New type.
13506 (opt_mode<T>): Allow construction from anything that can be
13508 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13509 (mode_for_size): Return an opt_machine_mode.
13510 * stor-layout.c (mode_for_size): Likewise.
13511 (mode_for_size_tree): Update call accordingly.
13512 (bitwise_mode_for_mode): Likewise.
13513 (make_fract_type): Likewise.
13514 (make_accum_type): Likewise.
13515 * caller-save.c (replace_reg_with_saved_mem): Update call
13517 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13518 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13519 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13520 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13521 * expmed.c (extract_bit_field_1): Likewise.
13522 * reload.c (get_secondary_mem): Likewise.
13523 * varasm.c (assemble_integer): Likewise.
13524 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13527 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13529 * machmode.h (decimal_float_mode_for_size): New function.
13530 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13531 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13532 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13533 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13534 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13535 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13537 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13539 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13540 (get_builtin_sync_mode): Likewise.
13541 (expand_ifn_atomic_compare_exchange): Likewise.
13542 (expand_builtin_atomic_clear): Likewise.
13543 (expand_builtin_atomic_test_and_set): Likewise.
13544 (fold_builtin_atomic_always_lock_free): Likewise.
13545 * calls.c (compute_argument_addresses): Likewise.
13546 (emit_library_call_value_1): Likewise.
13547 (store_one_arg): Likewise.
13548 * combine.c (combine_instructions): Likewise.
13549 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13550 * config/arm/arm.c (arm_function_value): Likewise.
13551 (aapcs_allocate_return_reg): Likewise.
13552 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13553 * config/i386/i386.c (construct_container): Likewise.
13554 (ix86_gimplify_va_arg): Likewise.
13555 (ix86_expand_sse_cmp): Likewise.
13556 (emit_memmov): Likewise.
13557 (emit_memset): Likewise.
13558 (expand_small_movmem_or_setmem): Likewise.
13559 (ix86_expand_pextr): Likewise.
13560 (ix86_expand_pinsr): Likewise.
13561 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13562 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13564 * config/mips/mips.c (mips_function_value_1) Likewise.
13565 (mips_block_move_straight): Likewise.
13566 (mips_expand_ins_as_unaligned_store): Likewise.
13567 * config/powerpcspe/powerpcspe.c
13568 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13569 (rs6000_darwin64_record_arg_flush): Likewise.
13570 * config/rs6000/rs6000.c
13571 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13572 (rs6000_darwin64_record_arg_flush): Likewise.
13573 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13574 (sparc_function_value_1): Likewise.
13575 * config/spu/spu.c (adjust_operand): Likewise.
13576 (spu_emit_branch_or_set): Likewise.
13577 (arith_immediate_p): Likewise.
13578 * emit-rtl.c (gen_lowpart_common): Likewise.
13579 * expr.c (expand_expr_real_1): Likewise.
13580 * function.c (assign_parm_setup_block): Likewise.
13581 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13582 * reload1.c (alter_reg): Likewise.
13583 * stor-layout.c (mode_for_vector): Likewise.
13584 (layout_type): Likewise.
13586 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13588 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13589 (spu_convert_move): Likewise.
13590 * lower-subreg.c (resolve_simple_move): Likewise.
13592 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13595 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13596 define_insn to a define_expand.
13597 (altivec_vsum2sws_direct): New define_insn.
13598 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13600 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13602 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13605 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13610 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13611 realign stack if stack alignment needed is less than incoming
13614 2017-09-05 Marek Polacek <polacek@redhat.com>
13617 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13620 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13622 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13624 2017-09-05 Richard Biener <rguenther@suse.de>
13626 PR tree-optimization/82084
13627 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13629 2017-09-05 Richard Biener <rguenther@suse.de>
13631 PR tree-optimization/82102
13632 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13634 2017-09-05 Martin Liska <mliska@suse.cz>
13636 PR tree-optimization/82032
13637 * tree-cfg.c (generate_range_test): New function.
13638 * tree-cfg.h (generate_range_test): Declared here.
13639 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13640 (cleanup_control_expr_graph): Use it.
13641 * tree-switch-conversion.c (try_switch_expansion): Remove
13643 (emit_case_nodes): Use generate_range_test.
13645 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13648 * config/i386/i386.md (*<btsc><mode>_mask): Add
13649 TARGET_USE_BT to insn constraint.
13650 (*btr<mode>_mask): Ditto.
13652 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13654 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13655 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13657 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13660 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13661 TARGET_NEON and TARGET_IWMMXT.
13662 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13663 TARGET_NEON and TARGET_IWMMXT.
13664 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13666 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13668 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13669 (ix86_rewrite_tls_address): Ditto.
13670 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13671 (ix86_rewrite_tls_address_1): Ditto.
13672 (ix86_rewrite_tls_address): Ditto.
13673 * config/i386/predicates.md (tls_address_pattern): New predicate.
13674 * config/i386/i386.md (TLS address splitter): New splitter.
13676 2017-09-04 Richard Biener <rguenther@suse.de>
13678 PR tree-optimization/82084
13679 * fold-const.h (can_native_encode_string_p): Declare.
13680 * fold-const.c (can_native_encode_string_p): Factor out from ...
13681 (native_encode_string): ... here.
13682 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13683 vectorizing stores from constants we later cannot handle.
13685 2017-09-04 Marek Polacek <polacek@redhat.com>
13688 * doc/invoke.texi: Update -Wtautological-compare documentation.
13690 2017-09-04 Jeff Law <law@redhat.com>
13692 PR tree-optimization/64910
13693 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13694 swap the first and last operand if the last is a constant.
13696 2017-09-04 Marek Polacek <polacek@redhat.com>
13699 * convert.c (do_narrow): When sanitizing signed integer overflows,
13700 bail out for signed types.
13701 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13703 2017-09-04 Richard Biener <rguenther@suse.de>
13705 PR tree-optimization/82060
13706 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13707 Move devirtualization after stmt folding and before EH/AB/noreturn
13708 cleanup to get the stmt refs canonicalized. Use a bool instead
13709 of gimple_modified_p since that doesn't work for NOPs. Schedule
13710 NOPs generated by folding for removal.
13712 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13713 Alan Hayward <alan.hayward@arm.com>
13714 David Sherwood <david.sherwood@arm.com>
13716 * coretypes.h (pad_direction): New enum.
13717 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13718 (FUNCTION_ARG_PADDING): Likewise.
13719 * target.def (function_arg_padding): New hook.
13720 * targhooks.h (default_function_arg_padding): Declare.
13721 * targhooks.c (default_function_arg_padding): New function.
13722 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13723 (TARGET_FUNCTION_ARG_PADDING): ...this.
13724 * doc/tm.texi: Regenerate.
13725 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13726 instead of direction.
13727 (compute_argument_addresses): Likewise.
13728 (load_register_parameters): Likewise.
13729 (emit_library_call_value_1): Likewise.
13730 (store_one_arg): Use targetm.calls.function_arg_padding instead
13731 of FUNCTION_ARG_PADDING.
13732 (must_pass_in_stack_var_size_or_pad): Likewise.
13733 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13734 (emit_group_store): Likewise.
13735 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13736 instead of FUNCTION_ARG_PADDING.
13737 (emit_push_insn): Likewise, and propagate enum change throughout
13739 * function.h (direction): Delete.
13740 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13742 * function.c (assign_parm_find_stack_rtl): Likewise.
13743 (assign_parm_setup_block_p): Likewise.
13744 (assign_parm_setup_block): Likewise.
13745 (gimplify_parameters): Likewise.
13746 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13747 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13749 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13750 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13751 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13752 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13753 (aarch64_function_arg_padding): ...this new function.
13754 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13755 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13756 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13757 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13758 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13759 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13760 (arm_pad_arg_upward): Replace with...
13761 (arm_function_arg_padding): ...this new function.
13762 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13764 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13765 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13766 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13767 (ia64_hpux_function_arg_padding): Replace with...
13768 (ia64_function_arg_padding): ...this new function. Use pad_direction
13769 instead of direction. Check for TARGET_HPUX.
13770 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13771 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13772 (iq2000_function_arg_padding): New function.
13773 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13774 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13775 (mips_function_arg_padding): ...this new function.
13776 (mips_pad_reg_upward): Update accordingly.
13777 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13778 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13779 targetm.calls.function_arg_padding.
13780 (FUNCTION_ARG_PADDING): Delete.
13781 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13782 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13783 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13784 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13785 (nios2_block_reg_padding): Return pad_direction instead of direction.
13786 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13787 instead of direction.
13788 (nios2_function_arg_padding): Likewise. Make static.
13789 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13790 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13791 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13792 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13793 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13794 (pa_function_arg_padding): Make static. Return pad_direction instead
13796 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13797 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13798 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13799 instead of direction. Use targetm.calls.function_arg_padding.
13800 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13801 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13802 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13803 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13804 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13806 (function_arg_padding): Rename to...
13807 (rs6000_function_arg_padding): ...this. Make static. Return
13808 pad_direction instead of direction.
13809 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13810 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13811 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13812 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13813 instead of direction. Use targetm.calls.function_arg_padding.
13814 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13815 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13816 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13817 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13818 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13819 (function_arg_padding): Rename to...
13820 (rs6000_function_arg_padding): ...this. Make static. Return
13821 pad_direction instead of direction.
13822 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13823 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13824 * config/s390/s390.c (s390_function_arg_padding): New function.
13825 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13826 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13827 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13828 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13829 (function_arg_padding): Rename to...
13830 (sparc_function_arg_padding): ...this. Make static. Return
13831 pad_direction instead of direction.
13832 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13833 * config/spu/spu.c (spu_function_arg_padding): New function.
13834 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13835 * system.h (FUNCTION_ARG_PADDING): Poison.
13837 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13838 Alan Hayward <alan.hayward@arm.com>
13839 David Sherwood <david.sherwood@arm.com>
13841 * target.def (modes_tieable_p): New hook.
13842 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13843 (TARGET_MODES_TIEABLE_P): ...this.
13844 * doc/tm.texi.in: Regenerate.
13845 * hooks.h (hook_bool_mode_mode_true): Declare.
13846 * hooks.c (hook_bool_mode_mode_true): New function.
13847 * combine.c (subst): Use targetm.modes_tieable_p instead of
13849 * dse.c (find_shift_sequence): Likewise.
13850 * expmed.c (extract_low_bits): Likewise.
13851 * lower-subreg.c: Include target.h.
13852 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13854 * rtlanal.c (rtx_cost): Likewise.
13855 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13856 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13857 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13858 (TARGET_MODES_TIEABLE_P): Redefine.
13859 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13860 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13861 (TARGET_MODES_TIEABLE_P): Redefine.
13862 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13863 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13864 (arc_modes_tieable_p): New function.
13865 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13866 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13867 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13868 (arm_modes_tieable_p): Make static.
13869 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13870 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13871 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13872 (TARGET_MODES_TIEABLE_P): Redefine.
13873 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13874 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13875 (TARGET_MODES_TIEABLE_P): Redefine.
13876 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13877 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13878 (cr16_modes_tieable_p): New function.
13879 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13880 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13881 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13882 (TRULY_NOOP_TRUNCATION): Update comment.
13883 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13884 (TRULY_NOOP_TRUNCATION): Update comment.
13885 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13886 (frv_modes_tieable_p): New function.
13887 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13888 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13889 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13890 (TARGET_MODES_TIEABLE_P): Redefine.
13891 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13892 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13893 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13894 (TARGET_MODES_TIEABLE_P): Redefine.
13895 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13896 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13897 (ia64_modes_tieable_p): New function.
13898 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13899 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13900 (iq2000_modes_tieable_p): New function.
13901 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13902 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13903 (lm32_modes_tieable_p): New function.
13904 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13905 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13906 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13907 (TARGET_MODES_TIEABLE_P): Redefine.
13908 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13909 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13910 (m32r_modes_tieable_p): New function.
13911 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13912 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13913 (m68k_modes_tieable_p): New function.
13914 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13915 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13916 (mcore_modes_tieable_p): New function.
13917 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13918 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13920 (TARGET_MODES_TIEABLE_P): Redefine.
13921 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13922 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13923 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13924 (TARGET_MODES_TIEABLE_P): Redefine.
13925 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13926 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13927 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13928 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13929 (mn10300_modes_tieable_p): ...this and make static.
13930 (TARGET_MODES_TIEABLE_P): Redefine.
13931 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13932 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13933 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13934 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13935 (msp430_modes_tieable_p): Make static.
13936 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13937 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13938 (TARGET_MODES_TIEABLE_P): Redefine.
13939 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13940 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13941 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13942 (TARGET_MODES_TIEABLE_P): Redefine.
13943 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13944 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13945 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13946 (TARGET_MODES_TIEABLE_P): Redefine.
13947 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13948 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13949 (pdp11_modes_tieable_p): New function.
13950 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13951 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13952 (rs6000_modes_tieable_p): New function.
13953 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13954 * config/powerpcspe/powerpcspe.md: Update comment.
13955 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13956 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13957 (TARGET_MODES_TIEABLE_P): Redefine.
13958 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13959 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13960 (rl78_modes_tieable_p): New function.
13961 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13962 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13963 (rs6000_modes_tieable_p): New function.
13964 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13965 * config/rs6000/rs6000.md: Update comment.
13966 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13967 * config/rx/rx.c (rx_modes_tieable_p): New function.
13968 (TARGET_MODES_TIEABLE_P): Redefine.
13969 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13970 * config/s390/s390.c (s390_modes_tieable_p): New function.
13971 (TARGET_MODES_TIEABLE_P): Redefine.
13972 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13973 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13974 (sh_modes_tieable_p): New function.
13975 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13976 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13977 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13978 (sparc_modes_tieable_p): Make static.
13979 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13980 * config/spu/spu.c (spu_modes_tieable_p): New function.
13981 (TARGET_MODES_TIEABLE_P): Redefine.
13982 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13983 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13984 (TARGET_MODES_TIEABLE_P): Redefine.
13985 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13986 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13987 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13988 * config/v850/v850.c (v850_modes_tieable_p): New function.
13989 (TARGET_MODES_TIEABLE_P): Redefine.
13990 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13991 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13992 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13993 (visium_modes_tieable_p): New function.
13994 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13995 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13996 (xtensa_modes_tieable_p): New function.
13997 * system.h (MODES_TIEABLE_P): Poison.
13999 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14000 Alan Hayward <alan.hayward@arm.com>
14001 David Sherwood <david.sherwood@arm.com>
14003 * target.def (hard_regno_mode_ok): New hook.
14004 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14005 (TARGET_HARD_REGNO_MODE_OK): ...this.
14006 * doc/tm.texi.in: Regenerate.
14007 * hooks.h (hook_bool_uint_mode_true): Declare.
14008 * hooks.c (hook_bool_uint_mode_true): New function.
14009 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14010 HARD_REGNO_MODE_OK.
14011 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14012 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14013 instead of HARD_REGNO_MODE_OK.
14014 * caller-save.c: Include target.h.
14015 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14016 HARD_REGNO_MODE_OK.
14017 * combine.c (can_combine_p): Likewise.
14018 (combinable_i3pat): Likewise.
14019 (can_change_dest_mode): Likewise.
14020 * expr.c (init_expr_target): Likewise.
14021 (convert_move): Likewise.
14022 (convert_modes): Likewise.
14023 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14024 (setup_prohibited_mode_move_regs): Likewise.
14025 * ira.h (target_ira): Likewise.
14026 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14027 * lra-constraints.c (process_alt_operands): Likewise.
14028 (split_reg): Likewise.
14029 * recog.c (peep2_find_free_register): Likewise.
14030 * ree.c (combine_reaching_defs): Likewise.
14031 * regcprop.c (maybe_mode_change): Likewise.
14032 * reginfo.c (init_reg_sets_1): Likewise.
14033 (choose_hard_reg_mode): Likewise.
14034 (simplifiable_subregs): Likewise.
14035 * regrename.c (check_new_reg_p): Likewise.
14036 * reload.c (find_valid_class): Likewise.
14037 (find_valid_class_1): Likewise.
14038 (reload_inner_reg_of_subreg): Likewise.
14039 (push_reload): Likewise.
14040 (combine_reloads): Likewise.
14041 (find_dummy_reload): Likewise.
14042 (find_reloads): Likewise.
14043 * reload1.c (find_reg): Likewise.
14044 (set_reload_reg): Likewise.
14045 (allocate_reload_reg): Likewise.
14046 (choose_reload_regs): Likewise.
14047 (reload_adjust_reg_for_temp): Likewise.
14048 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14049 (simplify_subreg_regno): Likewise.
14050 * sel-sched.c (init_regs_for_mode): Likewise.
14051 * varasm.c (make_decl_rtl): Likewise.
14052 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14053 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14054 HARD_REGNO_MODE_OK.
14055 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14056 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14057 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14058 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14059 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14060 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14061 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14062 (arc_mode_class): Delete.
14063 (HARD_REGNO_MODE_OK): Delete.
14064 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14065 (arc_hard_regno_mode_ok): Rename old array to...
14066 (arc_hard_regno_mode_ok_modes): ...this.
14067 (arc_conditional_register_usage): Update accordingly.
14068 (arc_mode_class): Make static.
14069 (arc_hard_regno_mode_ok): New function.
14070 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14071 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14072 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14073 (arm_hard_regno_mode_ok): Make static.
14074 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14075 HARD_REGNO_MODE_OK.
14076 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14077 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14078 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14080 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14081 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14082 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14083 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14084 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14085 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14086 * config/bfin/predicates.md (valid_reg_operand): Use
14087 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14088 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14089 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14090 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14091 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14092 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14093 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14094 (cr16_hard_regno_mode_ok): Make static and return a bool.
14095 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14096 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14097 (cris_hard_regno_mode_ok): New function.
14098 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14099 (epiphany_mode_class): Delete.
14100 (HARD_REGNO_MODE_OK): Delete.
14101 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14102 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14103 (hard_regno_mode_ok): Rename to...
14104 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14105 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14106 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14107 HARD_REGNO_MODE_OK.
14108 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14109 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14110 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14111 (frv_hard_regno_mode_ok): Make static and return a bool.
14112 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14113 HARD_REGNO_MODE_OK.
14114 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14115 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14116 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14117 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14119 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14120 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14121 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14122 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14124 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14125 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14126 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14127 (ia64_hard_regno_mode_ok): New function.
14128 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14129 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14130 (iq2000_hard_regno_mode_ok): New function.
14131 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14132 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14133 (lm32_hard_regno_mode_ok): New function.
14134 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14135 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14136 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14137 instead of HARD_REGNO_MODE_OK.
14138 (m32c_hard_regno_ok): Rename to...
14139 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14140 (m32c_cannot_change_mode_class): Update accordingly.
14141 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14142 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14143 (m32r_mode_class): Delete.
14144 (HARD_REGNO_MODE_OK): Delete.
14145 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14146 (m32r_hard_regno_mode_ok): Rename to...
14147 (m32r_hard_regno_modes): ...this.
14148 (m32r_mode_class): Make static.
14149 (m32r_hard_regno_mode_ok): New function.
14150 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14151 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14152 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14153 (m68k_hard_regno_mode_ok): Make static.
14154 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14155 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14156 (mcore_hard_regno_mode_ok): New function.
14157 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14158 (HARD_REGNO_MODE_OK): Delete.
14159 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14161 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14162 (microblaze_hard_regno_mode_ok): New function.
14163 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14164 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14165 (mips_hard_regno_mode_ok): Delete.
14166 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14167 (mips_hard_regno_mode_ok_p): ...this and make static.
14168 (mips_hard_regno_mode_ok_p): Rename to...
14169 (mips_hard_regno_mode_ok_uncached): ...this.
14170 (mips_hard_regno_mode_ok): New function.
14171 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14172 of HARD_REGNO_MODE_OK.
14173 (mips_option_override): Update after above name changes.
14174 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14175 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14176 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14177 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14178 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14179 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14180 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14181 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14182 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14183 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14184 (msp430_hard_regno_mode_ok): Make static and return a bool.
14185 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14186 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14187 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14190 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14191 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14192 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14193 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14194 (PA_HARD_REGNO_MODE_OK): ...this
14195 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14196 (PA_HARD_REGNO_MODE_OK): ...this.
14197 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14198 (pa_hard_regno_mode_ok): New function.
14199 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14200 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14201 (pdp11_hard_regno_mode_ok): New function.
14202 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14203 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14205 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14207 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14208 (rs6000_hard_regno_mode_ok): Rename to...
14209 (rs6000_hard_regno_mode_ok_uncached): ...this.
14210 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14211 (rs6000_hard_regno_mode_ok): New function.
14212 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14213 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14214 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14215 (riscv_hard_regno_mode_ok): ...this and make static.
14216 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14217 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14218 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14219 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14220 (rl78_hard_regno_mode_ok): Make static and return bool.
14221 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14222 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14224 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14225 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14226 (rs6000_hard_regno_mode_ok): Rename to...
14227 (rs6000_hard_regno_mode_ok_uncached): ...this.
14228 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14229 (rs6000_hard_regno_mode_ok): New function.
14230 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14231 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14232 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14233 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14234 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14235 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14236 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14237 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14238 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14239 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14240 (sh_hard_regno_mode_ok): Make static.
14241 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14242 instead of HARD_REGNO_MODE_OK.
14243 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14244 (sparc_mode_class): Delete.
14245 (HARD_REGNO_MODE_OK): Delete.
14246 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14247 (hard_regno_mode_classes): Make static.
14248 (sparc_mode_class): Likewise.
14249 (sparc_hard_regno_mode_ok): New function.
14250 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14251 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14252 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14254 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14255 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14256 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14257 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14258 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14259 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14260 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14261 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14262 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14263 (visium_hard_regno_mode_ok): New function.
14264 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14265 instead of HARD_REGNO_MODE_OK.
14266 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14267 (HARD_REGNO_MODE_OK): Delete.
14268 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14269 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14270 (xtensa_option_override): Update accordingly.
14271 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14272 (xtensa_hard_regno_mode_ok): New function.
14273 * system.h (HARD_REGNO_MODE_OK): Poison.
14275 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14276 Alan Hayward <alan.hayward@arm.com>
14277 David Sherwood <david.sherwood@arm.com>
14279 * target.def (hard_regno_call_part_clobbered): New hook.
14280 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14281 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14282 * doc/tm.texi: Regenerate.
14283 * hooks.h (hook_bool_uint_mode_false): Declare.
14284 * hooks.c (hook_bool_uint_mode_false): New function.
14285 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14286 * cselib.c (cselib_process_insn): Use
14287 targetm.hard_regno_call_part_clobbered instead of
14288 HARD_REGNO_CALL_PART_CLOBBERED.
14289 * ira-conflicts.c (ira_build_conflicts): Likewise.
14290 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14291 * lra-constraints.c (need_for_call_save_p): Likewise.
14292 * lra-lives.c: Include target.h.
14293 (check_pseudos_live_through_calls): Use
14294 targetm.hard_regno_call_part_clobbered instead of
14295 HARD_REGNO_CALL_PART_CLOBBERED.
14296 * regcprop.c: Include target.h.
14297 (copyprop_hardreg_forward_1): Use
14298 targetm.hard_regno_call_part_clobbered instead of
14299 HARD_REGNO_CALL_PART_CLOBBERED.
14300 * reginfo.c (choose_hard_reg_mode): Likewise.
14301 * regrename.c (check_new_reg_p): Likewise.
14302 * reload.c (find_equiv_reg): Likewise.
14303 * reload1.c (emit_reload_insns): Likewise.
14304 * sched-deps.c (deps_analyze_insn): Likewise.
14305 * sel-sched.c (init_regs_for_mode): Likewise.
14306 (mark_unavailable_hard_regs): Likewise.
14307 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14308 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14309 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14311 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14312 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14313 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14315 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14317 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14318 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14319 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14321 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14322 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14323 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14325 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14326 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14328 * config/powerpcspe/powerpcspe.c
14329 (rs6000_hard_regno_call_part_clobbered): New function.
14330 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14331 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14332 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14334 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14335 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14336 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14338 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14339 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14340 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14342 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14343 Alan Hayward <alan.hayward@arm.com>
14344 David Sherwood <david.sherwood@arm.com>
14346 * rtl.h (subreg_memory_offset): Declare.
14347 * emit-rtl.c (subreg_memory_offset): New function.
14348 * expmed.c (store_bit_field_1): Use it.
14349 * expr.c (undefined_operand_subword_p): Likewise.
14350 * simplify-rtx.c (simplify_subreg): Likewise.
14352 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14354 PR rtl-optimization/57448
14357 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14358 using atomic_load pattern.
14359 (expand_atomic_store): Likewise.
14361 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14364 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14365 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14366 handling. Use replace_call_with_value with NULL instead of
14367 gsi_replace, unlink_stmt_vdef and release_defs.
14369 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14372 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14374 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14377 * rtl.h (emit_library_call_value_1): Declare.
14378 (emit_library_call): Replace declaration with a series of overloads.
14379 Remove the parameter count argument.
14380 (emit_library_call_value): Likewise.
14381 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14382 with an "rtx_mode_t *".
14383 (emit_library_call_value): Delete.
14384 (emit_library_call): Likewise.
14385 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14386 (asan_emit_allocas_unpoison): Likewise.
14387 * builtins.c (expand_builtin_powi): Likewise.
14388 (expand_asan_emit_allocas_unpoison): Likewise.
14389 * cfgexpand.c (expand_main_function): Likewise.
14390 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14391 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14392 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14393 * config/arm/arm.c (arm_trampoline_init): Likewise.
14394 (arm_call_tls_get_addr): Likewise.
14395 (arm_expand_divmod_libfunc): Likewise.
14396 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14397 (smulsi3_highpart): Likewise.
14398 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14399 (c6x_expand_compare): Likewise.
14400 (c6x_expand_movmem): Likewise.
14401 * config/frv/frv.c (frv_trampoline_init): Likewise.
14402 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14403 (ix86_expand_divmod_libfunc): Likewise.
14404 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14405 (ia64_expand_compare): Likewise.
14406 (ia64_profile_hook): Likewise.
14407 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14408 (nonlocal_goto): Likewise.
14409 (restore_stack_nonlocal): Likewise.
14410 * config/m32r/m32r.c (block_move_call): Likewise.
14411 (m32r_trampoline_init): Likewise.
14412 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14413 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14414 (m68k_call_m68k_read_tp): Likewise.
14415 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14416 (microblaze_expand_divide): Likewise.
14417 * config/mips/mips.h (mips_args): Likewise.
14418 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14419 (MIPS_ICACHE_SYNC): Likewise.
14420 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14421 (nios2_trampoline_init): Likewise.
14422 * config/pa/pa.c (hppa_tls_call): Likewise.
14423 (pa_trampoline_init): Likewise.
14424 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14425 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14426 (expand_strn_compare): Likewise.
14427 (rs6000_generate_compare): Likewise.
14428 (rs6000_expand_float128_convert): Likewise.
14429 (output_profile_hook): Likewise.
14430 (rs6000_trampoline_init): Likewise.
14431 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14432 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14433 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14434 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14435 (rs6000_generate_compare): Likewise.
14436 (rs6000_expand_float128_convert): Likewise.
14437 (output_profile_hook): Likewise.
14438 (rs6000_trampoline_init): Likewise.
14439 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14440 * config/sh/sh.c (sh_trampoline_init): Likewise.
14441 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14442 (sparc_emit_float_lib_cmp): Likewise.
14443 (sparc32_initialize_trampoline): Likewise.
14444 (sparc64_initialize_trampoline): Likewise.
14445 (sparc_profile_hook): Likewise.
14446 * config/spu/spu.c (ea_load_store): Likewise.
14447 * config/spu/spu.md (floatunssidf2): Likewise.
14448 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14449 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14450 * config/visium/visium.c (expand_block_move_4): Likewise.
14451 (expand_block_move_2): Likewise.
14452 (expand_block_move_1): Likewise.
14453 (expand_block_set_4): Likewise.
14454 (expand_block_set_2): Likewise.
14455 (expand_block_set_1): Likewise.
14456 (visium_trampoline_init): Likewise.
14457 (visium_profile_hook): Likewise.
14458 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14459 (xtensa_setup_frame_addresses): Likewise.
14460 (xtensa_trampoline_init): Likewise.
14461 * except.c (sjlj_emit_function_enter): Likewise.
14462 (sjlj_emit_function_exit): Likewise.
14463 * explow.c (allocate_dynamic_stack_space): Likewise.
14464 (probe_stack_range): Likewise.
14465 * expr.c (convert_mode_scalar): Likewise.
14466 * optabs.c (expand_binop): Likewise.
14467 (expand_twoval_binop_libfunc): Likewise.
14468 (expand_unop): Likewise.
14469 (prepare_cmp_insn): Likewise.
14470 (prepare_float_lib_cmp): Likewise.
14471 (expand_float): Likewise.
14472 (expand_fix): Likewise.
14473 (expand_fixed_convert): Likewise.
14474 (maybe_emit_sync_lock_test_and_set): Likewise.
14475 (expand_atomic_compare_and_swap): Likewise.
14476 (expand_mem_thread_fence): Likewise.
14477 (expand_atomic_fetch_op): Likewise.
14479 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14481 * doc/generic.texi (OpenACC): Adjust URL.
14482 * doc/invoke.texi (C Dialect Options): Ditto.
14484 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14486 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14487 predicate for operand 1. Add (m,<S>) constraint.
14488 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14489 Prevent memory operand 1 with register operand 2.
14491 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14493 PR rtl-optimization/82024
14494 * combine.c (try_combine): If the combination result is a PARALLEL,
14495 and we only need to retain the SET in there that would be placed
14496 at I2, check that we can place that at I3 instead, before doing so.
14498 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14501 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14503 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14504 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14507 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14508 Jeff Law <law@redhat.com>
14510 * varasm.c (bss_initializer_p): Do not put constants into .bss
14511 (categorize_decl_for_section): Handle bss_initializer_p returning
14512 false when DECL_INITIAL is NULL.
14514 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14517 * config/s390/s390.c (s390_can_inline_p): New function.
14519 2017-09-01 Jeff Law <law@redhat.com>
14521 PR tree-optimization/82052
14522 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14523 Always initialize the returned slot after a hash table miss
14524 when INSERT is true.
14526 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14528 * config/s390/s390.md (mem_signal_fence): Remove.
14529 * doc/md.texi (mem_signal_fence): Remove.
14530 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14532 * target-insns.def (mem_signal_fence): Remove.
14534 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14537 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14540 * asan.c (create_odr_indicator): Strip name encoding from assembler
14541 name before appending it after __odr_asan_.
14543 2017-09-01 Martin Liska <mliska@suse.cz>
14545 PR tree-optimization/82059
14546 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14547 frequency only when an edge is redirected.
14549 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14551 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14552 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14553 (arc_conditional_register_usage): Remove ARC600 lp_count
14555 (arc_file_start): Emit Tag_ARC_CPU_variation.
14556 (arc_can_use_doloop_p): New conditions to use ZOLs.
14557 (hwloop_fail): New function.
14558 (hwloop_optimize): Likewise.
14559 (hwloop_pattern_reg): Likewise.
14560 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14561 (arc_reorg_loops): New function, calls reorg_loops.
14562 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14563 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14565 (arc_loop_hazard): Remove function, functionality moved into
14567 (arc_hazard): Remove arc_loop_hazard call.
14568 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14569 into hwloop_optimize.
14570 (arc_label_align): Remove ZOL handling.
14571 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14572 * config/arc/arc.md (doloop_begin): Remove pattern.
14573 (doloop_begin_i): Likewise.
14574 (doloop_end_i): Likewise.
14575 (doloop_fallback): Likewise.
14576 (doloop_fallback_m): Likewise.
14577 (doloop_end): Reimplement expand.
14578 (arc_lp): New pattern for LP instruction.
14579 (loop_end): New pattern.
14580 (loop_fail): Likewise.
14581 (decrement_and_branch_until_zero): Likewise.
14582 * config/arc/arc.opt (mlpc-width): New option.
14583 * doc/invoke.texi (mlpc-width): Document option.
14585 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14587 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14588 (arc_ccfsm_advance): Fix checking for delay slots.
14589 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14591 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14593 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14595 (movhi_insn): Update store instruction constraint which are saving
14596 6-bit short immediates.
14597 (movsi_insn): Consider also short scaled load operations.
14598 (zero_extendhisi2_i): Use Usd constraint instead of T.
14599 (extendhisi2_i): Add q constraint.
14600 (arc_clzsi2): Add type and length attributes.
14601 (arc_ctzsi2): Likewise.
14602 * config/arc/constraints.md (Usc): Update constraint, the
14603 assembler can parse two relocations for a single instruction.
14605 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14607 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14608 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14610 2017-08-31 Olivier Hainque <hainque@adacore.com>
14612 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14613 match as powerpc-wrs-vxworks*.
14615 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14617 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14618 register constraint for by-element operand.
14619 (aarch64_mls_elt_merge<mode>): Likewise.
14621 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14623 * config/arc/arc.c (arc_can_follow_jump): Check for short
14626 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14628 * config.gcc: Use g.opt for arc.
14629 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14630 functionality moved to ...
14631 (legitimate_scaled_address_p): New function, ...here.
14632 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14633 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14634 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14636 (arc_override_options): Handle G option.
14637 (arc_output_pic_addr_const): Correct function definition.
14638 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14639 (arc_decl_anon_ns_mem_p): Delete.
14640 (arc_in_small_data_p): Overhaul this function to take into
14641 consideration the value given via G option.
14642 (arc_rewrite_small_data_1): Renamed and corrected old
14643 arc_rewrite_small_data function.
14644 (arc_rewrite_small_data): New function.
14645 (small_data_pattern): Don't use pic_offset_table_rtx.
14646 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14647 * config/arc/simdext.md (movmisalignv2hi): Use
14648 prepare_move_operands function.
14650 (movmisalign*): Likewise.
14651 * doc/invoke.texi (ARC options): Document -G option.
14653 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14655 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14657 * config/arc/arc.c (arc_print_operand): Output scalled address for
14658 sdata whenever is possible.
14659 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14660 load/stores are available.
14661 (compact_sda_memory_operand): Check for the alignment required by
14662 code density instructions.
14663 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14665 * config/arc/constraints.md (Usd): Update constraint.
14666 (Us0): New constraint.
14667 (Usc): Update constraint.
14669 2017-08-31 Richard Biener <rguenther@suse.de>
14671 PR middle-end/82054
14672 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14673 function only once.
14675 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14677 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14678 Resize type_signature.
14680 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14681 Alan Hayward <alan.hayward@arm.com>
14682 David Sherwood <david.sherwood@arm.com>
14684 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14685 subregs whose inner modes can be stored in GPRs.
14686 (aarch64_classify_index): Likewise.
14688 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14689 Alan Hayward <alan.hayward@arm.com>
14690 David Sherwood <david.sherwood@arm.com>
14692 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14693 (V_INT_EQUIV): ...this.
14694 (v_cmp_result): Rename to...
14695 (v_int_equiv): ...this.
14696 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14697 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14698 (copysign<mode>3): Likewise.
14699 (aarch64_simd_bsl<mode>_internal): Likewise.
14700 (aarch64_simd_bsl<mode>): Likewise.
14701 (vec_cmp<mode><mode>): Likewise.
14702 (vcond<mode><mode>): Likewise.
14703 (vcond<v_cmp_mixed><mode>): Likewise.
14704 (vcondu<mode><v_cmp_mixed>): Likewise.
14705 (aarch64_cm<optab><mode>): Likewise.
14706 (aarch64_cmtst<mode>): Likewise.
14707 (aarch64_fac<optab><mode>): Likewise.
14708 (vec_perm_const<mode>): Likewise.
14709 (vcond_mask_<mode><v_cmp_result>): Rename to...
14710 (vcond_mask_<mode><v_int_equiv>): ...this.
14711 (vec_cmp<mode><v_cmp_result>): Rename to...
14712 (vec_cmp<mode><v_int_equiv>): ...this.
14714 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14715 Alan Hayward <alan.hayward@arm.com>
14716 David Sherwood <david.sherwood@arm.com>
14718 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14720 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14721 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14722 (UNSPEC_LD4_DREG): New unspecs.
14723 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14724 (aarch64_ld2<mode>_dreg_be): Replace with...
14725 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14727 (aarch64_ld3<mode>_dreg_le)
14728 (aarch64_ld3<mode>_dreg_be): Replace with...
14729 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14731 (aarch64_ld4<mode>_dreg_le)
14732 (aarch64_ld4<mode>_dreg_be): Replace with...
14733 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14736 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14738 PR tree-optimization/81987
14739 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14740 insert an initializer in a location not dominated by the stride
14743 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14745 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14746 on the entire header of the finally block in the fallthru case.
14748 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14750 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14752 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14754 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14755 rs6000_emit_move_from_cr and call renamed function.
14756 (rs6000_emit_prologue): Call renamed functions.
14757 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14758 movesi_from_cr, remove volatile CRs.
14760 2017-08-30 Jon Beniston <jon@beniston.com>
14761 Richard Biener <rguenther@suse.de>
14763 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14764 of VECTOR_MODE_P check.
14765 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14766 element vector types.
14768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14770 * df.h (df_read_modify_subreg_p): Remove in favor of...
14771 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14772 const_rtx instead of an rtx.
14773 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14774 * df-problems.c (df_word_lr_mark_ref): Likewise.
14775 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14776 (mark_pseudo_reg_dead): Likewise.
14777 (mark_ref_dead): Likewise.
14778 * reginfo.c (init_subregs_of_mode): Likewise.
14779 * sched-deps.c (sched_analyze_1): Likewise.
14780 * df-scan.c (df_def_record_1): Likewise.
14781 (df_uses_record): Likewise.
14782 (df_read_modify_subreg_p): Remove in favor of...
14783 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14784 const_rtx instead of an rtx.
14786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14787 Alan Hayward <alan.hayward@arm.com>
14788 David Sherwood <david.sherwood@arm.com>
14790 * rtl.h (partial_subreg_p): New function.
14791 * caller-save.c (save_call_clobbered_regs): Use it.
14792 * calls.c (expand_call): Likewise.
14793 * combine.c (combinable_i3pat): Likewise.
14794 (simplify_set): Likewise.
14795 (make_extraction): Likewise.
14796 (make_compound_operation_int): Likewise.
14797 (gen_lowpart_or_truncate): Likewise.
14798 (force_to_mode): Likewise.
14799 (make_field_assignment): Likewise.
14800 (reg_truncated_to_mode): Likewise.
14801 (record_truncated_value): Likewise.
14802 (move_deaths): Likewise.
14803 * cse.c (record_jump_cond): Likewise.
14804 (cse_insn): Likewise.
14805 * cselib.c (cselib_lookup_1): Likewise.
14806 * expmed.c (extract_bit_field_using_extv): Likewise.
14807 * function.c (assign_parm_setup_reg): Likewise.
14808 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14809 * ira-build.c (create_insn_allocnos): Likewise.
14810 * lra-coalesce.c (merge_pseudos): Likewise.
14811 * lra-constraints.c (match_reload): Likewise.
14812 (simplify_operand_subreg): Likewise.
14813 (curr_insn_transform): Likewise.
14814 * lra-lives.c (process_bb_lives): Likewise.
14815 * lra.c (new_insn_reg): Likewise.
14816 (lra_substitute_pseudo): Likewise.
14817 * regcprop.c (mode_change_ok): Likewise.
14818 (maybe_mode_change): Likewise.
14819 (copyprop_hardreg_forward_1): Likewise.
14820 * reload.c (push_reload): Likewise.
14821 (find_reloads): Likewise.
14822 (find_reloads_subreg_address): Likewise.
14823 * reload1.c (alter_reg): Likewise.
14824 (eliminate_regs_1): Likewise.
14825 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14827 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14829 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14830 back to if statements, including unpack.
14832 2017-08-30 Martin Liska <mliska@suse.cz>
14834 PR inline-asm/82001
14835 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14837 (func_checker::compare_asm_inputs_outputs): ... this function.
14838 (func_checker::compare_gimple_asm): Use the function to compare
14839 also ASM constrains.
14840 * ipa-icf-gimple.h: Rename the function.
14842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14843 Alan Hayward <alan.hayward@arm.com>
14844 David Sherwood <david.sherwood@arm.com>
14846 * coretypes.h (complex_mode): New type.
14847 * gdbhooks.py (build_pretty_printer): Handle it.
14848 * machmode.h (complex_mode): New class.
14849 (complex_mode::includes_p): New function.
14850 (is_complex_int_mode): Likewise.
14851 (is_complex_float_mode): Likewise.
14852 * genmodes.c (get_mode_class): Handle complex mode classes.
14853 * function.c (expand_function_end): Use is_complex_int_mode.
14855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14856 Alan Hayward <alan.hayward@arm.com>
14857 David Sherwood <david.sherwood@arm.com>
14859 * coretypes.h (scalar_mode_pod): New typedef.
14860 * gdbhooks.py (build_pretty_printer): Handle it.
14861 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14862 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14863 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14864 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14865 as_a <scalar_mode>.
14867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14868 Alan Hayward <alan.hayward@arm.com>
14869 David Sherwood <david.sherwood@arm.com>
14871 * machmode.h (mode_for_vector): Take a scalar_mode instead
14873 * stor-layout.c (mode_for_vector): Likewise.
14874 * explow.c (promote_mode): Use as_a <scalar_mode>.
14875 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14878 Alan Hayward <alan.hayward@arm.com>
14879 David Sherwood <david.sherwood@arm.com>
14881 * target.def (preferred_simd_mode): Take a scalar_mode
14882 instead of a machine_mode.
14883 * targhooks.h (default_preferred_simd_mode): Likewise.
14884 * targhooks.c (default_preferred_simd_mode): Likewise.
14885 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14886 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14887 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14888 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14889 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14890 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14891 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14892 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14894 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14895 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14896 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14897 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14898 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14899 * doc/tm.texi: Regenerate.
14900 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14904 Alan Hayward <alan.hayward@arm.com>
14905 David Sherwood <david.sherwood@arm.com>
14907 * target.def (scalar_mode_supported_p): Take a scalar_mode
14908 instead of a machine_mode.
14909 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14910 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14911 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14912 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14913 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14914 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14915 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14916 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14917 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14918 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14919 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14920 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14921 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14922 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14924 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14925 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14926 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14927 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14928 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14930 * doc/tm.texi: Regenerate.
14932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14933 Alan Hayward <alan.hayward@arm.com>
14934 David Sherwood <david.sherwood@arm.com>
14936 * coretypes.h (opt_scalar_mode): New typedef.
14937 * gdbhooks.py (build_pretty_printers): Handle it.
14938 * machmode.h (mode_iterator::get_2xwider): Add overload for
14940 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14942 * expr.c (convert_mode_scalar): Likewise.
14943 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14944 * optabs.c (expand_float): Likewise.
14945 (expand_fix): Likewise.
14946 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14949 Alan Hayward <alan.hayward@arm.com>
14950 David Sherwood <david.sherwood@arm.com>
14952 * optabs.c (expand_float): Explicitly check for scalars before
14953 using a branching expansion.
14954 (expand_fix): Likewise.
14956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14957 Alan Hayward <alan.hayward@arm.com>
14958 David Sherwood <david.sherwood@arm.com>
14960 * expr.c (convert_mode): Split scalar handling out into...
14961 (convert_mode_scalar): ...this new function. Treat the modes
14964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14965 Alan Hayward <alan.hayward@arm.com>
14966 David Sherwood <david.sherwood@arm.com>
14968 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14970 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14973 Alan Hayward <alan.hayward@arm.com>
14974 David Sherwood <david.sherwood@arm.com>
14976 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14977 rather than a machine_mode.
14978 (fixed_from_string): Likewise.
14979 (fixed_convert): Likewise.
14980 (fixed_convert_from_int): Likewise.
14981 (fixed_convert_from_real): Likewise.
14982 (real_convert_from_fixed): Likewise.
14983 * fixed-value.c (fixed_from_double_int): Likewise.
14984 (fixed_from_string): Likewise.
14985 (fixed_convert): Likewise.
14986 (fixed_convert_from_int): Likewise.
14987 (fixed_convert_from_real): Likewise.
14988 (real_convert_from_fixed): Likewise.
14989 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14992 Alan Hayward <alan.hayward@arm.com>
14993 David Sherwood <david.sherwood@arm.com>
14995 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14996 of separate mode class checks. Do not allow vector modes here.
14997 (immed_wide_int_const): Use as_a <scalar_mode>.
14998 * explow.c (trunc_int_for_mode): Likewise.
14999 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15000 (wi::shwi): Likewise.
15001 (wi::min_value): Likewise.
15002 (wi::max_value): Likewise.
15003 * dwarf2out.c (loc_descriptor): Likewise.
15004 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15005 for CONST_WIDE_INT.
15007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15008 Alan Hayward <alan.hayward@arm.com>
15009 David Sherwood <david.sherwood@arm.com>
15011 * tree.h (SCALAR_TYPE_MODE): New macro.
15012 * expr.c (expand_expr_addr_expr_1): Use it.
15013 (expand_expr_real_2): Likewise.
15014 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15015 (fold_convert_const_fixed_from_int): Likewise.
15016 (fold_convert_const_fixed_from_real): Likewise.
15017 (native_encode_fixed): Likewise
15018 (native_encode_complex): Likewise
15019 (native_encode_vector): Likewise.
15020 (native_interpret_fixed): Likewise.
15021 (native_interpret_real): Likewise.
15022 (native_interpret_complex): Likewise.
15023 (native_interpret_vector): Likewise.
15024 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15025 (simd_clone_adjust_argument_types): Likewise.
15026 (simd_clone_init_simd_arrays): Likewise.
15027 (simd_clone_adjust): Likewise.
15028 * stor-layout.c (layout_type): Likewise.
15029 * tree.c (build_minus_one_cst): Likewise.
15030 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15031 * tree-inline.c (estimate_move_cost): Likewise.
15032 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15033 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15034 (vectorizable_reduction): Likewise.
15035 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15036 (vect_recog_mixed_size_cond_pattern): Likewise.
15037 (check_bool_pattern): Likewise.
15038 (adjust_bool_pattern): Likewise.
15039 (search_type_for_mask_1): Likewise.
15040 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15041 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15042 (vectorizable_load): Likewise.
15043 (vectorizable_store): Likewise.
15044 * ubsan.c (ubsan_encode_value): Likewise.
15045 * varasm.c (output_constant): Likewise.
15047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15048 Alan Hayward <alan.hayward@arm.com>
15049 David Sherwood <david.sherwood@arm.com>
15051 * coretypes.h (scalar_mode): New class.
15052 * machmode.h (scalar_mode): Likewise.
15053 (scalar_mode::includes_p): New function.
15054 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15055 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15056 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15057 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15058 * expmed.c (store_bit_field_1): Likewise.
15059 (extract_bit_field_1): Likewise.
15060 * expr.c (write_complex_part): Likewise.
15061 (read_complex_part): Likewise.
15062 (emit_move_complex_push): Likewise.
15063 (expand_expr_real_2): Likewise.
15064 * function.c (assign_parm_setup_reg): Likewise.
15065 (assign_parms_unsplit_complex): Likewise.
15066 * optabs.c (expand_binop): Likewise.
15067 * rtlanal.c (subreg_get_info): Likewise.
15068 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15069 * varasm.c (output_constant_pool_2): Likewise.
15071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15072 Alan Hayward <alan.hayward@arm.com>
15073 David Sherwood <david.sherwood@arm.com>
15075 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15077 (expmed_mult_highpart_optab): Likewise.
15078 (expmed_mult_highpart): Likewise.
15080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15081 Alan Hayward <alan.hayward@arm.com>
15082 David Sherwood <david.sherwood@arm.com>
15084 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15085 instead of a machine_mode.
15086 (builtin_memset_read_str): Likewise.
15087 * builtins.c (c_readstr): Likewise.
15088 (builtin_memcpy_read_str): Likewise.
15089 (builtin_strncpy_read_str): Likewise.
15090 (builtin_memset_read_str): Likewise.
15091 (builtin_memset_gen_str): Likewise.
15092 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15093 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15094 instead of a machine_mode.
15095 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15097 (make_extraction): Likewise.
15098 (try_widen_shift_mode): Take and return scalar_int_modes instead
15100 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15101 a scalar_int_mode instead of a machine_mode.
15102 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15103 (avr_addr_space_pointer_mode): Likewise.
15104 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15105 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15106 (msp430_unwind_word_mode): Likewise.
15107 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15108 (spu_addr_space_pointer_mode): Likewise.
15109 (spu_addr_space_address_mode): Likewise.
15110 (spu_libgcc_cmp_return_mode): Likewise.
15111 (spu_libgcc_shift_count_mode): Likewise.
15112 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15113 (rl78_addr_space_pointer_mode): Likewise.
15114 (fl78_unwind_word_mode): Likewise.
15115 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15117 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15118 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15119 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15120 (mips_valid_pointer_mode): Likewise.
15121 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15122 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15123 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15125 (ft32_addr_space_address_mode): Likewise.
15126 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15127 scalar_int_mode instead of a machine_mode.
15128 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15130 (m32c_addr_space_address_mode): Likewise.
15131 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15132 (rs6000_eh_return_filter_mode): Likewise.
15133 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15134 (rs6000_eh_return_filter_mode): Likewise.
15135 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15136 (s390_libgcc_shift_count_mode): Likewise.
15137 (s390_unwind_word_mode): Likewise.
15138 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15140 * target.def (mode_rep_extended): Likewise.
15141 (valid_pointer_mode): Likewise.
15142 (addr_space.valid_pointer_mode): Likewise.
15143 (eh_return_filter_mode): Return a scalar_int_mode rather than
15145 (libgcc_cmp_return_mode): Likewise.
15146 (libgcc_shift_count_mode): Likewise.
15147 (unwind_word_mode): Likewise.
15148 (addr_space.pointer_mode): Likewise.
15149 (addr_space.address_mode): Likewise.
15150 * doc/tm.texi: Regenerate.
15151 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15153 (do_jump): Use scalar_int_mode for local variables.
15154 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15155 rather than a machine_mode.
15156 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15157 (scompare_loc_descriptor_wide): Likewise.
15158 (scompare_loc_descriptor_narrow): Likewise.
15159 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15161 * except.c (sjlj_emit_dispatch_table): Likewise.
15162 (expand_builtin_eh_copy_values): Likewise.
15163 * explow.c (convert_memory_address_addr_space_1): Likewise.
15164 Take a scalar_int_mode rather than a machine_mode.
15165 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15166 than a machine_mode.
15167 (memory_address_addr_space): Use scalar_int_mode for local variables.
15168 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15169 rather than a machine_mode.
15170 * expmed.c (mask_rtx): Likewise.
15171 (init_expmed_one_conv): Likewise.
15172 (expand_mult_highpart_adjust): Likewise.
15173 (extract_high_half): Likewise.
15174 (expmed_mult_highpart_optab): Likewise.
15175 (expmed_mult_highpart): Likewise.
15176 (expand_smod_pow2): Likewise.
15177 (expand_sdiv_pow2): Likewise.
15178 (emit_store_flag_int): Likewise.
15179 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15181 (extract_low_bits): Likewise.
15182 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15184 * expr.c (pieces_addr::adjust): Likewise.
15185 (can_store_by_pieces): Likewise.
15186 (store_by_pieces): Likewise.
15187 (clear_by_pieces_1): Likewise.
15188 (expand_expr_addr_expr_1): Likewise.
15189 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15190 (expand_expr_real_1): Likewise.
15191 (try_casesi): Likewise.
15192 * final.c (shorten_branches): Likewise.
15193 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15194 type of "mode" to machine_mode.
15195 * internal-fn.c (expand_arith_overflow_result_store): Take a
15196 scalar_int_mode rather than a machine_mode.
15197 (expand_mul_overflow): Use scalar_int_mode for local variables.
15198 * loop-doloop.c (doloop_modify): Likewise.
15199 (doloop_optimize): Likewise.
15200 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15201 than a machine_mode.
15202 (expand_doubleword_shift_condmove): Likewise.
15203 (expand_doubleword_shift): Likewise.
15204 (expand_doubleword_clz): Likewise.
15205 (expand_doubleword_popcount): Likewise.
15206 (expand_doubleword_parity): Likewise.
15207 (expand_absneg_bit): Use scalar_int_mode for local variables.
15208 (prepare_float_lib_cmp): Likewise.
15209 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15210 rather than a machine_mode.
15211 (convert_memory_address_addr_space): Likewise.
15212 (get_mode_bounds): Likewise.
15213 (get_address_mode): Return a scalar_int_mode rather than a
15215 * rtlanal.c (get_address_mode): Likewise.
15216 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15217 than a machine_mode.
15218 * targhooks.c (default_mode_rep_extended): Likewise.
15219 (default_valid_pointer_mode): Likewise.
15220 (default_addr_space_valid_pointer_mode): Likewise.
15221 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15222 than a machine_mode.
15223 (default_libgcc_cmp_return_mode): Likewise.
15224 (default_libgcc_shift_count_mode): Likewise.
15225 (default_unwind_word_mode): Likewise.
15226 (default_addr_space_pointer_mode): Likewise.
15227 (default_addr_space_address_mode): Likewise.
15228 * targhooks.h (default_eh_return_filter_mode): Likewise.
15229 (default_libgcc_cmp_return_mode): Likewise.
15230 (default_libgcc_shift_count_mode): Likewise.
15231 (default_unwind_word_mode): Likewise.
15232 (default_addr_space_pointer_mode): Likewise.
15233 (default_addr_space_address_mode): Likewise.
15234 (default_mode_rep_extended): Take a scalar_int_mode rather than
15236 (default_valid_pointer_mode): Likewise.
15237 (default_addr_space_valid_pointer_mode): Likewise.
15238 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15240 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15241 rather than a machine_mode.
15242 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15243 for local variables.
15244 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15245 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15246 than a machine_mode.
15248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15249 Alan Hayward <alan.hayward@arm.com>
15250 David Sherwood <david.sherwood@arm.com>
15252 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15253 the mode argument to scalar_int_mode.
15254 (do_jump_by_parts_zero_rtx): Likewise.
15255 (do_jump_by_parts_equality_rtx): Likewise.
15256 (do_jump_by_parts_greater): Take a mode argument.
15257 (do_jump_by_parts_equality): Likewise.
15258 (do_jump_1): Update calls accordingly.
15260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15261 Alan Hayward <alan.hayward@arm.com>
15262 David Sherwood <david.sherwood@arm.com>
15264 * is-a.h (safe_dyn_cast): New function.
15265 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15266 (jump_table_for_label): Likewise.
15267 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15268 instead of an rtx_insn *.
15269 (shorten_branches): Use dyn_cast instead of LABEL_P and
15270 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15271 rtx_jump_table_data::get_data_mode.
15272 (final_scan_insn): Likewise.
15274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15275 Alan Hayward <alan.hayward@arm.com>
15276 David Sherwood <david.sherwood@arm.com>
15278 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15279 trying to combine a full-register integer set with a subreg
15282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15283 Alan Hayward <alan.hayward@arm.com>
15284 David Sherwood <david.sherwood@arm.com>
15286 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15287 that is always either address_mode or pointer_mode.
15289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15290 Alan Hayward <alan.hayward@arm.com>
15291 David Sherwood <david.sherwood@arm.com>
15293 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15294 when the two are known to be equal.
15296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15297 Alan Hayward <alan.hayward@arm.com>
15298 David Sherwood <david.sherwood@arm.com>
15300 * simplify-rtx.c (simplify_const_unary_operation): Use
15301 is_a <scalar_int_mode> instead of checking for a nonzero
15302 precision. Forcibly convert op_mode to a scalar_int_mode
15303 in that case. More clearly differentiate the operand and
15304 result modes and use the former when deciding what the value
15305 of a count-bits operation should be. Use is_int_mode instead
15306 of checking for a MODE_INT. Remove redundant check for whether
15307 this mode has a zero precision.
15309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15310 Alan Hayward <alan.hayward@arm.com>
15311 David Sherwood <david.sherwood@arm.com>
15313 * optabs.c (widen_leading): Change the type of the mode argument
15314 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15315 (widen_bswap): Likewise.
15316 (expand_parity): Likewise.
15317 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15318 (expand_ffs): Likewise.
15319 (epand_unop): Check for scalar integer modes before calling the
15322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15323 Alan Hayward <alan.hayward@arm.com>
15324 David Sherwood <david.sherwood@arm.com>
15326 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15328 (expand_expr_real_1): Update call accordingly.
15330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15331 Alan Hayward <alan.hayward@arm.com>
15332 David Sherwood <david.sherwood@arm.com>
15334 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15335 value_mode arguments. Use scalar_int_mode internally.
15336 (store_bit_field_1): Rename the new integer mode from imode
15337 to op0_mode and use it instead of GET_MODE (op0). Update calls
15338 to store_split_bit_field, store_bit_field_using_insv and
15339 store_fixed_bit_field.
15340 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15341 Use scalar_int_mode internally. Use a bit count rather than a mode
15342 when calculating the largest bit size for get_best_mode.
15343 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15344 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15345 Remove assertion that OP0 has a scalar integer mode.
15346 (store_split_bit_field): Add op0_mode and value_mode arguments.
15347 Update calls to extract_fixed_bit_field.
15348 (extract_bit_field_using_extv): Add an op0_mode argument.
15349 Use scalar_int_mode internally.
15350 (extract_bit_field_1): Rename the new integer mode from imode to
15351 op0_mode and use it instead of GET_MODE (op0). Update calls to
15352 extract_split_bit_field, extract_bit_field_using_extv and
15353 extract_fixed_bit_field.
15354 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15355 to extract_split_bit_field and extract_fixed_bit_field_1.
15356 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15357 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15358 on the target mode.
15359 (extract_split_bit_field): Add an op0_mode argument. Update call
15360 to extract_fixed_bit_field.
15362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15363 Alan Hayward <alan.hayward@arm.com>
15364 David Sherwood <david.sherwood@arm.com>
15366 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15367 * explow.c (hard_function_value): Likewise.
15368 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15369 convert_to_mode call outside the loop.
15370 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15371 for the mode iterator. Require the mode specified by max_pieces
15373 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15375 (copy_blkmode_to_reg): Likewise.
15376 (set_storage_via_setmem): Likewise.
15377 * optabs.c (prepare_cmp_insn): Likewise.
15378 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15379 * stor-layout.c (finish_bitfield_representative): Likewise.
15381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15382 Alan Hayward <alan.hayward@arm.com>
15383 David Sherwood <david.sherwood@arm.com>
15385 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15386 * expr.c (convert_move): Use them.
15387 (convert_modes): Likewise.
15388 (store_expr_with_bounds): Likewise.
15390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15391 Alan Hayward <alan.hayward@arm.com>
15392 David Sherwood <david.sherwood@arm.com>
15394 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15395 parameter for the mode of "x". Remove the "known_x", "known_mode"
15396 and "known_ret" arguments. Change the type of the mode argument
15397 to scalar_int_mode.
15398 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15399 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15400 (reg_num_sign_bit_copies_for_combine): Likewise.
15401 * rtlanal.c (nonzero_bits1): Likewise.
15402 (num_sign_bit_copies1): Likewise.
15403 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15404 (reg_num_sign_bit_copies_general): Likewise.
15405 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15406 (reg_nonzero_bits_general): Likewise.
15408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15409 Alan Hayward <alan.hayward@arm.com>
15410 David Sherwood <david.sherwood@arm.com>
15412 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15413 than in subroutines. Return 1 for non-integer modes.
15414 (cached_num_sign_bit_copies): Change the type of the mode parameter
15415 to scalar_int_mode.
15416 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15417 classes. Handle CONST_INT_P first and then check whether X also
15418 has a scalar integer mode. Check the same thing for inner registers
15419 of a SUBREG and for values that are being extended or truncated.
15421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15422 Alan Hayward <alan.hayward@arm.com>
15423 David Sherwood <david.sherwood@arm.com>
15425 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15426 in subroutines. Return the mode mask for non-integer modes.
15427 (cached_nonzero_bits): Change the type of the mode parameter
15428 to scalar_int_mode.
15429 (nonzero_bits1): Likewise. Remove early exit for other mode
15430 classes. Handle CONST_INT_P first and then check whether X
15431 also has a scalar integer mode.
15433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15434 Alan Hayward <alan.hayward@arm.com>
15435 David Sherwood <david.sherwood@arm.com>
15437 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15438 Return a scalar_int_mode and assert that the size is greater than
15440 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15441 (op_by_pieces_d::op_by_pieces_d): Likewise.
15442 (op_by_pieces_d::run): Likewise.
15443 (can_store_by_pieces): Likewise.
15445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15446 Alan Hayward <alan.hayward@arm.com>
15447 David Sherwood <david.sherwood@arm.com>
15449 * combine.c (extract_left_shift): Add a mode argument and update
15451 (make_compound_operation_int): Change the type of the mode parameter
15452 to scalar_int_mode and update the call to extract_left_shift.
15454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15455 Alan Hayward <alan.hayward@arm.com>
15456 David Sherwood <david.sherwood@arm.com>
15458 * combine.c (simplify_and_const_int): Change the type of the mode
15459 parameter to scalar_int_mode.
15460 (simplify_and_const_int_1): Likewise. Update recursive call.
15462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15463 Alan Hayward <alan.hayward@arm.com>
15464 David Sherwood <david.sherwood@arm.com>
15466 * combine.c (simplify_compare_const): Check that the mode is a
15467 scalar_int_mode (rather than VOIDmode) before testing its
15469 (simplify_comparison): Move COMPARISON_P handling out of the
15470 loop and restrict the latter part of the loop to scalar_int_modes.
15471 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15472 and when considering SUBREG_REGs. Use is_int_mode instead of
15473 checking GET_MODE_CLASS against MODE_INT.
15475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15476 Alan Hayward <alan.hayward@arm.com>
15477 David Sherwood <david.sherwood@arm.com>
15479 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15480 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15481 shift_unit_mode and for modes involved in scalar shifts.
15483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15484 Alan Hayward <alan.hayward@arm.com>
15485 David Sherwood <david.sherwood@arm.com>
15487 * combine.c (force_int_to_mode): New function, split out from...
15488 (force_to_mode): ...here. Keep xmode up-to-date and use it
15489 instead of GET_MODE (x).
15491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15492 Alan Hayward <alan.hayward@arm.com>
15493 David Sherwood <david.sherwood@arm.com>
15495 * optabs-query.h (extraction_insn::struct_mode): Change type to
15496 opt_scalar_int_mode and update comment.
15497 (extraction_insn::field_mode): Change type to scalar_int_mode.
15498 (extraction_insn::pos_mode): Likewise.
15499 * combine.c (make_extraction): Update accordingly.
15500 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15501 (get_optab_extraction_insn): Likewise.
15502 * recog.c (simplify_while_replacing): Likewise.
15503 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15504 parameter to opt_scalar_int_mode.
15506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15507 Alan Hayward <alan.hayward@arm.com>
15508 David Sherwood <david.sherwood@arm.com>
15510 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15511 to a scalar_int_mode instead of a machine_mode.
15512 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15513 (get_best_mode): Return a boolean and use a pointer argument to store
15514 the selected mode. Replace the limit mode parameter with a bit limit.
15515 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15516 for the values returned by bit_field_mode_iterator::next_mode.
15517 (store_bit_field): Update call to get_best_mode.
15518 (store_fixed_bit_field): Likewise.
15519 (extract_fixed_bit_field): Likewise.
15520 * expr.c (optimize_bitfield_assignment_op): Likewise.
15521 * fold-const.c (optimize_bit_field_compare): Likewise.
15522 (fold_truth_andor_1): Likewise.
15523 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15524 Update for new type of m_mode.
15525 (get_best_mode): As above.
15527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15528 Alan Hayward <alan.hayward@arm.com>
15529 David Sherwood <david.sherwood@arm.com>
15531 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15532 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15533 (store_bit_field): Check is_a <scalar_int_mode> before calling
15534 strict_volatile_bitfield_p.
15535 (extract_bit_field): Likewise.
15537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15538 Alan Hayward <alan.hayward@arm.com>
15539 David Sherwood <david.sherwood@arm.com>
15541 * target.def (cstore_mode): Return a scalar_int_mode.
15542 * doc/tm.texi: Regenerate.
15543 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15544 * targhooks.h (default_cstore_mode): Likewise.
15545 * targhooks.c (default_cstore_mode): Likewise, using a forced
15547 * expmed.c (emit_cstore): Expect the target of the cstore to be
15550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15551 Alan Hayward <alan.hayward@arm.com>
15552 David Sherwood <david.sherwood@arm.com>
15554 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15556 (niter_desc): Likewise mode.
15557 (iv_analyze): Add a mode parameter.
15559 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15560 and change its type to scalar_int_mode.
15561 * loop-iv.c: Update commentary at head of file.
15562 (iv_constant): Pass the mode paraeter before the rtx it describes
15563 and change its type to scalar_int_mode. Remove VOIDmode handling.
15564 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15565 (iv_extend): Likewise.
15566 (shorten_into_mode): Likewise.
15567 (iv_add): Use scalar_int_mode.
15568 (iv_mult): Likewise.
15569 (iv_shift): Likewise.
15570 (canonicalize_iv_subregs): Likewise.
15571 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15572 it describes and change its mode to scalar_int_mode. Also change
15573 the type of the returned inner_mode to scalar_int_mode.
15574 (get_biv_step): Likewise, turning outer_mode from a pointer
15575 into a direct parameter. Update call to get_biv_step_1.
15576 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15577 iv_constant and get_biv_step.
15578 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15579 and change its type to scalar_int_mode. Don't initialise iv->mode
15580 to VOIDmode and remove later checks for its still being VOIDmode.
15581 Update calls to iv_analyze_op and iv_analyze_expr. Check
15582 is_a <scalar_int_mode> when changing the mode under consideration.
15583 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15584 Update call to iv_analyze_expr.
15585 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15586 inner register is not also a scalar_int_mode. Update call to
15588 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15589 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15590 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15591 separate mode class checks. Update calls to iv_analyze. Remove
15592 fix-up of VOIDmodes after iv_analyze_biv.
15593 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15594 don't have a scalar_int_mode. Update call to biv_p.
15596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15597 Alan Hayward <alan.hayward@arm.com>
15598 David Sherwood <david.sherwood@arm.com>
15600 * cfgexpand.c (convert_debug_memory_address): Use
15601 as_a <scalar_int_mode>.
15602 * combine.c (expand_compound_operation): Likewise.
15603 (make_extraction): Likewise.
15604 (change_zero_ext): Likewise.
15605 (simplify_comparison): Likewise.
15606 * cse.c (cse_insn): Likewise.
15607 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15608 (mem_loc_descriptor): Likewise.
15609 (loc_descriptor): Likewise.
15610 * expmed.c (init_expmed_one_mode): Likewise.
15611 (synth_mult): Likewise.
15612 (emit_store_flag_1): Likewise.
15613 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15614 of a comparison with size.
15615 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15616 (reduce_to_bit_field_precision): Likewise.
15617 * function.c (expand_function_end): Likewise.
15618 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15619 * loop-doloop.c (doloop_modify): Likewise.
15620 * optabs.c (expand_binop): Likewise.
15621 (expand_unop): Likewise.
15622 (expand_copysign_absneg): Likewise.
15623 (prepare_cmp_insn): Likewise.
15624 (maybe_legitimize_operand): Likewise.
15625 * recog.c (const_scalar_int_operand): Likewise.
15626 * rtlanal.c (get_address_mode): Likewise.
15627 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15628 (simplify_cond_clz_ctz): Likewise.
15629 * tree-nested.c (get_nl_goto_field): Likewise.
15630 * tree.c (build_vector_type_for_mode): Likewise.
15631 * var-tracking.c (use_narrower_mode): Likewise.
15633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15634 Alan Hayward <alan.hayward@arm.com>
15635 David Sherwood <david.sherwood@arm.com>
15637 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15638 * builtins.c (expand_builtin_signbit): Use it.
15639 * cfgexpand.c (expand_debug_expr): Likewise.
15640 * dojump.c (do_jump): Likewise.
15641 (do_compare_and_jump): Likewise.
15642 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15643 * expmed.c (make_tree): Likewise.
15644 * expr.c (expand_expr_real_2): Likewise.
15645 (expand_expr_real_1): Likewise.
15646 (try_casesi): Likewise.
15647 * fold-const-call.c (fold_const_call_ss): Likewise.
15648 * fold-const.c (unextend): Likewise.
15649 (extract_muldiv_1): Likewise.
15650 (fold_single_bit_test): Likewise.
15651 (native_encode_int): Likewise.
15652 (native_encode_string): Likewise.
15653 (native_interpret_int): Likewise.
15654 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15655 * internal-fn.c (expand_addsub_overflow): Likewise.
15656 (expand_neg_overflow): Likewise.
15657 (expand_mul_overflow): Likewise.
15658 (expand_arith_overflow): Likewise.
15659 * match.pd: Likewise.
15660 * stor-layout.c (layout_type): Likewise.
15661 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15662 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15663 * tree-ssanames.c (get_range_info): Likewise.
15664 * tree-switch-conversion.c (array_value_type) Likewise.
15665 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15666 (vect_recog_divmod_pattern): Likewise.
15667 (vect_recog_mixed_size_cond_pattern): Likewise.
15668 * tree-vrp.c (extract_range_basic): Likewise.
15669 (simplify_float_conversion_using_ranges): Likewise.
15670 * tree.c (int_fits_type_p): Likewise.
15671 * ubsan.c (instrument_bool_enum_load): Likewise.
15672 * varasm.c (mergeable_string_section): Likewise.
15673 (narrowing_initializer_constant_valid_p): Likewise.
15674 (output_constant): Likewise.
15676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15677 Alan Hayward <alan.hayward@arm.com>
15678 David Sherwood <david.sherwood@arm.com>
15680 * machmode.h (NARROWEST_INT_MODE): New macro.
15681 * expr.c (alignment_for_piecewise_move): Use it instead of
15682 GET_CLASS_NARROWEST_MODE (MODE_INT).
15683 (push_block): Likewise.
15684 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15686 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15689 Alan Hayward <alan.hayward@arm.com>
15690 David Sherwood <david.sherwood@arm.com>
15692 * postreload.c (move2add_valid_value_p): Change the type of the
15693 mode parameter to scalar_int_mode.
15694 (move2add_use_add2_insn): Add a mode parameter and use it instead
15696 (move2add_use_add3_insn): Likewise.
15697 (reload_cse_move2add): Update accordingly.
15699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15700 Alan Hayward <alan.hayward@arm.com>
15701 David Sherwood <david.sherwood@arm.com>
15703 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15705 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15706 * optabs.c (expand_unop): Likewise.
15708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15709 Alan Hayward <alan.hayward@arm.com>
15710 David Sherwood <david.sherwood@arm.com>
15712 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15713 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15714 (popcount_loc_descriptor): Likewise.
15715 (bswap_loc_descriptor): Likewise.
15716 (rotate_loc_descriptor): Likewise.
15717 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15718 calling the functions above.
15720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15721 Alan Hayward <alan.hayward@arm.com>
15722 David Sherwood <david.sherwood@arm.com>
15724 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15726 (try_combine): Likewise.
15727 (simplify_if_then_else): Likewise.
15728 * cse.c (cse_insn): Likewise.
15729 * dwarf2out.c (mem_loc_descriptor): Likewise.
15730 * emit-rtl.c (gen_lowpart_common): Likewise.
15731 * simplify-rtx.c (simplify_truncation): Likewise.
15732 (simplify_binary_operation_1): Likewise.
15733 (simplify_const_relational_operation): Likewise.
15734 (simplify_ternary_operation): Likewise.
15735 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15738 Alan Hayward <alan.hayward@arm.com>
15739 David Sherwood <david.sherwood@arm.com>
15741 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15742 * reload.c (push_reload): Likewise.
15743 (find_reloads): Likewise.
15745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15746 Alan Hayward <alan.hayward@arm.com>
15747 David Sherwood <david.sherwood@arm.com>
15749 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15750 (make_compound_operation_int): Likewise.
15751 (change_zero_ext): Likewise.
15752 * expr.c (convert_move): Likewise.
15753 (convert_modes): Likewise.
15754 * fwprop.c (forward_propagate_subreg): Likewise.
15755 * loop-iv.c (get_biv_step_1): Likewise.
15756 * optabs.c (widen_operand): Likewise.
15757 * postreload.c (move2add_valid_value_p): Likewise.
15758 * recog.c (simplify_while_replacing): Likewise.
15759 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15760 (simplify_binary_operation_1): Likewise. Remove redundant
15761 mode equality check.
15763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15764 Alan Hayward <alan.hayward@arm.com>
15765 David Sherwood <david.sherwood@arm.com>
15767 * combine.c (combine_simplify_rtx): Add checks for
15768 is_a <scalar_int_mode>.
15769 (simplify_if_then_else): Likewise.
15770 (make_field_assignment): Likewise.
15771 (simplify_comparison): Likewise.
15772 * ifcvt.c (noce_try_bitop): Likewise.
15773 * loop-invariant.c (canonicalize_address_mult): Likewise.
15774 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15777 Alan Hayward <alan.hayward@arm.com>
15778 David Sherwood <david.sherwood@arm.com>
15780 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15781 is_a <scalar_int_mode> instead of != BLKmode.
15783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15784 Alan Hayward <alan.hayward@arm.com>
15785 David Sherwood <david.sherwood@arm.com>
15787 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15788 instead of != VOIDmode.
15789 * combine.c (if_then_else_cond): Likewise.
15790 (change_zero_ext): Likewise.
15791 * dwarf2out.c (mem_loc_descriptor): Likewise.
15792 (loc_descriptor): Likewise.
15793 * rtlanal.c (canonicalize_condition): Likewise.
15794 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15797 Alan Hayward <alan.hayward@arm.com>
15798 David Sherwood <david.sherwood@arm.com>
15800 * simplify-rtx.c (simplify_binary_operation_1): Use
15801 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15804 Alan Hayward <alan.hayward@arm.com>
15805 David Sherwood <david.sherwood@arm.com>
15807 * wide-int.h (int_traits<unsigned char>) New class.
15808 (int_traits<unsigned short>) Likewise.
15809 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15810 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15812 * combine.c (set_nonzero_bits_and_sign_copies): Use
15813 is_a <scalar_int_mode>.
15814 (find_split_point): Likewise.
15815 (combine_simplify_rtx): Likewise.
15816 (simplify_logical): Likewise.
15817 (expand_compound_operation): Likewise.
15818 (expand_field_assignment): Likewise.
15819 (make_compound_operation): Likewise.
15820 (extended_count): Likewise.
15821 (change_zero_ext): Likewise.
15822 (simplify_comparison): Likewise.
15823 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15824 (ucompare_loc_descriptor): Likewise.
15825 (minmax_loc_descriptor): Likewise.
15826 (mem_loc_descriptor): Likewise.
15827 (loc_descriptor): Likewise.
15828 * expmed.c (init_expmed_one_mode): Likewise.
15829 * lra-constraints.c (lra_constraint_offset): Likewise.
15830 * optabs.c (prepare_libcall_arg): Likewise.
15831 * postreload.c (move2add_note_store): Likewise.
15832 * reload.c (operands_match_p): Likewise.
15833 * rtl.h (load_extend_op): Likewise.
15834 * rtlhooks.c (gen_lowpart_general): Likewise.
15835 * simplify-rtx.c (simplify_truncation): Likewise.
15836 (simplify_unary_operation_1): Likewise.
15837 (simplify_binary_operation_1): Likewise.
15838 (simplify_const_binary_operation): Likewise.
15839 (simplify_const_relational_operation): Likewise.
15840 (simplify_subreg): Likewise.
15841 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15842 * var-tracking.c (adjust_mems): Likewise.
15843 (prepare_call_arguments): Likewise.
15845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15846 Alan Hayward <alan.hayward@arm.com>
15847 David Sherwood <david.sherwood@arm.com>
15849 * machmode.h (is_int_mode): New fuction.
15850 * combine.c (find_split_point): Use it.
15851 (combine_simplify_rtx): Likewise.
15852 (simplify_if_then_else): Likewise.
15853 (simplify_set): Likewise.
15854 (simplify_shift_const_1): Likewise.
15855 (simplify_comparison): Likewise.
15856 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15857 * cse.c (notreg_cost): Likewise.
15858 (cse_insn): Likewise.
15859 * cselib.c (cselib_lookup_1): Likewise.
15860 * dojump.c (do_jump_1): Likewise.
15861 (do_compare_rtx_and_jump): Likewise.
15862 * dse.c (get_call_args): Likewise.
15863 * dwarf2out.c (rtl_for_decl_init): Likewise.
15864 (native_encode_initializer): Likewise.
15865 * expmed.c (emit_store_flag_1): Likewise.
15866 (emit_store_flag): Likewise.
15867 * expr.c (convert_modes): Likewise.
15868 (store_field): Likewise.
15869 (expand_expr_real_1): Likewise.
15870 * fold-const.c (fold_read_from_constant_string): Likewise.
15871 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15872 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15873 * optabs.c (expand_binop): Likewise.
15874 (expand_unop): Likewise.
15875 (expand_abs_nojump): Likewise.
15876 (expand_one_cmpl_abs_nojump): Likewise.
15877 * simplify-rtx.c (mode_signbit_p): Likewise.
15878 (val_signbit_p): Likewise.
15879 (val_signbit_known_set_p): Likewise.
15880 (val_signbit_known_clear_p): Likewise.
15881 (simplify_relational_operation_1): Likewise.
15882 * tree.c (vector_type_mode): Likewise.
15884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15885 Alan Hayward <alan.hayward@arm.com>
15886 David Sherwood <david.sherwood@arm.com>
15888 * machmode.h (smallest_mode_for_size): Fix formatting.
15889 (smallest_int_mode_for_size): New function.
15890 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15891 instead of smallest_mode_for_size.
15892 * combine.c (make_extraction): Likewise.
15893 * config/arc/arc.c (arc_expand_movmem): Likewise.
15894 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15895 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15896 * config/s390/s390.c (s390_expand_insv): Likewise.
15897 * config/sparc/sparc.c (assign_int_registers): Likewise.
15898 * config/spu/spu.c (spu_function_value): Likewise.
15899 (spu_function_arg): Likewise.
15900 * coverage.c (get_gcov_type): Likewise.
15901 (get_gcov_unsigned_t): Likewise.
15902 * dse.c (find_shift_sequence): Likewise.
15903 * expmed.c (store_bit_field_1): Likewise.
15904 * expr.c (convert_move): Likewise.
15905 (store_field): Likewise.
15906 * internal-fn.c (expand_arith_overflow): Likewise.
15907 * optabs-query.c (get_best_extraction_insn): Likewise.
15908 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15909 * stor-layout.c (layout_type): Likewise.
15910 (initialize_sizetypes): Likewise.
15911 * targhooks.c (default_get_mask_mode): Likewise.
15912 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15915 Alan Hayward <alan.hayward@arm.com>
15916 David Sherwood <david.sherwood@arm.com>
15918 * machmode.h (opt_mode::else_blk): New function.
15919 (int_mode_for_mode): Declare.
15920 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15921 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15923 * cfgexpand.c (expand_debug_expr): Likewise.
15924 * combine.c (gen_lowpart_or_truncate): Likewise.
15925 (gen_lowpart_for_combine): Likewise.
15926 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15927 * config/avr/avr.c (avr_to_int_mode): Likewise.
15928 (avr_out_plus_1): Likewise.
15929 (avr_out_plus): Likewise.
15930 (avr_out_round): Likewise.
15931 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15932 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15933 (s390_expand_vcond): Likewise.
15934 * config/spu/spu.c (spu_split_immediate): Likewise.
15935 (spu_expand_mov): Likewise.
15936 * dse.c (get_stored_val): Likewise.
15937 * expmed.c (store_bit_field_1): Likewise.
15938 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15940 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15941 (extract_low_bits): Likewise.
15942 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15943 handling rather than repeating the check.
15944 (emit_group_store): Likewise.
15945 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15946 * optabs.c (expand_absneg_bit): Likewise.
15947 (expand_copysign_absneg): Likewise.
15948 (expand_copysign_bit): Likewise.
15949 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15950 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15951 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15952 * var-tracking.c (prepare_call_arguments): Likewise.
15953 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15954 int_mode_for_mode instead of mode_for_size.
15955 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15958 Alan Hayward <alan.hayward@arm.com>
15959 David Sherwood <david.sherwood@arm.com>
15961 * machmode.h (int_mode_for_size): New function.
15962 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15963 instead of mode_for_size.
15964 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15966 * combine.c (expand_field_assignment): Use int_mode_for_size
15967 instead of mode_for_size.
15968 (make_extraction): Likewise.
15969 (simplify_shift_const_1): Likewise.
15970 (simplify_comparison): Likewise.
15971 * dojump.c (do_jump): Likewise.
15972 * dwarf2out.c (mem_loc_descriptor): Likewise.
15973 * emit-rtl.c (init_derived_machine_modes): Likewise.
15974 * expmed.c (flip_storage_order): Likewise.
15975 (convert_extracted_bit_field): Likewise.
15976 * expr.c (copy_blkmode_from_reg): Likewise.
15977 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15978 * internal-fn.c (expand_mul_overflow): Likewise.
15979 * lower-subreg.c (simple_move): Likewise.
15980 * optabs-libfuncs.c (init_optabs): Likewise.
15981 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15982 * tree.c (vector_type_mode): Likewise.
15983 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15984 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15985 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15986 * tree-vect-stmts.c (vectorizable_load): Likewise.
15987 (vectorizable_store): Likewise.
15989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15990 Alan Hayward <alan.hayward@arm.com>
15991 David Sherwood <david.sherwood@arm.com>
15993 * coretypes.h (pod_mode): New type.
15994 (scalar_int_mode_pod): New typedef.
15995 * machmode.h (pod_mode): New class.
15996 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15997 * genmodes.c (emit_mode_int_n): Update accordingly.
15998 * lower-subreg.h (target_lower_subreg): Change type to
15999 scalar_int_mode_pod.
16000 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16001 scalar_int_mode_pod.
16003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16004 Alan Hayward <alan.hayward@arm.com>
16005 David Sherwood <david.sherwood@arm.com>
16007 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16008 machine_mode to scalar_int_mode.
16009 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16010 (rs6000_option_override_internal): Remove cast to int.
16011 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16012 machine_mode to scalar_int_mode.
16013 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16014 (rs6000_option_override_internal): Remove cast to int.
16015 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16016 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16018 * config/s390/s390.c (s390_expand_builtin): Likewise.
16019 * coretypes.h (scalar_int_mode): New type.
16020 (opt_scalar_int_mode): New typedef.
16021 * machmode.h (scalar_int_mode): New class.
16022 (scalar_int_mode::includes_p): New function.
16023 (byte_mode): Change type to scalar_int_mode.
16024 (word_mode): Likewise.
16025 (ptr_mode): Likewise.
16026 * emit-rtl.c (byte_mode): Likewise.
16027 (word_mode): Likewise.
16028 (ptr_mode): Likewise.
16029 (init_derived_machine_modes): Update accordingly.
16030 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16031 and MODE_PARTIAL_INT.
16032 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16033 opt_scalar_int_mode.
16035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16036 Alan Hayward <alan.hayward@arm.com>
16037 David Sherwood <david.sherwood@arm.com>
16039 * target.def (libgcc_floating_mode_supported_p): Take a
16041 * doc/tm.texi: Regenerate.
16042 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16044 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16045 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16049 Alan Hayward <alan.hayward@arm.com>
16050 David Sherwood <david.sherwood@arm.com>
16052 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16053 * doc/tm.texi: Regenerate.
16054 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16055 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16056 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16057 * targhooks.h (default_floatn_mode): Likewise.
16058 * targhooks.c (default_floatn_mode): Likewise.
16059 * tree.c (build_common_tree_nodes): Update accordingly.
16061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16062 Alan Hayward <alan.hayward@arm.com>
16063 David Sherwood <david.sherwood@arm.com>
16065 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16066 (mode_iterator::iterate_p): Likewise.
16067 (mode_iterator::get_wider): Likewise.
16068 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16071 Alan Hayward <alan.hayward@arm.com>
16072 David Sherwood <david.sherwood@arm.com>
16074 * coretypes.h (opt_scalar_float_mode): New typedef.
16075 * machmode.h (float_mode_for_size): New function.
16076 * emit-rtl.c (double_mode): Delete.
16077 (init_emit_once): Use float_mode_for_size.
16078 * stor-layout.c (layout_type): Likewise.
16079 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16082 Alan Hayward <alan.hayward@arm.com>
16083 David Sherwood <david.sherwood@arm.com>
16085 * output.h (assemble_real): Take a scalar_float_mode.
16086 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16087 * config/arm/arm.md (consttable_4): Likewise.
16088 (consttable_8): Likewise.
16089 (consttable_16): Likewise.
16090 * config/mips/mips.md (consttable_float): Likewise.
16091 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16092 * varasm.c (assemble_real): Take a scalar_float_mode.
16093 (output_constant_pool_2): Update accordingly.
16094 (output_constant): Likewise.
16096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16097 Alan Hayward <alan.hayward@arm.com>
16098 David Sherwood <david.sherwood@arm.com>
16100 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16101 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16102 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16103 (native_encode_real): Likewise.
16104 (native_interpret_real): Likewise.
16105 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16106 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16109 Alan Hayward <alan.hayward@arm.com>
16110 David Sherwood <david.sherwood@arm.com>
16112 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16113 <scalar_float_mode>. Simplify.
16114 (gen_extend_conv_libfunc): Likewise.
16116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16117 Alan Hayward <alan.hayward@arm.com>
16118 David Sherwood <david.sherwood@arm.com>
16120 * coretypes.h (scalar_float_mode): New type.
16121 * machmode.h (mode_traits::from_int): Use machine_mode if
16122 USE_ENUM_MODES is defined.
16123 (is_a): New function.
16125 (dyn_cast): Likewise.
16126 (scalar_float_mode): New class.
16127 (scalar_float_mode::includes_p): New function.
16128 (is_float_mode): Likewise.
16129 * gdbhooks.py (MachineModePrinter): New class.
16130 (build_pretty_printer): Use it for scalar_float_mode.
16131 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16132 (format_helper::format_helper): Turn into a template.
16133 * genmodes.c (get_mode_class): New function.
16134 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16135 or machine_mode if none.
16136 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16137 as_a <scalar_float_mode>.
16138 * dwarf2out.c (mem_loc_descriptor): Likewise.
16139 (insert_float): Likewise.
16140 (add_const_value_attribute): Likewise.
16141 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16142 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16143 (expand_unop): Update accordingly.
16144 (expand_abs_nojump): Likewise.
16145 (expand_copysign_absneg): Take a scalar_float_mode.
16146 (expand_copysign_bit): Likewise.
16147 (expand_copysign): Update accordingly.
16149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16150 Alan Hayward <alan.hayward@arm.com>
16151 David Sherwood <david.sherwood@arm.com>
16153 * coretypes.h (opt_mode): New class.
16154 * machmode.h (opt_mode): Likewise.
16155 (opt_mode::else_void): New function.
16156 (opt_mode::require): Likewise.
16157 (opt_mode::exists): Likewise.
16158 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16159 (GET_MODE_2XWIDER_MODE): Likewise.
16160 (mode_iterator::get_wider): Update accordingly.
16161 (mode_iterator::get_2xwider): Likewise.
16162 (mode_iterator::get_known_wider): Likewise, turning into a template.
16163 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16164 forcing a wider mode to exist.
16165 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16166 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16167 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16168 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16169 * lower-subreg.c (init_lower_subreg): Likewise.
16170 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16171 on the final iteration.
16172 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16173 a wider mode exists before asking for a move pattern.
16174 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16175 forcing a wider mode to exist.
16176 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16177 returning false if no such mode exists.
16178 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16179 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16180 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16181 Avoid checking for a MODE_INT if we already know the mode is not a
16183 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16184 forcing a wider mode to exist.
16185 (expmed_mult_highpart_optab): Likewise.
16186 (expmed_mult_highpart): Likewise.
16187 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16189 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16190 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16191 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16192 * internal-fn.c (expand_mul_overflow): Update use of
16193 GET_MODE_2XWIDER_MODE.
16194 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16195 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16196 GET_MODE_WIDER_MODE.
16197 (convert_plusminus_to_widen): Likewise.
16198 * tree-switch-conversion.c (array_value_type): Likewise.
16199 * var-tracking.c (emit_note_insn_var_location): Likewise.
16200 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16201 Return false inside rather than outside the loop if no wider mode
16203 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16204 and GET_MODE_2XWIDER_MODE
16205 (can_compare_p): Use else_void.
16206 * gdbhooks.py (OptMachineModePrinter): New class.
16207 (build_pretty_printer): Use it for opt_mode.
16209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16210 Alan Hayward <alan.hayward@arm.com>
16211 David Sherwood <david.sherwood@arm.com>
16213 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16214 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16217 Alan Hayward <alan.hayward@arm.com>
16218 David Sherwood <david.sherwood@arm.com>
16220 * machmode.h (mode_traits): New structure.
16221 (get_narrowest_mode): New function.
16222 (mode_iterator::start): Likewise.
16223 (mode_iterator::iterate_p): Likewise.
16224 (mode_iterator::get_wider): Likewise.
16225 (mode_iterator::get_known_wider): Likewise.
16226 (mode_iterator::get_2xwider): Likewise.
16227 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16228 (FOR_EACH_MODE): Likewise.
16229 (FOR_EACH_MODE_FROM): Likewise.
16230 (FOR_EACH_MODE_UNTIL): Likewise.
16231 (FOR_EACH_WIDER_MODE): Likewise.
16232 (FOR_EACH_2XWIDER_MODE): Likewise.
16233 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16234 * combine.c (simplify_comparison): Likewise
16235 * config/i386/i386.c (type_natural_mode): Likewise.
16236 * cse.c (cse_insn): Likewise.
16237 * dse.c (find_shift_sequence): Likewise.
16238 * emit-rtl.c (init_derived_machine_modes): Likewise.
16239 (init_emit_once): Likewise.
16240 * explow.c (hard_function_value): Likewise.
16241 * expmed.c (extract_fixed_bit_field_1): Likewise.
16242 (extract_bit_field_1): Likewise.
16243 (expand_divmod): Likewise.
16244 (emit_store_flag_1): Likewise.
16245 * expr.c (init_expr_target): Likewise.
16246 (convert_move): Likewise.
16247 (alignment_for_piecewise_move): Likewise.
16248 (widest_int_mode_for_size): Likewise.
16249 (emit_block_move_via_movmem): Likewise.
16250 (copy_blkmode_to_reg): Likewise.
16251 (set_storage_via_setmem): Likewise.
16252 (compress_float_constant): Likewise.
16253 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16254 * optabs-query.c (get_best_extraction_insn): Likewise.
16255 * optabs.c (expand_binop): Likewise.
16256 (expand_twoval_unop): Likewise.
16257 (expand_twoval_binop): Likewise.
16258 (widen_leading): Likewise.
16259 (widen_bswap): Likewise.
16260 (expand_parity): Likewise.
16261 (expand_unop): Likewise.
16262 (prepare_cmp_insn): Likewise.
16263 (prepare_float_lib_cmp): Likewise.
16264 (expand_float): Likewise.
16265 (expand_fix): Likewise.
16266 (expand_sfix_optab): Likewise.
16267 * postreload.c (move2add_use_add2_insn): Likewise.
16268 * reg-stack.c (reg_to_stack): Likewise.
16269 * reginfo.c (choose_hard_reg_mode): Likewise.
16270 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16271 * stor-layout.c (mode_for_size): Likewise.
16272 (smallest_mode_for_size): Likewise.
16273 (mode_for_vector): Likewise.
16274 (finish_bitfield_representative): Likewise.
16275 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16276 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16277 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16278 * var-tracking.c (prepare_call_arguments): Likewise.
16280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16281 Alan Hayward <alan.hayward@arm.com>
16282 David Sherwood <david.sherwood@arm.com>
16284 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16285 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16286 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16287 * machmode.h (mode_size): Move earlier in file.
16288 (mode_precision): Likewise.
16289 (mode_inner): Likewise.
16290 (mode_nunits): Likewise.
16291 (mode_unit_size): Likewise.
16292 (unit_unit_precision): Likewise.
16293 (mode_wider): Likewise.
16294 (mode_2xwider): Likewise.
16295 (machine_mode): New class.
16296 (mode_to_bytes): New function.
16297 (mode_to_bits): Likewise.
16298 (mode_to_precision): Likewise.
16299 (mode_to_inner): Likewise.
16300 (mode_to_unit_size): Likewise.
16301 (mode_to_unit_precision): Likewise.
16302 (mode_to_nunits): Likewise.
16303 (GET_MODE_SIZE): Use mode_to_bytes.
16304 (GET_MODE_BITSIZE): Use mode_to_bits.
16305 (GET_MODE_PRECISION): Use mode_to_precision.
16306 (GET_MODE_INNER): Use mode_to_inner.
16307 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16308 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16309 (GET_MODE_NUNITS): Use mode_to_nunits.
16310 * system.h (ALWAYS_INLINE): New macro.
16311 * config/powerpcspe/powerpcspe-c.c
16312 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16313 int for arg1_mode and arg2_mode.
16315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16316 Alan Hayward <alan.hayward@arm.com>
16317 David Sherwood <david.sherwood@arm.com>
16319 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16320 Prefix mode names with E_ in case statements.
16321 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16322 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16323 (aarch64_split_simd_move): Likewise.
16324 (aarch64_gen_storewb_pair): Likewise.
16325 (aarch64_gen_loadwb_pair): Likewise.
16326 (aarch64_gen_store_pair): Likewise.
16327 (aarch64_gen_load_pair): Likewise.
16328 (aarch64_get_condition_code_1): Likewise.
16329 (aarch64_constant_pool_reload_icode): Likewise.
16330 (get_rsqrte_type): Likewise.
16331 (get_rsqrts_type): Likewise.
16332 (get_recpe_type): Likewise.
16333 (get_recps_type): Likewise.
16334 (aarch64_gimplify_va_arg_expr): Likewise.
16335 (aarch64_simd_container_mode): Likewise.
16336 (aarch64_emit_load_exclusive): Likewise.
16337 (aarch64_emit_store_exclusive): Likewise.
16338 (aarch64_expand_compare_and_swap): Likewise.
16339 (aarch64_gen_atomic_cas): Likewise.
16340 (aarch64_emit_bic): Likewise.
16341 (aarch64_emit_atomic_swap): Likewise.
16342 (aarch64_emit_atomic_load_op): Likewise.
16343 (aarch64_evpc_trn): Likewise.
16344 (aarch64_evpc_uzp): Likewise.
16345 (aarch64_evpc_zip): Likewise.
16346 (aarch64_evpc_ext): Likewise.
16347 (aarch64_evpc_rev): Likewise.
16348 (aarch64_evpc_dup): Likewise.
16349 (aarch64_gen_ccmp_first): Likewise.
16350 (aarch64_gen_ccmp_next): Likewise.
16351 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16352 (alpha_emit_xfloating_libcall): Likewise.
16353 (emit_insxl): Likewise.
16354 (alpha_arg_type): Likewise.
16355 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16356 (arc_preferred_simd_mode): Likewise.
16357 (arc_secondary_reload): Likewise.
16358 (get_arc_condition_code): Likewise.
16359 (arc_print_operand): Likewise.
16360 (arc_legitimate_constant_p): Likewise.
16361 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16362 * config/arc/arc.md (casesi_load): Likewise.
16363 (casesi_compact_jump): Likewise.
16364 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16365 (cc_use_register): Likewise.
16366 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16367 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16368 (arm_init_iwmmxt_builtins): Likewise.
16369 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16370 (neon_expand_vector_init): Likewise.
16371 (arm_attr_length_move_neon): Likewise.
16372 (maybe_get_arm_condition_code): Likewise.
16373 (arm_emit_vector_const): Likewise.
16374 (arm_preferred_simd_mode): Likewise.
16375 (arm_output_iwmmxt_tinsr): Likewise.
16376 (thumb1_output_casesi): Likewise.
16377 (thumb2_output_casesi): Likewise.
16378 (arm_emit_load_exclusive): Likewise.
16379 (arm_emit_store_exclusive): Likewise.
16380 (arm_expand_compare_and_swap): Likewise.
16381 (arm_evpc_neon_vuzp): Likewise.
16382 (arm_evpc_neon_vzip): Likewise.
16383 (arm_evpc_neon_vrev): Likewise.
16384 (arm_evpc_neon_vtrn): Likewise.
16385 (arm_evpc_neon_vext): Likewise.
16386 (arm_validize_comparison): Likewise.
16387 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16388 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16389 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16390 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16391 (c6x_preferred_simd_mode): Likewise.
16392 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16393 (epiphany_rtx_costs): Likewise.
16394 * config/epiphany/predicates.md (proper_comparison_operator):
16396 * config/frv/frv.c (condexec_memory_operand): Likewise.
16397 (frv_emit_move): Likewise.
16398 (output_move_single): Likewise.
16399 (output_condmove_single): Likewise.
16400 (frv_hard_regno_mode_ok): Likewise.
16401 (frv_matching_accg_mode): Likewise.
16402 * config/h8300/h8300.c (split_adds_subs): Likewise.
16403 (h8300_rtx_costs): Likewise.
16404 (h8300_print_operand): Likewise.
16405 (compute_mov_length): Likewise.
16406 (output_logical_op): Likewise.
16407 (compute_logical_op_length): Likewise.
16408 (compute_logical_op_cc): Likewise.
16409 (h8300_shift_needs_scratch_p): Likewise.
16410 (output_a_shift): Likewise.
16411 (compute_a_shift_length): Likewise.
16412 (compute_a_shift_cc): Likewise.
16413 (expand_a_rotate): Likewise.
16414 (output_a_rotate): Likewise.
16415 * config/i386/i386.c (classify_argument): Likewise.
16416 (function_arg_advance_32): Likewise.
16417 (function_arg_32): Likewise.
16418 (function_arg_64): Likewise.
16419 (function_value_64): Likewise.
16420 (ix86_gimplify_va_arg): Likewise.
16421 (ix86_legitimate_constant_p): Likewise.
16422 (put_condition_code): Likewise.
16423 (split_double_mode): Likewise.
16424 (ix86_avx256_split_vector_move_misalign): Likewise.
16425 (ix86_expand_vector_logical_operator): Likewise.
16426 (ix86_split_idivmod): Likewise.
16427 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16428 (ix86_build_const_vector): Likewise.
16429 (ix86_build_signbit_mask): Likewise.
16430 (ix86_match_ccmode): Likewise.
16431 (ix86_cc_modes_compatible): Likewise.
16432 (ix86_expand_branch): Likewise.
16433 (ix86_expand_sse_cmp): Likewise.
16434 (ix86_expand_sse_movcc): Likewise.
16435 (ix86_expand_int_sse_cmp): Likewise.
16436 (ix86_expand_vec_perm_vpermi2): Likewise.
16437 (ix86_expand_vec_perm): Likewise.
16438 (ix86_expand_sse_unpack): Likewise.
16439 (ix86_expand_int_addcc): Likewise.
16440 (ix86_split_to_parts): Likewise.
16441 (ix86_vectorize_builtin_gather): Likewise.
16442 (ix86_vectorize_builtin_scatter): Likewise.
16443 (avx_vpermilp_parallel): Likewise.
16444 (inline_memory_move_cost): Likewise.
16445 (ix86_tieable_integer_mode_p): Likewise.
16446 (x86_maybe_negate_const_int): Likewise.
16447 (ix86_expand_vector_init_duplicate): Likewise.
16448 (ix86_expand_vector_init_one_nonzero): Likewise.
16449 (ix86_expand_vector_init_one_var): Likewise.
16450 (ix86_expand_vector_init_concat): Likewise.
16451 (ix86_expand_vector_init_interleave): Likewise.
16452 (ix86_expand_vector_init_general): Likewise.
16453 (ix86_expand_vector_set): Likewise.
16454 (ix86_expand_vector_extract): Likewise.
16455 (emit_reduc_half): Likewise.
16456 (ix86_emit_i387_round): Likewise.
16457 (ix86_mangle_type): Likewise.
16458 (ix86_expand_round_sse4): Likewise.
16459 (expand_vec_perm_blend): Likewise.
16460 (canonicalize_vector_int_perm): Likewise.
16461 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16462 (expand_vec_perm_1): Likewise.
16463 (expand_vec_perm_interleave3): Likewise.
16464 (expand_vec_perm_even_odd_pack): Likewise.
16465 (expand_vec_perm_even_odd_1): Likewise.
16466 (expand_vec_perm_broadcast_1): Likewise.
16467 (ix86_vectorize_vec_perm_const_ok): Likewise.
16468 (ix86_expand_vecop_qihi): Likewise.
16469 (ix86_expand_mul_widen_hilo): Likewise.
16470 (ix86_expand_sse2_abs): Likewise.
16471 (ix86_expand_pextr): Likewise.
16472 (ix86_expand_pinsr): Likewise.
16473 (ix86_preferred_simd_mode): Likewise.
16474 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16475 * config/i386/sse.md (*andnot<mode>3): Likewise.
16476 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16477 (*<code><mode>3): Likewise.
16478 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16479 (ia64_expand_atomic_op): Likewise.
16480 (ia64_arg_type): Likewise.
16481 (ia64_mode_to_int): Likewise.
16482 (ia64_scalar_mode_supported_p): Likewise.
16483 (ia64_vector_mode_supported_p): Likewise.
16484 (expand_vec_perm_broadcast): Likewise.
16485 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16486 (iq2000_function_arg_advance): Likewise.
16487 (iq2000_function_arg): Likewise.
16488 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16489 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16490 (m68k_libcall_value): Likewise.
16491 (m68k_function_value): Likewise.
16492 (sched_attr_op_type): Likewise.
16493 * config/mcore/mcore.c (mcore_output_move): Likewise.
16494 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16496 (microblaze_function_arg): Likewise.
16497 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16498 (mips_print_operand): Likewise.
16499 (mips_mode_ok_for_mov_fmt_p): Likewise.
16500 (mips_vector_mode_supported_p): Likewise.
16501 (mips_preferred_simd_mode): Likewise.
16502 (mips_expand_vpc_loongson_even_odd): Likewise.
16503 (mips_expand_vec_unpack): Likewise.
16504 (mips_expand_vi_broadcast): Likewise.
16505 (mips_expand_vector_init): Likewise.
16506 (mips_expand_vec_reduc): Likewise.
16507 (mips_expand_msa_cmp): Likewise.
16508 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16509 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16510 (cc_flags_for_mode): Likewise.
16511 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16512 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16513 (nds32_output_casesi_pc_relative): Likewise.
16514 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16515 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16516 (nvptx_gen_unpack): Likewise.
16517 (nvptx_gen_pack): Likewise.
16518 (nvptx_gen_shuffle): Likewise.
16519 (nvptx_gen_wcast): Likewise.
16520 (nvptx_preferred_simd_mode): Likewise.
16521 * config/pa/pa.c (pa_secondary_reload): Likewise.
16522 * config/pa/predicates.md (base14_operand): Likewise.
16523 * config/powerpcspe/powerpcspe-c.c
16524 (altivec_resolve_overloaded_builtin): Likewise.
16525 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16527 (rs6000_preferred_simd_mode): Likewise.
16528 (output_vec_const_move): Likewise.
16529 (rs6000_expand_vector_extract): Likewise.
16530 (rs6000_split_vec_extract_var): Likewise.
16531 (reg_offset_addressing_ok_p): Likewise.
16532 (rs6000_legitimate_offset_address_p): Likewise.
16533 (rs6000_legitimize_address): Likewise.
16534 (rs6000_emit_set_const): Likewise.
16535 (rs6000_const_vec): Likewise.
16536 (rs6000_emit_move): Likewise.
16537 (spe_build_register_parallel): Likewise.
16538 (rs6000_darwin64_record_arg_recurse): Likewise.
16539 (swap_selector_for_mode): Likewise.
16540 (spe_init_builtins): Likewise.
16541 (paired_init_builtins): Likewise.
16542 (altivec_init_builtins): Likewise.
16543 (do_load_for_compare): Likewise.
16544 (rs6000_generate_compare): Likewise.
16545 (rs6000_expand_float128_convert): Likewise.
16546 (emit_load_locked): Likewise.
16547 (emit_store_conditional): Likewise.
16548 (rs6000_output_function_epilogue): Likewise.
16549 (rs6000_handle_altivec_attribute): Likewise.
16550 (rs6000_function_value): Likewise.
16551 (emit_fusion_gpr_load): Likewise.
16552 (emit_fusion_p9_load): Likewise.
16553 (emit_fusion_p9_store): Likewise.
16554 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16555 (fusion_gpr_mem_load): Likewise.
16556 (fusion_addis_mem_combo_load): Likewise.
16557 (fusion_addis_mem_combo_store): Likewise.
16558 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16559 (fusion_gpr_mem_load): Likewise.
16560 (fusion_addis_mem_combo_load): Likewise.
16561 (fusion_addis_mem_combo_store): Likewise.
16562 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16564 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16565 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16566 (rs6000_preferred_simd_mode): Likewise.
16567 (output_vec_const_move): Likewise.
16568 (rs6000_expand_vector_extract): Likewise.
16569 (rs6000_split_vec_extract_var): Likewise.
16570 (reg_offset_addressing_ok_p): Likewise.
16571 (rs6000_legitimate_offset_address_p): Likewise.
16572 (rs6000_legitimize_address): Likewise.
16573 (rs6000_emit_set_const): Likewise.
16574 (rs6000_const_vec): Likewise.
16575 (rs6000_emit_move): Likewise.
16576 (rs6000_darwin64_record_arg_recurse): Likewise.
16577 (swap_selector_for_mode): Likewise.
16578 (paired_init_builtins): Likewise.
16579 (altivec_init_builtins): Likewise.
16580 (rs6000_expand_float128_convert): Likewise.
16581 (emit_load_locked): Likewise.
16582 (emit_store_conditional): Likewise.
16583 (rs6000_output_function_epilogue): Likewise.
16584 (rs6000_handle_altivec_attribute): Likewise.
16585 (rs6000_function_value): Likewise.
16586 (emit_fusion_gpr_load): Likewise.
16587 (emit_fusion_p9_load): Likewise.
16588 (emit_fusion_p9_store): Likewise.
16589 * config/rx/rx.c (rx_gen_move_template): Likewise.
16590 (flags_from_mode): Likewise.
16591 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16592 (s390_slb_comparison): Likewise.
16593 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16594 (s390_vector_mode_supported_p): Likewise.
16595 (s390_cc_modes_compatible): Likewise.
16596 (s390_match_ccmode_set): Likewise.
16597 (s390_canonicalize_comparison): Likewise.
16598 (s390_emit_compare_and_swap): Likewise.
16599 (s390_branch_condition_mask): Likewise.
16600 (s390_rtx_costs): Likewise.
16601 (s390_secondary_reload): Likewise.
16602 (__SECONDARY_RELOAD_CASE): Likewise.
16603 (s390_expand_cs): Likewise.
16604 (s390_preferred_simd_mode): Likewise.
16605 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16606 * config/sh/sh.c (sh_print_operand): Likewise.
16607 (dump_table): Likewise.
16608 (sh_secondary_reload): Likewise.
16609 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16610 * config/sh/sh.md (casesi_worker_1): Likewise.
16611 (casesi_worker_2): Likewise.
16612 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16613 (fcc_comparison_operator): Likewise.
16614 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16615 (emit_soft_tfmode_cvt): Likewise.
16616 (sparc_preferred_simd_mode): Likewise.
16617 (output_cbranch): Likewise.
16618 (sparc_print_operand): Likewise.
16619 (sparc_expand_vec_perm_bmask): Likewise.
16620 (vector_init_bshuffle): Likewise.
16621 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16622 (spu_vector_mode_supported_p): Likewise.
16623 (spu_expand_insv): Likewise.
16624 (spu_emit_branch_or_set): Likewise.
16625 (spu_handle_vector_attribute): Likewise.
16626 (spu_builtin_splats): Likewise.
16627 (spu_builtin_extract): Likewise.
16628 (spu_builtin_promote): Likewise.
16629 (spu_expand_sign_extend): Likewise.
16630 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16631 (tilegx_simd_int): Likewise.
16632 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16633 (tilepro_simd_int): Likewise.
16634 * config/v850/v850.c (const_double_split): Likewise.
16635 (v850_print_operand): Likewise.
16636 (ep_memory_offset): Likewise.
16637 * config/vax/vax.c (vax_rtx_costs): Likewise.
16638 (vax_output_int_move): Likewise.
16639 (vax_output_int_add): Likewise.
16640 (vax_output_int_subtract): Likewise.
16641 * config/visium/predicates.md (visium_branch_operator): Likewise.
16642 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16643 (visium_print_operand_address): Likewise.
16644 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16645 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16646 (xtensa_expand_conditional_branch): Likewise.
16647 (xtensa_copy_incoming_a7): Likewise.
16648 (xtensa_output_literal): Likewise.
16649 * dfp.c (decimal_real_maxval): Likewise.
16650 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16653 Alan Hayward <alan.hayward@arm.com>
16654 David Sherwood <david.sherwood@arm.com>
16656 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16657 (mode_nunits_inline): Likewise.
16658 (mode_inner_inline): Likewise.
16659 (mode_unit_size_inline): Likewise.
16660 (mode_unit_precision_inline): Likewise.
16661 (emit_insn_modes_h): Likewise. Also emit a #define of the
16663 (emit_mode_wider): Add an E_ prefix to mode names.
16664 (emit_mode_complex): Likewise.
16665 (emit_mode_inner): Likewise.
16666 (emit_mode_adjustments): Likewise.
16667 (emit_mode_int_n): Likewise.
16668 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16669 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16670 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16671 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16672 (CRC32_BUILTIN, ENTRY): Likewise.
16673 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16674 (aarch64_pop_regs): Likewise.
16675 (aarch64_process_components): Likewise.
16676 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16677 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16678 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16679 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16680 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16681 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16683 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16684 (pcmpistr): Likewise.
16685 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16686 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16687 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16689 * config/rl78/rl78.c (mduc_regs): Likewise.
16690 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16691 (htm_expand_builtin): Likewise.
16692 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16693 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16694 * config/xtensa/xtensa.c (print_operand): Likewise.
16695 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16696 (NUM_MODE_VECTOR_INT): Likewise.
16697 * genoutput.c (null_operand): Likewise.
16698 (output_operand_data): Likewise.
16699 * genrecog.c (print_parameter_value): Likewise.
16700 * lra.c (debug_operand_data): Likewise.
16702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16704 * dwarf2out.c (scompare_loc_descriptor_wide)
16705 (scompare_loc_descriptor_narrow): New functions, split out from...
16706 (scompare_loc_descriptor): ...here.
16707 * expmed.c (emit_store_flag_int): New function, split out from...
16708 (emit_store_flag): ...here.
16710 2017-08-30 Richard Biener <rguenther@suse.de>
16712 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16713 (dwarf2out_early_finish): Move setting of AT_pubnames from
16714 early debug output to early finish.
16716 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16718 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16719 and -mdata-region to the assembler.
16721 2017-08-30 Richard Biener <rguenther@suse.de>
16723 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16725 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16726 (dwarf2out_early_global_decl): Only generate a DIE for the
16727 abstract origin if it doesn't already exist or is a declaration DIE.
16728 (resolve_addr): Do not add the linkage name twice when
16729 generating a stub DIE for the DW_TAG_GNU_call_site target.
16731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16733 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16734 Use machine_mode rather than int for arg1_mode.
16736 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16739 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16740 that the second argument of the built-in functions to unpack
16741 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16742 switch statement instead a lot of if statements.
16743 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16744 Allow 64-bit values to be in Altivec registers as well as
16745 traditional floating point registers.
16746 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16748 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16750 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16751 MAX_REGS_PER_ADDRESS == 1.
16753 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16755 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16756 * config/i386/i386.c (ix86_option_override_internal): Simplify
16757 setting of opts->x_flag_entry.
16759 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16760 Jakub Jelinek <jakub@redhat.com>
16761 Richard Biener <rguenther@suse.de>
16763 PR tree-optimization/81503
16764 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16765 folded constant fits in the target type; reorder tests for clarity.
16767 2017-08-29 Martin Liska <mliska@suse.cz>
16769 * passes.def: Include pass_lower_switch.
16770 * stmt.c (dump_case_nodes): Remove and move to
16771 tree-switch-conversion.
16772 (case_values_threshold): Likewise.
16773 (expand_switch_as_decision_tree_p): Likewise.
16774 (emit_case_decision_tree): Likewise.
16775 (expand_case): Likewise.
16776 (balance_case_nodes): Likewise.
16777 (node_has_low_bound): Likewise.
16778 (node_has_high_bound): Likewise.
16779 (node_is_bounded): Likewise.
16780 (emit_case_nodes): Likewise.
16781 (struct simple_case_node): New struct.
16782 (add_case_node): Remove.
16783 (emit_case_dispatch_table): Use vector instead of case_list.
16784 (reset_out_edges_aux): Remove.
16785 (compute_cases_per_edge): Likewise.
16786 (expand_case): Build list of simple_case_node.
16787 (expand_sjlj_dispatch_table): Use it.
16788 * tree-switch-conversion.c (struct case_node): Moved from
16789 stmt.c and adjusted.
16790 (emit_case_nodes): Likewise.
16791 (node_has_low_bound): Likewise.
16792 (node_has_high_bound): Likewise.
16793 (node_is_bounded): Likewise.
16794 (case_values_threshold): Likewise.
16795 (reset_out_edges_aux): Likewise.
16796 (compute_cases_per_edge): Likewise.
16797 (add_case_node): Likewise.
16798 (dump_case_nodes): Likewise.
16799 (balance_case_nodes): Likewise.
16800 (expand_switch_as_decision_tree_p): Likewise.
16801 (emit_jump): Likewise.
16802 (emit_case_decision_tree): Likewise.
16803 (try_switch_expansion): Likewise.
16804 (do_jump_if_equal): Likewise.
16805 (emit_cmp_and_jump_insns): Likewise.
16806 (fix_phi_operands_for_edge): New function.
16807 (record_phi_operand_mapping): Likewise.
16808 (class pass_lower_switch): New pass.
16809 (pass_lower_switch::execute): New function.
16810 (make_pass_lower_switch): Likewise.
16811 (conditional_probability):
16812 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16813 * tree-pass.h: Add make_pass_lower_switch.
16815 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16818 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16821 2017-08-29 Richard Biener <rguenther@suse.de>
16823 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16824 we do not add a DW_AT_inline attribute twice.
16825 (gen_subprogram_die): Remove code setting DW_AT_inline on
16826 DECL_ABSTRACT_P nodes.
16828 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16830 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16831 calls to internal functions.
16832 (gimplify_modify_expr): Likewise.
16833 * tree-call-cdce.c (use_internal_fn): Likewise.
16834 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16835 (convert_to_divmod): Set the nothrow flag.
16836 * tree-if-conv.c (predicate_mem_writes): Likewise.
16837 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16838 (vectorizable_call): Likewise.
16839 (vectorizable_store): Likewise.
16840 (vectorizable_load): Likewise.
16841 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16842 (vect_recog_mask_conversion_pattern): Likewise.
16844 2017-08-29 Martin Liska <mliska@suse.cz>
16847 * gcc.c (driver_handle_option): Add new argument.
16848 * opts-common.c (handle_option): Pass
16849 target_option_override_hook.
16850 * opts-global.c (lang_handle_option): Add new option.
16851 (set_default_handlers): Add new argument.
16852 (decode_options): Likewise.
16853 * opts.c (target_handle_option): Likewise.
16854 (common_handle_option): Call target_option_override_hook.
16855 * opts.h (struct cl_option_handler_func): Add hook for
16856 target option override.
16857 (struct cl_option_handlers): Likewise.
16858 (set_default_handlers): Add new argument.
16859 (decode_options): Likewise.
16860 (common_handle_option): Likewise.
16861 (target_handle_option): Likewise.
16862 * toplev.c (toplev::main): Pass targetm.target_option.override
16865 2017-08-29 Richard Biener <rguenther@suse.de>
16866 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16868 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16869 life to the active subtree.
16871 2017-08-28 Jeff Law <law@redhat.com>
16873 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16874 derive_equivalences.
16875 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16876 Code moved into....
16877 (edge_info::derive_equivalences): New private member function
16879 * tree-ssa-dom.c (class edge_info): Changed from a struct
16880 to a class. Add ctor/dtor, methods and data members.
16881 (edge_info::edge_info): Renamed from allocate_edge_info.
16882 Initialize additional members.
16883 (edge_info::~edge_info): New.
16884 (free_dom_edge_info): Delete the edge info.
16885 (record_edge_info): Use new class & associated member functions.
16886 Tighten forms for testing for edge equivalences.
16887 (record_temporary_equivalences): Iterate over the simple
16888 equivalences rather than assuming there's only one per edge.
16889 (cprop_into_successor_phis): Iterate over the simple
16890 equivalences rather than assuming there's only one per edge.
16891 (optimize_stmt): Use operand_equal_p rather than pointer
16892 equality for mini-DSE code.
16894 2017-08-28 Nathan Sidwell <nathan@acm.org>
16896 * gcc.c (execute): Fold SIGPIPE handling into switch
16897 statement. Adjust internal error message.
16899 2017-08-28 Richard Biener <rguenther@suse.de>
16902 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16903 Do nothing for removed DIEs.
16905 2017-08-28 Richard Biener <rguenther@suse.de>
16907 PR tree-optimization/81977
16908 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16911 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16914 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16915 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16916 using targetm.gen_mem_thread_fence.
16918 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16921 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16922 predicate to register_operand. Reorder operands.
16923 (*btr<mode>): Ditto.
16924 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16925 (*btr<mode>_mask): Ditto.
16927 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16929 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16930 * config/rs6000/xmmintrin.h: New file.
16931 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16933 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16936 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16937 parameter and_insn and return it.
16938 (recombine_lvx_pattern): Insert a copy to ensure availability of
16939 the base register of the copied masking operation at the point of
16940 the instruction replacement.
16941 (recombine_stvx_pattern): Likewise.
16943 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16945 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16946 undocumented switches.
16947 (-mpower9-dform-vector): Likewise.
16948 (-mpower9-dform): Likewise.
16949 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16950 comments to delete references to -mpower9-dform* switches.
16951 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16952 Delete reference to -mpower9-dform* switches, test for
16953 -mpower9-vector instead.
16954 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16955 (OTHER_P9_VECTOR_MASKS): Likewise.
16956 (POWERPC_MASKS): Likewise.
16957 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16958 tests against -mpower9-dform* to -mpower9-vector. Delete code
16959 that checked for -mpower9-dform* consistancy with other options.
16960 Add test for -mpower9-misc to enable other power9 switches.
16961 (rs6000_init_hard_regno_mode_ok): Likewise.
16962 (rs6000_option_override_internal): Likewise.
16963 (rs6000_emit_prologue): Likewise.
16964 (rs6000_emit_epilogue): Likewise.
16965 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16966 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16967 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16969 (emit_fusion_p9_store): Likewise.
16970 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16971 resetting these macros if the assembler does not support ISA 3.0
16973 (TARGET_P9_DFORM_VECTOR): Likewise.
16974 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16975 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16977 2017-08-25 Alan Modra <amodra@gmail.com>
16979 PR rtl-optimization/81747
16980 * cse.c (cse_extended_basic_block): Don't attempt to record
16981 equivalences for degenerate conditional branches that branch
16982 to their fall-through.
16984 2017-08-24 Martin Sebor <msebor@redhat.com>
16986 PR middle-end/81908
16987 * gimple-fold.c (size_must_be_zero_p): New function.
16988 (gimple_fold_builtin_memory_op): Call it.
16990 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16992 * config/rs6000/mm_malloc.h: New file.
16994 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16996 PR tree-optimization/81913
16997 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16998 analysis when either IVs in condition can wrap.
17000 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17002 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17003 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17005 2017-08-24 Richard Biener <rguenther@suse.de>
17008 * targhooks.c (default_target_can_inline_p): Properly
17009 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17010 is present and always compare.
17011 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17012 infer -mfpmath=sse from TARGET_SSE_P.
17013 (ix86_can_inline_p): Properly use target_option_default_node when
17014 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17016 2017-08-24 Richard Biener <rguenther@suse.de>
17019 * dwarf2out.c (output_die): Handle flag_generate_offload like
17021 (output_comp_unit): Likewise.
17022 (gen_array_type_die): Likewise.
17023 (dwarf2out_early_finish): Likewise.
17024 (note_variable_value_in_expr): Likewise.
17025 (dwarf2out_finish): Likewise. Adjust assert.
17026 * cgraphunit.c (symbol_table::compile): Move setting of
17027 flag_generate_offload earlier ...
17028 (symbol_table::finalize_compilation_unit): ... here, before
17029 early debug finalization.
17031 2017-08-24 Richard Biener <rguenther@suse.de>
17033 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17034 and ipa-fnsummary.h.
17035 (ix86_can_inline_p): When ix86_fpmath flags do not match
17036 check whether the callee uses FP math at all.
17038 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17040 PR middle-end/81931
17041 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17042 instead of TYPE_PRECISION.
17044 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17045 Alan Hayward <alan.hayward@arm.com>
17046 David Sherwood <david.sherwood@arm.com>
17048 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17050 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17052 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17054 (ix86_frame::stack_realign_allocate): New field.
17055 (struct machine_frame_state): Modify comments.
17056 (machine_frame_state::sp_realigned_fp_end): New field.
17057 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17058 layout calculation.
17059 (sp_valid_at): Add assertion to assure no attempt to access invalid
17060 offset of a realigned stack.
17061 (fp_valid_at): Likewise.
17062 (choose_baseaddr): Modify comments.
17063 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17064 ix86_expand_prologue.
17065 (ix86_expand_prologue): Modify stack realignment and allocation.
17066 (ix86_expand_epilogue): Modify comments.
17067 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17068 avx2_runtime, avx512f, and avx512f_runtime.
17070 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17072 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17073 (mstackrealign): Do not init to -1.
17074 * config/i386/i386.c (ix86_option_override_internal):
17075 Check opts_set, not opts when setting default value of
17076 opts->x_ix86_force_align_arg_pointer.
17078 2017-08-23 Richard Biener <rguenther@suse.de>
17080 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17081 lang_hooks.decl_printable_name.
17082 * print-rtl-function.c (print_rtx_function): Likewise.
17083 * tree-pretty-print.c (dump_function_header): Likewise.
17085 2017-08-23 Richard Biener <rguenther@suse.de>
17088 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17089 -g0 at compile-time.
17091 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17093 PR middle-end/19706
17094 * doc/sourcebuild.texi (Other hardware attributes):
17097 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17099 PR middle-end/19706
17100 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17101 Fix single-use check.
17103 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17105 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17107 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17109 * doc/install.texi: Modify to add more details on running selected
17112 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17114 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17115 is combined with -mabi=ms.
17116 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17119 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17121 PR tree-optimization/81488
17122 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17123 and cached_basis fields.
17124 (MAX_SPREAD): New constant.
17125 (alloc_cand_and_find_basis): Initialize new fields.
17126 (clear_visited): New function.
17127 (create_phi_basis_1): Rename from create_phi_basis, set visited
17128 and cached_basis fields.
17129 (create_phi_basis): New wrapper function.
17130 (phi_add_costs_1): Rename from phi_add_costs, add spread
17131 parameter, set visited field, short-circuit when limits reached.
17132 (phi_add_costs): New wrapper function.
17133 (record_phi_increments_1): Rename from record_phi_increments, set
17135 (record_phi_increments): New wrapper function.
17136 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17137 (phi_incr_cost): New wrapper function.
17138 (all_phi_incrs_profitable_1): Rename from
17139 all_phi_incrs_profitable, set visited field.
17140 (all_phi_incrs_profitable): New wrapper function.
17142 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17143 Alan Hayward <alan.hayward@arm.com>
17144 David Sherwood <david.sherwood@arm.com>
17146 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17147 that takes the outer and inner modes.
17148 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17149 comparison as the canonical test for a paradoxical subreg.
17150 * combine.c (simplify_set): Use paradoxical_subreg_p.
17151 (make_extraction): Likewise.
17152 (force_to_mode): Likewise.
17153 (rtx_equal_for_field_assignment_p): Likewise.
17154 (gen_lowpart_for_combine): Likewise.
17155 (simplify_comparison): Likewise.
17156 * cse.c (equiv_constant): Likewise.
17157 * expmed.c (store_bit_field_1): Likewise.
17158 * final.c (alter_subreg): Likewise.
17159 * fwprop.c (propagate_rtx): Likewise.
17160 (forward_propagate_subreg): Likewise.
17161 * ira-conflicts.c (ira_build_conflicts): Likewise.
17162 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17163 * lra-constraints.c (curr_insn_transform): Likewise.
17164 (split_reg): Likewise.
17165 * lra-eliminations.c (move_plus_up): Likewise.
17166 (lra_eliminate_regs_1): Likewise.
17167 * recog.c (general_operand): Likewise.
17168 * ree.c (combine_reaching_defs): Likewise.
17169 * reload.c (push_reload): Likewise.
17170 (find_reloads): Likewise.
17171 * reload1.c (elimination_effects): Likewise.
17172 (compute_reload_subreg_offset): Likewise.
17173 (choose_reload_regs): Likewise.
17174 * rtlanal.c (subreg_lsb_1): Likewise.
17175 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17176 (simplify_subreg): Likewise.
17177 * var-tracking.c (track_loc_p): Likewise.
17178 * emit-rtl.c (byte_lowpart_offset): Likewise.
17179 (paradoxical_subreg_p): Delete out-of-line definition.
17181 2017-08-22 Jeff Law <law@redhat.com>
17183 PR tree-optimization/81741
17184 PR tree-optimization/71947
17185 * tree-ssa-dom.c: Include tree-inline.h.
17186 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17187 equivalences if one is more expensive to compute than the other.
17188 * tree-ssa-scopedtables.h (class const_or_copies): Make
17189 record_const_or_copy_raw method private.
17190 (class avail_exprs_stack): New method simplify_binary_operation.
17191 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17192 avail_exprs_stack::simplify_binary_operation as needed.
17193 (avail_exprs_stack::simplify_binary_operation): New function.
17195 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17197 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17198 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17199 (DOT_SYMBOLS): Likewise.
17200 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17201 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17202 (RS6000_ABI_NAME): Likewise.
17203 (TARGET_CMODEL): Likewise.
17204 (TOC_SECTION_ASM_OP): Likewise.
17205 (SET_CMODEL): New macro.
17206 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17208 2017-08-22 Richard Biener <rguenther@suse.de>
17210 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17211 to allow for free-lang-data replacements similar to verify_type_variant.
17213 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17214 Alan Hayward <alan.hayward@arm.com>
17215 David Sherwood <david.sherwood@arm.com>
17217 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17218 VOIDmode for the LABEL_REF.
17220 2017-08-22 Richard Biener <rguenther@suse.de>
17222 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17224 2017-08-22 Richard Biener <rguenther@suse.de>
17226 * common.opt (feliminate-dwarf2-dups): Ignore.
17227 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17228 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17229 same_die_p_wrap, compute_section_prefix,
17230 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17231 (comdat_symbol_id, comdat_symbol_number): Likewise.
17232 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17234 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17235 (output_die): Mark unreachable path unreachable.
17236 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17237 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17238 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17239 (dwarf2out_early_finish): Likewise.
17241 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17243 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17245 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17248 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17249 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17250 (avr_attribute_table) <io, io_low, address>: Initialize
17251 .decl_required with true.
17253 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17255 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17256 undocumented debugging options.
17257 (-mvsx-scalar-double): Likewise.
17258 (-mallow-df-permute): Likewise.
17259 (-mvectorize-builtins): Likewise.
17260 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17261 (rs6000_builtin_vectorized_function): Likewise.
17262 (rs6000_builtin_md_vectorized_function): Likewise.
17263 (rs6000_opt_vars): Likewise.
17265 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17268 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17269 (*btrq_imm): Rename from *btrq.
17270 (*btcq_imm): Rename from *btcq.
17271 (btsc): New code attribute.
17272 (*<btsc><mode>): New insn pattern.
17273 (*btr<mode>): Ditto.
17274 (*<btsc><mode>_mask): New insn_and_split pattern.
17275 (*btr<mode>_mask): Ditto.
17277 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17278 Alan Hayward <alan.hayward@arm.com>
17279 David Sherwood <david.sherwood@arm.com>
17281 * function.c (pad_below): Simplify padding calculation.
17283 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17284 Alan Hayward <alan.hayward@arm.com>
17285 David Sherwood <david.sherwood@arm.com>
17287 * target.def (function_prologue): Remove frame size argument.
17288 (function_epilogue): Likewise.
17289 * doc/tm.texi: Regenerate.
17290 * final.c (final_start_function): Update call to function_prologue.
17291 (final_end_function): Update call to function_epilogue.
17292 (default_function_pro_epilogue): Remove frame size argument.
17293 * output.h (default_function_pro_epilogue): Likewise.
17294 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17295 (arm_output_function_prologue): Likewise.
17296 * config/frv/frv.c (frv_function_prologue): Likewise.
17297 (frv_function_epilogue): Likewise.
17298 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17299 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17300 (ia64_output_function_epilogue): Likewise.
17301 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17302 (m32r_output_function_epilogue): Likewise.
17303 * config/microblaze/microblaze.c (microblaze_function_prologue)
17304 (microblaze_function_epilogue): Likewise.
17305 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17306 (mips_output_function_epilogue): Likewise.
17307 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17308 (mmix_target_asm_function_epilogue): Likewise.
17309 * config/msp430/msp430.c (msp430_start_function): Likewise.
17310 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17311 (nds32_asm_function_epilogue): Likewise.
17312 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17313 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17314 (pa_output_function_epilogue): Likewise.
17315 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17316 (rs6000_output_function_epilogue): Likewise.
17317 * config/rl78/rl78.c (rl78_start_function): Likewise.
17318 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17319 (rs6000_output_function_epilogue): Likewise.
17320 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17321 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17322 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17323 (sparc_asm_function_epilogue): Likewise.
17325 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17327 * tree.h (type_has_mode_precision_p): New function.
17328 * convert.c (convert_to_integer_1): Use it.
17329 * expr.c (expand_expr_real_2): Likewise.
17330 (expand_expr_real_1): Likewise.
17331 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17332 * match.pd: Likewise.
17333 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17334 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17335 * tree-tailcall.c (process_assignment): Likewise.
17336 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17337 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17338 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17339 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17340 (vectorizable_assignment): Likewise.
17341 (vectorizable_shift): Likewise.
17342 (vectorizable_operation): Likewise.
17343 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17345 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17347 * match.pd: Add pow (C, x) simplification.
17349 2017-08-21 Richard Biener <rguenther@suse.de>
17351 PR tree-optimization/81900
17352 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17353 for blocks with abnormal predecessors.
17354 (compute_antic): Do not set visited flag prematurely.
17356 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17359 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17361 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17363 * stor-layout.h (vector_type_mode): Move to...
17364 * tree.h (vector_type_mode): ...here.
17365 * stor-layout.c (vector_type_mode): Move to...
17366 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17368 2017-08-21 Richard Biener <rguenther@suse.de>
17370 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17371 register_external_die hooks.
17372 (debug_false_tree_charstarstar_uhwistar): Declare.
17373 (debug_nothing_tree_charstar_uhwi): Likewise.
17374 * debug.c (do_nothing_debug_hooks): Adjust.
17375 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17376 (debug_nothing_tree_charstar_uhwi): Likewise.
17377 * dbxout.c (dbx_debug_hooks): Adjust.
17378 (xcoff_debug_hooks): Likewise.
17379 * sdbout.c (sdb_debug_hooks): Likewise.
17380 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17381 * dwarf2out.c (macinfo_label_base): New global.
17382 (dwarf2out_register_external_die): New function for the
17383 register_external_die hook.
17384 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17385 (dwarf2_debug_hooks): Use them.
17386 (dwarf2_lineno_debug_hooks): Adjust.
17387 (struct die_struct): Add with_offset flag.
17388 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17389 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17390 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17391 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17392 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17393 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17394 defining section names for the early LTO debug variants.
17395 (reset_indirect_string): New helper.
17396 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17397 (print_dw_val): Add support for offsetted symbol references.
17398 (get_ultimate_context): Split out from is_cxx.
17399 (is_cxx): Use get_ultimate_context.
17400 (is_fortran): Add decl overload.
17401 (compute_comp_unit_symbol): Split out worker from
17402 compute_section_prefix.
17403 (compute_section_prefix): Call compute_comp_unit_symbol and
17404 set comdat_type_p here.
17405 (output_die): Skip DIE symbol output for the LTO added one.
17406 Handle DIE symbol references with offset.
17407 (output_comp_unit): Guard section name mangling properly.
17408 For LTO debug sections emit a symbol at the section beginning
17409 which we use to refer to its DIEs.
17410 (add_abstract_origin_attribute): For DIEs registered via
17411 dwarf2out_register_external_die directly refer to the early
17412 DIE rather than indirectly through the shadow one we created.
17413 Remove obsolete call to dwarf2out_abstract_function for
17414 non-function/block origins.
17415 (gen_array_type_die): When generating early LTO debug do
17416 not emit DW_AT_string_length.
17417 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17418 late when in LTO. As suggested place a gcc_unreachable for
17419 the DECL_ABSTRACT_P case.
17420 (gen_subprogram_die): Avoid another specification DIE
17421 for early built declarations/definitions for the late LTO case.
17422 (gen_variable_die): Add type references for late duplicated VLA dies
17424 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17425 we have the abstract instance already.
17426 (process_scope_var): Adjust decl DIE contexts in LTO which
17427 first puts them in limbo.
17428 (gen_decl_die): Do not generate type DIEs late apart from
17429 types for VLAs or for decls we do not yet have a DIE. Do not
17430 call dwarf2out_abstract_function late.
17431 (dwarf2out_early_global_decl): Make sure to create DIEs
17432 for abstract instances of a decl first.
17433 (dwarf2out_late_global_decl): Adjust comment.
17434 (output_macinfo_op): With multiple macro sections use
17435 macinfo_label_base to distinguish labels.
17436 (output_macinfo): Likewise. Update macinfo_label_base.
17437 Pass in the line info label.
17438 (note_variable_value_in_expr): When generating LTO resolve
17439 all variable values here by generating DIEs as needed.
17440 (init_sections_and_labels): Add early LTO debug flag parameter
17441 and generate different sections and names if set. Add generation
17442 counter for the labels so we can have multiple of them.
17443 (reset_dies): Helper to allow DIEs to be output multiple times.
17444 (dwarf2out_finish): When outputting DIEs to the fat part of an
17445 LTO object first reset DIEs.
17446 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17447 (modified_type_die): Check for decl_ultimate_origin being self
17449 (gen_type_die_with_usage): Likewise.
17450 (gen_typedef_die): Allow decl_ultimate_origin being self.
17451 (set_decl_abstract_flags): Remove.
17452 (set_block_abstract_flags): Likewise.
17453 (dwarf2out_abstract_function): Treat the early generated DIEs
17454 as the abstract copy and only add DW_AT_inline and
17455 DW_AT_artificial here and call set_decl_origin_self.
17456 If the DIE has an abstract origin don't do anything.
17457 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17458 if we have none yet (Go fails to build one, PR78628).
17459 (variably_modified_type_p): Prevent endless recursion for Ada
17460 cyclic pointer types.
17461 * lto-streamer-in.c: Include debug.h.
17462 (dref_queue): New global.
17463 (lto_read_tree_1): Stream in DIE references.
17464 (lto_input_tree): Register DIE references.
17465 (input_function): Stream DECL_DEBUG_ARGS.
17466 * lto-streamer-out.c: Include debug.h.
17467 (lto_write_tree_1): Output DIE references.
17468 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17469 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17470 (output_function): Stream DECL_DEBUG_ARGS.
17471 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17472 Stream DECL_ABSTRACT_ORIGIN.
17473 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17474 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17475 DECL_CONTEXT for file-scope decls.
17476 * lto-streamer.h (struct dref_entry): Declare.
17477 (dref_queue): Likewise.
17478 * cfgexpand.c (pass_expand::execute): Do not call the
17479 outlining_inline_function hook here.
17480 * lto-wrapper.c (debug_obj): New global.
17481 (tool_cleanup): Unlink it if required.
17482 (debug_objcopy): New function.
17483 (run_gcc): Handle early debug sections in the IL files by
17484 extracting them to separate files, partially linkin them and
17485 feeding the result back as result to the linker.
17486 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17487 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17488 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17489 sections into a separate segment.
17490 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17492 (darwin_asm_dwarf_section): Likewise.
17493 (darwin_asm_output_dwarf_offset): Likewise.
17494 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17496 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17497 Alan Hayward <alan.hayward@arm.com>
17498 David Sherwood <david.sherwood@arm.com>
17500 * read-md.h (md_reader::record_potential_iterator_use): Replace
17501 pointer argument with an rtx and an index.
17502 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17503 (apply_mode_iterator): Likewise.
17504 (apply_code_iterator): Likewise.
17505 (apply_int_iterator): Likewise.
17506 (apply_subst_iterator): Likewise.
17507 (record_iterator_use): Likewise.
17508 (record_attribute_use): Likewise.
17509 (md_reader::record_potential_iterator_use): Likewise. Update calls
17510 to record_iterator_use and apply_iterator.
17511 (iterator_use): Replace ptr with x and index.
17512 (attribute_use): Likewise.
17513 (apply_attribute_uses): Update calls to apply_iterator.
17514 (apply_iterators): Likewise. Update initialization of iterator_use.
17515 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17516 and record_potential_iterator_use.
17517 (rtx_reader::read_rtx_operand): Likewise.
17519 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17520 Alan Hayward <alan.hayward@arm.com>
17521 David Sherwood <david.sherwood@arm.com>
17523 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17526 2017-08-21 Richard Biener <rguenther@suse.de>
17528 PR middle-end/81884
17529 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17530 at struct end conservatively when comparing common bases.
17532 2017-08-21 Richard Biener <rguenther@suse.de>
17534 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17535 (mem_ref_in_stmt): Remove.
17536 (determine_max_movement): Use ref index to get at the reference.
17537 (invariantness_dom_walker::before_dom_children): Deal with
17538 lim data already initialized.
17539 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17541 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17543 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17544 (TARGET_ISA_ROUND): Ditto.
17545 (TARGET_ROUND): Ditto.
17546 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17547 * config/i386/i386.md: Ditto.
17548 * config/i386/sse.md: Ditto.
17549 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17550 with OPTION_MASK_ISA_SSE4_1.
17552 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17555 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17556 __builtin_ia32_lzcnt_u16.
17558 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17561 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17562 (rs6000_set_current_function): Rewrite function to use it.
17564 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17567 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17568 and TYPE_WARN_IF_NOT_ALIGN.
17569 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17570 (handle_warn_if_not_align): New.
17571 (place_union_field): Call handle_warn_if_not_align.
17572 (place_field): Call handle_warn_if_not_align.
17573 Copy TYPE_WARN_IF_NOT_ALIGN.
17574 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17575 (layout_type): Likewise.
17576 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17578 (tree_decl_common): Add warn_if_not_align.
17579 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17580 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17581 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17582 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17583 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17584 * doc/extend.texi: Document warn_if_not_aligned attribute.
17585 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17587 2017-08-17 Martin Liska <mliska@suse.cz>
17590 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17591 (get_data_dependence): Use it as pointer type.
17592 (distribute_loop): Likewise.
17594 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17596 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17597 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17598 (p8_vmrgow_<mode>_direct): New define_insn.
17599 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17600 handle endianness for vmrgew and vmrgow permute patterns.
17602 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17604 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17605 * config/rs6000/rs6000-cpus.def: Remove comment.
17606 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17607 (POWERPC_MASKS): Likewise.
17608 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17609 use of TARGET_VSX_TIMODE.
17610 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17611 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17612 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17613 (rs6000_option_override_internal): Remove dead code.
17614 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17615 (rs6000_legitimize_reload_address): Likewise.
17616 (rs6000_legitimate_address_p): Likewise.
17617 (rs6000_opt_masks): Delete "vsx-timode".
17618 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17619 from function comment.
17620 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17621 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17622 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17624 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17625 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17626 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17627 (VSX_TI): Likewise.
17629 (define_peephole2): Likewise.
17631 2017-08-17 Martin Sebor <msebor@redhat.com>
17634 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17635 past the end of an array.
17636 (test_pp_format): Add test cases.
17638 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17640 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17641 missing ECF_NOTHROW flags.
17643 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17646 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17647 operands residing in integer registers.
17648 (*vsx_le_perm_load_<mode>): Likewise.
17649 (*vsx_le_perm_store_<mode>): Likewise.
17650 (define_peephole2): Add peepholes to optimize the above.
17652 2017-08-17 Marek Polacek <polacek@redhat.com>
17654 PR middle-end/81814
17655 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17656 to mimic what shorten_compare did. Change the return type to bool.
17657 (fold_cond_expr_with_comparison): Update call to
17658 operand_equal_for_comparison_p.
17659 (fold_ternary_loc): Likewise.
17661 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17663 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17665 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17667 2017-08-17 Richard Biener <rguenther@suse.de>
17669 * tree-ssa-structalias.c (solve_graph): When propagating
17670 to successors update the graphs succ edges and avoid duplicate work.
17672 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17675 * config/i386/i386.c (ix86_option_override_internal): Save target
17676 specific options after ix86_stack_protector_guard_reg was changed.
17678 2017-08-17 Richard Biener <rguenther@suse.de>
17680 PR tree-optimization/81827
17681 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17682 (new_var_info): Initialize it conservatively.
17683 (get_call_vi): Mark register vars.
17684 (new_scalar_tmp_constraint_exp): Likewise.
17685 (handle_rhs_call): Likewise.
17686 (handle_const_call): Likewise.
17687 (create_function_info_for): Likewise.
17688 (solve_constraints): Sort varinfos to separate register from
17689 non-register vars to pack points-to solution bitmaps during
17692 2017-08-17 Marek Polacek <polacek@redhat.com>
17694 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17696 2017-08-17 Richard Biener <rguenther@suse.de>
17698 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17699 to true when overflow is undefined and we saturated the result.
17701 2017-08-17 Alan Modra <amodra@gmail.com>
17704 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17705 Don't use store multiple if only one reg needs saving.
17706 (interesting_frame_related_regno): New function.
17707 (rs6000_frame_related): Don't emit frame info for regs that
17709 (rs6000_emit_epilogue): Likewise.
17711 2017-08-16 Nathan Sidwell <nathan@acm.org>
17713 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17714 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17715 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17716 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17717 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17718 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17719 (verify_type): Adjust for TYPE_BINFO move.
17720 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17721 process TYPE_BINFO directly.
17722 (hash_tree): Likewise.
17723 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17725 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17728 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17730 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17732 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17735 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17736 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17737 Add ix86_binary_operator_ok to insn constraint.
17738 (*iordi_1_bts): Ditto.
17739 (*xordi_1_btc): Ditto.
17740 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17741 Update corresponding peephole2 pattern.
17745 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17747 PR tree-optimization/81832
17748 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17749 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17751 2017-08-16 Marek Polacek <polacek@redhat.com>
17754 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17755 regarding size_zero_node.
17757 2017-08-16 Martin Liska <mliska@suse.cz>
17760 * config.gcc: Respect previously set extra_objs in case
17763 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17765 PR tree-optimization/81835
17766 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17767 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17768 not depend on the phi.
17770 2017-08-16 Alan Modra <amodra@gmail.com>
17772 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17775 2017-08-16 Alan Modra <amodra@gmail.com>
17777 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17778 (save_reg_p): ..into this. Update all callers.
17779 (first_reg_to_save): Simplify.
17781 2017-08-16 Alan Modra <amodra@gmail.com>
17783 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17786 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17790 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17791 constant count if that count is less than 32.
17793 2017-08-15 Nathan Sidwell <nathan@acm.org>
17795 * gcc.c (execute): Emit friendlier message if inferior is killed
17796 by an external cause.
17798 2017-08-15 Richard Biener <rguenther@suse.de>
17800 PR tree-optimization/81790
17801 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17802 CONSTRUCTORs from simplifying and VN.
17804 2017-08-14 Martin Sebor <msebor@redhat.com>
17806 * builtin-attrs.def: Add comments.
17808 2017-08-14 Martin Sebor <msebor@redhat.com>
17811 * doc/extend.texi (attribute nonstring): Document new attribute.
17813 2017-08-14 Martin Sebor <msebor@redhat.com>
17816 * tree-diagnostic.c (default_tree_printer): Handle %G.
17817 * gimple-pretty-print.h (percent_G_format): Declare new function.
17818 * gimple-pretty-print.c (percent_G_format): Define.
17819 * tree-pretty-print.c (percent_K_format): Add argument.
17821 2017-08-14 Martin Sebor <msebor@redhat.com>
17823 PR translation/79998
17824 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17825 Remove a stray space.
17827 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17830 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17831 (*iordi_1_bts): Ditto.
17832 (*xordi_1_btc): Ditto.
17834 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17837 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17838 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17840 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17841 quoted strings, and make more translator-friendly.
17842 (darwin_rs6000_override_options): Likewise.
17843 (rs6000_option_override_internal): Likewise.
17844 (rs6000_return_in_memory): Fix overlong line.
17845 (init_cmulative_args): Use quoted strings, and make more
17846 translator-friendly.
17847 (rs6000_pass_by_reference): Fix overlong line.
17848 (def_builtin): Use quoted strings.
17849 (altivec_expand_predicate_builtin): Use quoted strings, and make
17850 more translator-friendly.
17851 (htm_expand_builtin): Use quoted strings.
17852 (cpu_expand_builtin): Use quoted strings, and make more
17853 translator-friendly.
17854 (altivec_expand_builtin): Likewise.
17855 (paired_expand_predicate_builtin): Likewise.
17856 (rs6000_invalid_builtin): Likewise.
17857 (builtin_function_type): Use quoted strings.
17858 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17859 more translator-friendly.
17860 (rs6000_trampoline_init): Likewise.
17861 (rs6000_handle_altivec_attribute): Likewise.
17862 (rs6000_inner_target_options): Use quoted strings.
17863 (rs6000_disable_incompatible_switches): Likewise.
17864 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17865 strings, and make more translator-friendly.
17866 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17868 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17870 PR tree-optimization/81799
17871 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17872 cond_expr to simple gimple operand.
17874 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17876 PR middle-end/46932
17877 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17879 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17883 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17884 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17885 TARGET_GASISR_PROLOGUES.
17886 * config/avr/avr.c (avr_option_override): Same.
17887 (avr_pass_pre_proep::execute): Same.
17889 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17892 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17893 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17895 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17897 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17898 define_insn_and_split. Split to a memory load from 0 in
17899 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17900 using PTR mode iterator.
17901 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17902 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17903 (*add_tp_<mode>): Redefine as define_insn_and_split.
17904 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17905 address space. Merge with *add_tp_x32 using PTR mode iterator.
17906 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17907 Split to an add with a memory load from 0 in
17908 DEFAULT_TLS_SEG_REG address space.
17910 2017-08-12 Andrew Pinski <apinski@cavium.com>
17912 * config/aarch64/aarch64-option-extensions.def (rdma):
17913 Fix feature string to what Linux prints out in /proc/cpuinfo.
17915 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17918 * dwarf2out.c (modified_type_die): For C typedef types that have
17919 an ultimate origin, process the ultimate origin instead of the
17921 (gen_typedef_die): Assert that input DECLs have no ultimate
17923 (gen_type_die_with_usage): For typedef variants that have an
17924 ultimate origin, just call gen_decl_die on the original DECL.
17925 (process_scope_var): Avoid creating DIEs for local typedefs and
17926 concrete static variables.
17928 2017-08-12 Alan Modra <amodra@gmail.com>
17932 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17933 match gnu-user.h startfile.
17934 (ENDFILE_LINUX_SPEC): Similarly.
17936 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17939 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17941 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17943 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17944 * config/aarch64/aarch64.md (mov<mode>): Change.
17945 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17946 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17947 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17949 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17951 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17952 for storage order barriers.
17954 2017-08-11 Martin Liska <mliska@suse.cz>
17957 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17958 variables of void type.
17960 2017-08-11 Martin Liska <mliska@suse.cz>
17962 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17963 TARGET_SUPPORTS_ALIASES.
17964 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17965 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17966 (optimize_weakref): Likewise.
17967 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17968 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17969 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17971 2017-08-11 Martin Liska <mliska@suse.cz>
17974 * config/i386/i386.c (make_resolver_func): Do complete
17975 refactoring of the function.
17977 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17980 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17981 * config/i386/i386.c (ix86_stack_protect_guard): Use
17982 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17983 * doc/invoke.texi (x86 Options): Document
17984 -mstack-protector-guard-symbol= option.
17986 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17988 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17989 * config/i386/i386.c (ix86_split_stack_guard): New function.
17990 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17991 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17992 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17993 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17994 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17995 (split_stack_space_check): Call ix86_split_stack_guard.
17997 2017-08-10 Martin Sebor <msebor@redhat.com>
17999 * print-tree.c (print_node): Print location using the established
18001 Replace spaces with colons.
18002 (debug_raw, debug): Ditto.
18004 2017-08-10 Martin Sebor <msebor@redhat.com>
18007 * pretty-print.c (pp_format): Correct the handling of %s precision.
18009 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18012 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18014 (ix86_finalize_stack_frame_flags): This. Also clear
18015 frame_pointer_needed if -fno-omit-frame-pointer is used without
18017 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18018 with ix86_finalize_stack_frame_flags.
18019 (ix86_expand_epilogue): Likewise.
18020 (ix86_expand_split_stack_prologue): Likewise.
18021 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18023 2017-08-10 Martin Liska <mliska@suse.cz>
18026 * c-attribs.c (handle_target_attribute):
18027 Report warning for an empty string argument of target attribute.
18029 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18032 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18034 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18035 or DECL_NONLOCAL labels.
18036 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18037 or DECL_NONLOCAL labels here.
18039 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18041 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18042 to indicate when early gimple folding has been disabled.
18043 (rs6000_gimple_fold_builtin): Add debug content.
18044 (rs6000_invalid_builtin): Fix whitespace.
18045 (rs6000_expand_builtin): Fix whitespace.
18046 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18048 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18051 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18052 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18054 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18056 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18057 (qdf24xx): Likewise.
18058 * config/aarch64/aarch64.md: Include falkor.md.
18059 * config/aarch64/falkor.md: New.
18061 2017-08-09 Marek Polacek <polacek@redhat.com>
18064 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18065 * diagnostic.c (emit_diagnostic): Add a comment.
18066 (emit_diagnostic_valist): New function.
18068 2017-08-09 Marek Polacek <polacek@redhat.com>
18071 * input.c (make_location): New overload.
18072 * input.h (make_location): Declare.
18074 2017-08-08 Alan Modra <amodra@gmail.com>
18075 H.J. Lu <hongjiu.lu@intel.com>
18078 * gcc.c (NO_PIE_SPEC): Delete.
18079 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18081 (LINK_PIE_SPEC): ..to here.
18082 (LINK_COMMAND_SPEC): Support -no-pie.
18083 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18084 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18086 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18087 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18088 (ENDFILE_CRTEND_SPEC): Similarly.
18090 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18093 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18094 (mstack-protector-guard-offset=): Ditto.
18095 * config/i386/i386.c (ix86_option_override): Handle
18096 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18098 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18099 ix86_stack_protect_guard_offset variables.
18100 (TARGET_STACK_PROTECT_GUARD): Always define.
18101 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18102 and -mstack-protector-guard-offset= options.
18104 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18106 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18107 boundary case for the last candidate.
18109 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18111 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18112 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18114 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18116 PR middle-end/19706
18117 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18118 * config/aarch64/aarch64-builtins.c
18119 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18120 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18121 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18123 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18124 Andrew Pinski <pinskia@gmail.com>
18126 PR middle-end/19706
18127 * internal-fn.def (XORSIGN): New.
18128 * optabs.def (xorsign_optab): New.
18129 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18130 (convert_expand_mult_copysign): New.
18131 (pass_optimize_widening_mul::execute): Call
18132 convert_expand_mult_copysign.
18134 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18136 PR tree-optimization/81354
18137 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18138 Insert on edges rather than explicitly creating landing pads.
18139 (analyze_candidates_and_replace): Commit edge inserts.
18141 2017-08-08 Richard Biener <rguenther@suse.de>
18143 PR middle-end/81719
18144 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18145 (expand_simple_operations): Also look through ADDR_EXPRs with
18146 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18148 2017-08-08 Richard Biener <rguenther@suse.de>
18150 PR tree-optimization/81723
18151 * tree-vect-slp.c (struct bst_traits): New hash traits.
18152 (bst_fail): New global.
18153 (vect_build_slp_tree_2): New worker, split out from ...
18154 (vect_build_slp_tree): ... this now wrapping it with using
18155 bst_fail set to cache SLP tree build fails. Properly handle
18157 (vect_analyze_slp_instance): Allocate and free bst_fail.
18159 2017-08-08 Martin Liska <mliska@suse.cz>
18162 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18163 LABEL_DECLs that can be from a different function.
18165 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18167 PR tree-optimization/81744
18168 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18169 loop's number of iterations.
18171 2017-08-08 Martin Liska <mliska@suse.cz>
18173 * asan.c: Include header files.
18174 * attribs.c (build_decl_attribute_variant): New function moved
18176 (build_type_attribute_qual_variant): Likewise.
18177 (cmp_attrib_identifiers): Likewise.
18178 (simple_cst_list_equal): Likewise.
18179 (omp_declare_simd_clauses_equal): Likewise.
18180 (attribute_value_equal): Likewise.
18181 (comp_type_attributes): Likewise.
18182 (build_type_attribute_variant): Likewise.
18183 (lookup_ident_attribute): Likewise.
18184 (remove_attribute): Likewise.
18185 (merge_attributes): Likewise.
18186 (merge_type_attributes): Likewise.
18187 (merge_decl_attributes): Likewise.
18188 (merge_dllimport_decl_attributes): Likewise.
18189 (handle_dll_attribute): Likewise.
18190 (attribute_list_equal): Likewise.
18191 (attribute_list_contained): Likewise.
18192 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18193 (lookup_attribute_by_prefix): Likewise.
18194 * bb-reorder.c: Include header files.
18195 * builtins.c: Likewise.
18196 * calls.c: Likewise.
18197 * cfgexpand.c: Likewise.
18198 * cgraph.c: Likewise.
18199 * cgraphunit.c: Likewise.
18200 * convert.c: Likewise.
18201 * dwarf2out.c: Likewise.
18202 * final.c: Likewise.
18203 * fold-const.c: Likewise.
18204 * function.c: Likewise.
18205 * gimple-expr.c: Likewise.
18206 * gimple-fold.c: Likewise.
18207 * gimple-pretty-print.c: Likewise.
18208 * gimple.c: Likewise.
18209 * gimplify.c: Likewise.
18210 * hsa-common.c: Likewise.
18211 * hsa-gen.c: Likewise.
18212 * internal-fn.c: Likewise.
18213 * ipa-chkp.c: Likewise.
18214 * ipa-cp.c: Likewise.
18215 * ipa-devirt.c: Likewise.
18216 * ipa-fnsummary.c: Likewise.
18217 * ipa-inline.c: Likewise.
18218 * ipa-visibility.c: Likewise.
18220 * lto-cgraph.c: Likewise.
18221 * omp-expand.c: Likewise.
18222 * omp-general.c: Likewise.
18223 * omp-low.c: Likewise.
18224 * omp-offload.c: Likewise.
18225 * omp-simd-clone.c: Likewise.
18226 * opts-global.c: Likewise.
18227 * passes.c: Likewise.
18228 * predict.c: Likewise.
18229 * sancov.c: Likewise.
18230 * sanopt.c: Likewise.
18231 * symtab.c: Likewise.
18232 * toplev.c: Likewise.
18233 * trans-mem.c: Likewise.
18234 * tree-chkp.c: Likewise.
18235 * tree-eh.c: Likewise.
18236 * tree-into-ssa.c: Likewise.
18237 * tree-object-size.c: Likewise.
18238 * tree-parloops.c: Likewise.
18239 * tree-profile.c: Likewise.
18240 * tree-ssa-ccp.c: Likewise.
18241 * tree-ssa-live.c: Likewise.
18242 * tree-ssa-loop.c: Likewise.
18243 * tree-ssa-sccvn.c: Likewise.
18244 * tree-ssa-structalias.c: Likewise.
18245 * tree-ssa.c: Likewise.
18246 * tree-streamer-in.c: Likewise.
18247 * tree-vectorizer.c: Likewise.
18248 * tree-vrp.c: Likewise.
18249 * tsan.c: Likewise.
18250 * ubsan.c: Likewise.
18251 * varasm.c: Likewise.
18252 * varpool.c: Likewise.
18253 * tree.c: Remove functions moved to attribs.[ch].
18254 * tree.h: Likewise.
18255 * config/aarch64/aarch64.c: Add attrs.h header file.
18256 * config/alpha/alpha.c: Likewise.
18257 * config/arc/arc.c: Likewise.
18258 * config/arm/arm.c: Likewise.
18259 * config/avr/avr.c: Likewise.
18260 * config/bfin/bfin.c: Likewise.
18261 * config/c6x/c6x.c: Likewise.
18262 * config/cr16/cr16.c: Likewise.
18263 * config/cris/cris.c: Likewise.
18264 * config/darwin.c: Likewise.
18265 * config/epiphany/epiphany.c: Likewise.
18266 * config/fr30/fr30.c: Likewise.
18267 * config/frv/frv.c: Likewise.
18268 * config/ft32/ft32.c: Likewise.
18269 * config/h8300/h8300.c: Likewise.
18270 * config/i386/winnt.c: Likewise.
18271 * config/ia64/ia64.c: Likewise.
18272 * config/iq2000/iq2000.c: Likewise.
18273 * config/lm32/lm32.c: Likewise.
18274 * config/m32c/m32c.c: Likewise.
18275 * config/m32r/m32r.c: Likewise.
18276 * config/m68k/m68k.c: Likewise.
18277 * config/mcore/mcore.c: Likewise.
18278 * config/microblaze/microblaze.c: Likewise.
18279 * config/mips/mips.c: Likewise.
18280 * config/mmix/mmix.c: Likewise.
18281 * config/mn10300/mn10300.c: Likewise.
18282 * config/moxie/moxie.c: Likewise.
18283 * config/msp430/msp430.c: Likewise.
18284 * config/nds32/nds32-isr.c: Likewise.
18285 * config/nds32/nds32.c: Likewise.
18286 * config/nios2/nios2.c: Likewise.
18287 * config/nvptx/nvptx.c: Likewise.
18288 * config/pa/pa.c: Likewise.
18289 * config/pdp11/pdp11.c: Likewise.
18290 * config/powerpcspe/powerpcspe.c: Likewise.
18291 * config/riscv/riscv.c: Likewise.
18292 * config/rl78/rl78.c: Likewise.
18293 * config/rx/rx.c: Likewise.
18294 * config/s390/s390.c: Likewise.
18295 * config/sh/sh.c: Likewise.
18296 * config/sol2.c: Likewise.
18297 * config/sparc/sparc.c: Likewise.
18298 * config/spu/spu.c: Likewise.
18299 * config/stormy16/stormy16.c: Likewise.
18300 * config/tilegx/tilegx.c: Likewise.
18301 * config/tilepro/tilepro.c: Likewise.
18302 * config/v850/v850.c: Likewise.
18303 * config/vax/vax.c: Likewise.
18304 * config/visium/visium.c: Likewise.
18305 * config/xtensa/xtensa.c: Likewise.
18307 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18310 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18311 constraints since the -mupper-regs-* switches have been
18313 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18314 into a vector from a double word element that was extracted from
18315 another vector, and eliminate extra XXPERMDI instructions.
18316 (vsx_concat_<mode>_2): Likewise.
18317 (vsx_concat_<mode>_3): Likewise.
18318 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18319 concat to allow optimizing inserts from previous extracts.
18321 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18323 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18324 memory reference to a SSP offset in TLS address space.
18325 (ix86_print_operand) <case '@'>: Remove.
18326 (ix86_print_operand_punct_valid_p): Remove '@' code.
18327 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18328 UNSPEC_SP_TLS_TEST.
18329 (stack_tls_protect_set_<mode>): Remove.
18330 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18331 (stack_tls_protect_test_<mode>): Remove.
18332 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18334 2017-08-07 Olivier Hainque <hainque@adacore.com>
18337 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18339 2017-08-07 Douglas Rupp <rupp@adacore.com>
18341 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18342 variable was referenced as multidir in command.
18344 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18347 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18350 2017-08-07 Martin Liska <mliska@suse.cz>
18352 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18353 * config/rl78/rl78.c: Add include of attribs.h.
18354 * config/sh/sh.c: Likewise.
18355 * config/v850/v850.c: Likewise.
18357 2017-08-07 Tom de Vries <tom@codesourcery.com>
18359 PR middle-end/78266
18360 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18362 2017-08-07 Martin Liska <mliska@suse.cz>
18364 * config/mips/mips.c: Include attribs.h.
18366 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18369 * doc/invoke.texi: Document change in behvaior for -Ofast for
18372 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18374 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18376 (aarch64_pop_regs): Likewise.
18377 (aarch64_gen_load_pair): Likewise.
18378 (aarch64_save_callee_saves): Likewise.
18379 (aarch64_restore_callee_saves): Likewise.
18381 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18383 * config/i386/i386.c: Revert the last change.
18385 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18388 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18390 (ix86_finalize_stack_frame_flags): This. Also clear
18391 frame_pointer_needed if -fno-omit-frame-pointer is used without
18393 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18394 with ix86_finalize_stack_frame_flags.
18395 (ix86_expand_epilogue): Likewise.
18396 (ix86_expand_split_stack_prologue): Likewise.
18398 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18401 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18402 to P_AES for Westmere.
18404 2017-08-07 Jonathan Yong <10walls@gmail.com>
18406 * config/i386/mingw.opt (fset-stack-executable): Removed.
18407 * config/i386/cygming.opt (fset-stack-executable): Moved
18409 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18411 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18413 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18415 2017-08-07 Marek Polacek <polacek@redhat.com>
18417 PR middle-end/81737
18418 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18420 2017-08-07 Martin Liska <mliska@suse.cz>
18422 * attribs.h (canonicalize_attr_name): New function.
18423 (cmp_attribs): Move from c-format.c and adjusted.
18424 (is_attribute_p): Moved from tree.h.
18425 * tree-inline.c: Add new includes.
18426 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18427 (private_is_attribute_p): Remove.
18428 (private_lookup_attribute): Likewise.
18429 (private_lookup_attribute_by_prefix): Simplify.
18430 (remove_attribute): Use is_attribute_p.
18431 * tree.h: Remove removed declarations.
18433 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18435 PR middle-end/81698
18436 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18437 instead of computing it in the function. Formatting fix.
18438 (expand_case): Don't rely on default_edge being the first edge,
18439 clear it if removing it, pass default_edge to
18440 emit_case_dispatch_table.
18441 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18444 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18446 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18447 insn in the function, emit NOP after the insn.
18449 2017-08-06 Tom de Vries <tom@codesourcery.com>
18451 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18454 2017-08-06 Tom de Vries <tom@codesourcery.com>
18456 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18459 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18461 PR tree-optimization/57371
18462 * match.pd: New pattern.
18464 2017-08-04 Marek Polacek <polacek@redhat.com>
18466 PR middle-end/81695
18467 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18468 perform the computation in offset_int.
18470 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18472 PR tree-optimization/81136
18473 * tree-vectorizer.h: Include tree-hash-traits.h.
18474 (vec_base_alignments): New typedef.
18475 (vec_info): Add a base_alignments field.
18476 (vect_record_base_alignments): Declare.
18477 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18479 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18480 (create_data_ref): Add an is_conditional_in_stmt argument.
18481 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18482 the is_conditional_in_stmt field.
18483 (data_ref_loc): Add an is_conditional_in_stmt field.
18484 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18485 (find_data_references_in_stmt): Update call to create_data_ref.
18486 (graphite_find_data_references_in_stmt): Likewise.
18487 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18488 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18489 (vect_record_base_alignment): New function.
18490 (vect_record_base_alignments): Likewise.
18491 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18492 for nested statements even if we fail to compute a misalignment.
18493 Use pooled base alignments for unconditional references.
18494 (vect_find_same_alignment_drs): Compare base addresses instead
18496 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18497 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18499 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18501 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18502 Add an explicit name for the enum. Use auto_vec for slp_instances
18503 and grouped_stores.
18504 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18506 (_bb_vec_info): Add a constructor and destructor.
18507 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18508 (destroy_loop_vec_info): Delete.
18509 (vect_destroy_datarefs): Likewise.
18510 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18511 (vec_info::vec_info): New function.
18512 (vec_info::~vec_info): Likewise.
18513 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18514 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18515 destroy_loop_vec_info.
18516 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18517 (_loop_vec_info::_loop_vec_info): ...this.
18518 (destroy_loop_vec_info): Replace with...
18519 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18520 the stmt_vec_infos. Leave handling of vec_info information to its
18521 destructor. Remove explicit vector releases.
18522 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18523 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18524 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18525 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18526 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18527 (destroy_bb_vec_info): Replace with...
18528 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18529 information to its destructor.
18530 (vect_slp_analyze_bb_1): Use new and delete instead of
18531 new_bb_vec_info and destroy_bb_vec_info.
18532 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18535 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18537 * tree-data-ref.h (subscript): Add access_fn field.
18538 (data_dependence_relation): Add could_be_independent_p.
18539 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18540 (same_access_functions): Move to tree-data-ref.c.
18541 * tree-data-ref.c (ref_contains_union_access_p): New function.
18542 (access_fn_component_p): Likewise.
18543 (access_fn_components_comparable_p): Likewise.
18544 (dr_analyze_indices): Add a reference to access_fn_component_p.
18545 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18547 (constant_access_functions): Likewise.
18548 (add_other_self_distances): Likewise.
18549 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18550 (initialize_data_dependence_relation): Use XCNEW and remove
18551 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18552 of access functions that have the same type. Allow the
18553 subsequence to end with different bases in some circumstances.
18554 Record the chosen access functions in SUB_ACCESS_FN.
18555 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18556 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18557 (subscript_dependence_tester_1): Likewise dra and drb.
18558 (build_classic_dist_vector): Update calls accordingly.
18559 (subscript_dependence_tester): Likewise.
18560 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18561 DDR_COULD_BE_INDEPENDENT_P.
18562 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18563 comp_alias_ddrs instead of may_alias_ddrs.
18564 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18566 (vect_analyze_data_ref_dependence): Use it if
18567 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18568 distance vectors if that fails.
18569 (dependence_distance_ge_vf): New function.
18570 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18571 LOOP_VINFO_MAY_ALIAS_DDRS.
18573 2017-08-04 Richard Biener <rguenther@suse.de>
18575 PR middle-end/81705
18576 * fold-const.c (fold_binary_loc): Properly restrict
18577 minus_var0 && minus_var1 case when associating undefined overflow
18580 2017-08-04 Tom de Vries <tom@codesourcery.com>
18582 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18584 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18586 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18587 Don't start diagnostic messages with a capital letter.
18588 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18590 (rs6000_invalid_builtin): Likewise.
18591 (rs6000_trampoline_init): Likewise.
18593 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18596 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18597 after setting changeable df flags.
18599 2017-08-03 Richard Biener <rguenther@suse.de>
18601 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18602 up if the use is in USE - X.
18604 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18606 * toplev.c (dumpfile.h): New include.
18607 (internal_error_reentered): New static function. Use it...
18608 (internal_error_function): ...here to handle reentered internal_error.
18610 2017-08-03 Richard Biener <rguenther@suse.de>
18612 PR middle-end/81148
18613 * fold-const.c (split_tree): Add minus_var and minus_con
18614 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18615 here but always use minus_*.
18616 (associate_trees): Assert we never associate with MINUS_EXPR
18617 and NULL first operand. Do not recurse for PLUS_EXPR operands
18618 when associating as MINUS_EXPR either.
18619 (fold_binary_loc): Track minus_var and minus_con.
18621 2017-08-03 Tom de Vries <tom@codesourcery.com>
18624 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18625 ACCEL_COMPILER, apply finish_options on
18626 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18628 2017-08-03 Tom de Vries <tom@codesourcery.com>
18631 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18632 function_entry_patch_area_size > 0.
18634 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18637 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18638 instead of 10??LU, perform unit multiplication in wide_int,
18639 don't change alloc_object_size_limit if the limit is larger
18642 PR tree-optimization/81655
18643 PR tree-optimization/81588
18644 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18645 the case when ranges[i].low and high are 1 for unsigned type with
18648 PR middle-end/81052
18649 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18650 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18652 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18654 * tree-vrp.h: Add include guard.
18656 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18659 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18660 (ud2): New insn pattern.
18661 * config/i386/i386.c (ix86_expand_epilogue):
18662 For naked functions, generate ud2 instead of trap insn.
18664 2017-08-02 Marek Polacek <polacek@redhat.com>
18667 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18669 2017-08-02 Tom de Vries <tom@codesourcery.com>
18670 Cesar Philippidis <cesar@codesourcery.com>
18672 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18673 Add missing edge probabilities.
18675 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18677 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18678 Correct endianness.
18680 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18682 PR middle-end/79499
18683 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18684 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18685 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18686 prologue_seq sequences - if any.
18688 2017-08-02 Richard Biener <rguenther@suse.de>
18690 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18691 via vectors if supported, integer extracts via punning if supported
18692 or otherwise vector extracts.
18694 2017-08-02 Richard Biener <rguenther@suse.de>
18696 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18698 (bitmap_insert_into_set): ... this.
18700 2017-08-02 Richard Biener <rguenther@suse.de>
18702 PR tree-optimization/81633
18704 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18706 PR tree-optimization/71752
18707 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18709 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18711 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18712 (machine_function::call_ms2sysv_pad_out): Remove field.
18713 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18714 (ix86_compute_frame_layout): Likewise.
18716 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18719 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18720 attribute with interrupt attribute.
18722 2017-08-01 Andrew Pinski <apinski@cavium.com>
18724 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18725 BIT_INSERT_EXPR's operand 1
18726 to see if the types precision matches.
18728 2017-08-01 Martin Liska <mliska@suse.cz>
18730 PR middle-end/70140
18731 * builtins.c (expand_builtin_memcpy_args): Remove.
18732 (expand_builtin_memcpy): Call newly added function
18733 expand_builtin_memory_copy_args.
18734 (expand_builtin_memcpy_with_bounds): Likewise.
18735 (expand_builtin_mempcpy): Remove last argument.
18736 (expand_builtin_mempcpy_with_bounds): Likewise.
18737 (expand_builtin_memory_copy_args): New function created from
18738 expand_builtin_mempcpy_args with small modifications.
18739 (expand_builtin_mempcpy_args): Remove.
18740 (expand_builtin_stpcpy): Remove unused argument.
18741 (expand_builtin): Likewise.
18742 (expand_builtin_with_bounds): Likewise.
18744 2017-08-01 Martin Liska <mliska@suse.cz>
18747 Make mempcpy more optimal (PR middle-end/70140).
18749 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18752 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18753 __builtin_vec_cmpne verify both arguments are compatible vectors
18754 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18755 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18756 move computation of aligned to after checking the argument types.
18760 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18761 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18764 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18765 Doug Rupp <rupp@adacore.com>
18766 Olivier Hainque <hainque@adacore.com>
18768 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18769 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18771 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18772 for TARGET_OS_CPP_BUILTIN.
18773 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18774 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18776 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18777 passing required abi options to the assembler for EABI configurations.
18778 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18779 of .text.hot and .text.unlikely sections for kernel modules when
18780 using ARM style exceptions.
18781 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18782 options. Add EXTRA_CC1_SPEC.
18783 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18785 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18787 (ARM_TARGET2_DWARF_FORMAT): Define.
18788 * config/arm/t-vxworks: Adjust multilib control to removal of the
18789 Diab command line options.
18791 2017-08-01 Martin Liska <mliska@suse.cz>
18793 PR gcov-profile/81561
18794 * gcov.c (unblock): Make unblocking safe as we need to preserve
18795 index correspondence of blocks and block_lists.
18797 2017-08-01 Richard Biener <rguenther@suse.de>
18799 PR tree-optimization/81181
18800 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18801 (compute_antic): ... end of iteration here.
18803 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18805 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18806 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18807 (ftree-slp-vectorize): Likewise.
18808 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18809 can no longer be set independent of flag_tree_loop_vectorize.
18810 * omp-general.c (emp_max_vf): Likewise.
18811 * opts.c (enable_fdo_optimizations): Remove references to
18812 flag_tree_vectorize, these are now implicit.
18813 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18814 and leave it for the options machinery.
18816 2017-08-01 Martin Liska <mliska@suse.cz>
18818 PR middle-end/70140
18819 * builtins.c (expand_builtin_memcpy_args): Remove.
18820 (expand_builtin_memcpy): Call newly added function
18821 expand_builtin_memory_copy_args.
18822 (expand_builtin_memcpy_with_bounds): Likewise.
18823 (expand_builtin_mempcpy): Remove last argument.
18824 (expand_builtin_mempcpy_with_bounds): Likewise.
18825 (expand_builtin_memory_copy_args): New function created from
18826 expand_builtin_mempcpy_args with small modifications.
18827 (expand_builtin_mempcpy_args): Remove.
18828 (expand_builtin_stpcpy): Remove unused argument.
18829 (expand_builtin): Likewise.
18830 (expand_builtin_with_bounds): Likewise.
18832 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18835 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18836 print "ds:" only for immediates in generic address space.
18838 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18841 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18842 (ix86_function_ok_for_sibcall): Return false for naked functions.
18844 2017-08-01 Richard Biener <rguenther@suse.de>
18846 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18847 (compute_antic): Seed worklist with exit block predecessors.
18848 * cfganal.c (dfs_find_deadend): For a cycle return the source
18849 of the edge closing it.
18851 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18853 * config/aarch64/aarch64.c
18854 (aarch64_can_const_movi_rtx_p): Move 0 check.
18856 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18858 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18859 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18861 * match.pd: Ditto in address comparison pattern.
18863 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18865 PR tree-optimization/81627
18866 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18867 closed ssa form for store-store chain.
18869 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18871 PR tree-optimization/81620
18872 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18873 for store-store chain.
18875 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18877 PR tree-optimization/81588
18878 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18879 ranges[i].in_p, invert comparison code ccode. For >/>=,
18880 swap rhs1 and rhs2 and comparison code unconditionally,
18881 for </<= don't do that. Don't swap rhs1/rhs2 again if
18882 ranges[i].in_p, instead invert comparison code ccode if
18883 opcode or oe->rank is BIT_IOR_EXPR.
18886 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18887 a direct optab to conversion optab.
18888 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18889 with GET_MODE_INNER as last argument instead of optab_handler.
18890 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18891 vector extraction if possible and optab is available.
18892 * expr.c (store_constructor): Use convert_optab_handler instead
18893 of optab_handler. Use vector initialization from smaller
18894 vectors if possible and optab is available.
18895 * tree-vect-stmts.c (vectorizable_load): Likewise.
18896 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18897 now have two modes.
18898 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18899 of vec_init from half-sized vectors with the same element mode.
18900 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18901 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18902 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18903 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18904 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18905 after mode in gen_vec_extract* calls.
18906 (vec_extract<mode>): Renamed to ...
18907 (vec_extract<mode><ssescalarmodelower>): ... this.
18908 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18909 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18910 element mode after mode in gen_vec_init* calls.
18911 (VEC_INIT_HALF_MODE): New mode iterator.
18912 (vec_init<mode>): Renamed to ...
18913 (vec_init<mode><ssescalarmodelower>): ... this.
18914 (vec_init<mode><ssehalfvecmodelower>): New expander.
18915 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18916 (vec_extractv2sfsf): ... this.
18917 (vec_initv2sf): Renamed to ...
18918 (vec_initv2sfsf): ... this.
18919 (vec_extractv2si): Renamed to ...
18920 (vec_extractv2sisi): ... this.
18921 (vec_initv2si): Renamed to ...
18922 (vec_initv2sisi): ... this.
18923 (vec_extractv4hi): Renamed to ...
18924 (vec_extractv4hihi): ... this.
18925 (vec_initv4hi): Renamed to ...
18926 (vec_initv4hihi): ... this.
18927 (vec_extractv8qi): Renamed to ...
18928 (vec_extractv8qiqi): ... this.
18929 (vec_initv8qi): Renamed to ...
18930 (vec_initv8qiqi): ... this.
18931 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18932 (vec_init<mode>): Renamed to ...
18933 (vec_init<mode><VEC_base_l>): ... this.
18934 (vec_extract<mode>): Renamed to ...
18935 (vec_extract<mode><VEC_base_l>): ... this.
18936 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18937 (vec_initv2sfsf): ... this.
18938 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18939 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18940 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18941 element mode after mode in gen_vec_init* calls.
18942 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18943 (vec_init<mode><Vel>): ... this.
18944 (vec_extract<mode>): Renamed to ...
18945 (vec_extract<mode><Vel>): ... this.
18946 * config/aarch64/iterators.md (Vel): New mode attribute.
18947 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18948 Add element mode after mode in gen_vec_extract* calls.
18949 * config/s390/vector.md (non_vec_l): New mode attribute.
18950 (vec_extract<mode>): Renamed to ...
18951 (vec_extract<mode><non_vec_l>): ... this.
18952 (vec_init<mode>): Renamed to ...
18953 (vec_init<mode><non_vec_l>): ... this.
18954 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18955 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18957 * config/arm/iterators.md (V_elem_l): New mode attribute.
18958 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18959 (vec_extract<mode><V_elem_l>): ... this.
18960 (vec_extractv2di): Renamed to ...
18961 (vec_extractv2didi): ... this.
18962 (vec_init<mode>): Renamed to ...
18963 (vec_init<mode><V_elem_l>): ... this.
18964 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18965 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18966 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18967 Add element mode after gen_vec_extract* calls.
18968 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18969 (vec_init<mode><unitmode>): ... this.
18970 (vec_extract<mode>): Renamed to ...
18971 (vec_extract<mode><unitmode>): ... this.
18972 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18973 (vec_init<mode><unitmode>): ... this.
18974 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18975 (vec_initv2sfsf): ... this.
18976 (vec_extractv2sf): Renamed to ...
18977 (vec_extractv2sfsf): ... this.
18978 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18979 Add element mode after gen_vec_extract* calls.
18980 * config/mips/mips.md (unitmode): New mode iterator.
18981 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18982 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18983 * config/spu/spu.md (inner_l): New mode attribute.
18984 (vec_init<mode>): Renamed to ...
18985 (vec_init<mode><inner_l>): ... this.
18986 (vec_extract<mode>): Renamed to ...
18987 (vec_extract<mode><inner_l>): ... this.
18988 * config/sparc/sparc.md (veltmode): New mode iterator.
18989 (vec_init<VMALL:mode>): Renamed to ...
18990 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18991 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18992 (vec_initv2sisi): ... this.
18993 (vec_initv2sf): Renamed to ...
18994 (vec_initv2sfsf): ... this.
18995 (vec_extractv2sf): Renamed to ...
18996 (vec_extractv2sfsf): ... this.
18997 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18998 (vec_init<mode>): Renamed to ...
18999 (vec_init<mode><VEC_base_l>): ... this.
19000 (vec_extract<mode>): Renamed to ...
19001 (vec_extract<mode><VEC_base_l>): ... this.
19002 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19003 (vec_initv2sfsf): ... this.
19004 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19005 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19006 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19007 gen_vec_init* calls.
19009 2017-08-01 Richard Biener <rguenther@suse.de>
19011 PR tree-optimization/81297
19012 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19013 TREE_OVERFLOW from INTEGER_CSTs.
19015 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19017 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19019 2017-07-31 Carl Love <cel@us.ibm.com>
19021 * config/rs6000/rs6000-c: Add support for built-in functions
19022 vector signed char vec_xl_be (signed long long, signed char *);
19023 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19024 vector signed int vec_xl_be (signed long long, signed int *);
19025 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19026 vector signed long long vec_xl_be (signed long long, signed long long *);
19027 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19028 vector signed short vec_xl_be (signed long long, signed short *);
19029 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19030 vector double vec_xl_be (signed long long, double *);
19031 vector float vec_xl_be (signed long long, float *);
19032 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19033 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19034 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19036 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19037 (altivec_expand_builtin): Add switch statement to call
19038 altivec_expand_xl_be for each builtin.
19039 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19040 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19041 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19042 __builtin_vsx_le_be_v16qi.
19043 * doc/extend.texi: Update the built-in documentation file for the
19044 new built-in functions.
19046 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19049 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19051 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19053 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19055 * config.gcc: Add z14.
19056 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19057 CPU model numbers for z13s and z14.
19058 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19060 * config/s390/s390-opts.h (enum processor_type): Rename
19061 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19062 * config/s390/s390.c (processor_table): Add field for CPU name to
19063 be passed to Binutils.
19064 (s390_asm_output_machine_for_arch): Use the new field in
19065 processor_table for Binutils.
19066 (s390_expand_builtin): Replace arch12 with z14.
19067 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19068 (s390_get_sched_attrmask): Likewise.
19069 (s390_get_unit_mask): Likewise.
19070 * config/s390/s390.opt: Add z14 to processor_type enum.
19072 2017-07-31 Martin Jambor <mjambor@suse.cz>
19075 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19076 regardless of optimization level.
19078 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19079 Martin Liska <mliska@suse.cz>
19081 * predict.def: Remove old comment and adjust probability.
19082 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19083 PREDICT statements.
19085 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19088 * config/i386/i386.c (ix86_function_naked): New function.
19089 (ix86_can_use_return_insn_p): Return false for naked functions.
19090 (ix86_expand_prologue): Skip prologue for naked functions.
19091 (ix86_expand_epilogue): Skip epilogue for naked functions
19092 and emit trap instruction.
19093 (ix86_warn_func_return): New function.
19094 (ix86_attribute_table): Add "naked" attribute specification.
19095 (TARGET_WARN_FUNC_RETURN): Define.
19096 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19098 2017-07-31 Martin Liska <mliska@suse.cz>
19100 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19101 (dump_gimple_bb_header): Always dump BB info.
19102 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19104 2017-07-31 Martin Liska <mliska@suse.cz>
19107 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19108 also with current_function_decl non-null equality.
19110 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19113 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19114 change type to the element type, instead add eltype variable and
19115 use it where we are interested in the element type.
19117 PR tree-optimization/81603
19118 * ipa-polymorphic-call.c
19119 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19120 offset arithmetic in offset_int, bail out if the resulting bit offset
19121 doesn't fit into shwi.
19123 2017-07-31 Martin Liska <mliska@suse.cz>
19125 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19126 (gimplify_save_expr): Fix comment.
19128 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19131 * config/i386/i386.c (ix86_function_arg): Update arguments for
19133 (ix86_compute_frame_layout): Set the initial stack offset to
19134 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19135 INCOMING_FRAME_SP_OFFSET.
19136 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19137 stack before exception handler returns.
19138 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19139 the 'ERROR_CODE' for exception handler.
19141 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19143 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19144 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19145 (ASM_OUTPUT_REG_POP): Ditto.
19146 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19147 instead of asm_fprintf to output pure string.
19149 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19151 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19152 to imported_module_or_decl hook.
19153 (debug_nothing_tree_tree_tree_bool): Remove.
19154 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19155 * debug.c (do_nothing_debug_hooks): Use
19156 debug_nothing_tree_tree_tree_bool_bool instead of
19157 debug_nothing_tree_tree_tree_bool.
19158 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19159 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19160 * sdbout.c (sdb_debug_hooks): Likewise.
19161 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19162 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19164 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19165 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19166 attribute, don't add anything.
19168 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19170 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19171 (fold_build2_stat_loc): Likewise.
19172 (fold_build3_stat_loc): Likewise.
19173 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19174 (fold_build1_loc): Remove macro.
19175 (fold_build2_loc): Likewise.
19176 (fold_build3_loc): Likewise.
19178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19180 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19181 (gimple_build_debug_bind_source_stat): Likewise.
19182 * gimple.h (gimple_build_debug_bind): Remove macro.
19183 (gimple_build_debug_bind_source): Likewise.
19185 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19187 * bitmap.c (bitmap_alloc): Adjust.
19188 (bitmap_gc_alloc): Likewise.
19189 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19191 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19193 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19194 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19195 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19196 (bitmap_gc_alloc_stat): Likewise.
19197 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19199 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19201 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19202 * rtl.h (shallow_copy_rtx): Remove macro.
19204 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19206 * emit-rtl.c (gen_raw_REG): Adjust.
19207 * gengenrtl.c (gendef): Likewise.
19208 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19209 * rtl.h (rtx_alloc): Remove macro.
19211 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19213 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19214 (build_tree_list_stat): Likewise.
19215 * tree.h (build_tree_list): Remove macro.
19216 (build_tree_list_vec): Likewise.
19218 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19220 * tree.c (make_vector_stat): Remove _stat from name.
19221 (build_vector_stat): Likewise.
19222 * tree.h (make_vector_stat): Remove macro.
19223 (build_vector_stat): Likewise.
19225 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19227 * tree.h (build_var_debug_value): Remove prototype.
19229 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19231 * tree.c (tree_cons_stat): Remove _stat from name.
19232 * tree.h (tree_cons): Remove macro.
19234 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19236 * tree.c (build_vl_exp_stat): Remove _stat from name.
19237 * tree.h (build_vl_exp): Remove macro.
19239 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19241 * tree.c (build_decl_stat): Remove _stat from name.
19242 * tree.h (build_decl): Remove macro.
19244 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19246 * gimple.c (gimple_build_with_ops_stat): Adjust.
19247 (gimple_alloc_stat): Remove _stat from name.
19248 * gimple.h (gimple_alloc): Remove macro.
19250 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19252 * tree.c (make_tree_vec_stat): Remove _stat from name.
19253 (grow_tree_vec_stat): Likewise.
19254 * tree.h (make_tree_vec_stat): Adjust prototype.
19255 (grow_tree_vec_stat): Likewise.
19256 (make_tree_vec): Remove macro.
19257 (grow_tree_vec): Likewise.
19259 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19261 * fold-const.c (fold_build1_stat_loc): Adjust.
19262 (fold_build2_stat_loc): Likewise.
19263 (fold_build3_stat_loc): Likewise.
19264 * tree.c (build0_stat): Remove _stat from name.
19265 (build1_stat): Likewise.
19266 (build2_stat): Likewise.
19267 (build3_stat): Likewise.
19268 (build4_stat): Likewise.
19269 (build5_stat): Likewise.
19270 * tree.h (build1_loc): Remove macro, and rename _stat function
19272 (build2_loc): Likewise.
19273 (build3_loc): Likewise.
19274 (build4_loc): Likewise.
19275 (build5_loc): Likewise.
19277 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19279 * tree.c (make_int_cst_stat): Remove _stat from name.
19280 * tree.h (make_int_cst_stat): Adjust prototype.
19281 (make_int_cst): Remove macro.
19283 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19285 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19286 * tree.h (make_tree_binfo_stat): Adjust prototype.
19287 (make_tree_binfo): Remove.
19289 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19291 * tree.c (copy_node_stat): Rename to copy_node.
19292 (build_distinct_type_copy): Adjust.
19293 * tree.h (copy_node_stat): Adjust prototype.
19294 (copy_node): Remove macro.
19296 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19298 * tree.c (make_node_stat): rename to make_node.
19299 (build_tree_list_stat): Adjust.
19300 (build0_stat): Likewise.
19301 (build2_stat): Likewise.
19302 (build3_stat): Likewise.
19303 (build4_stat): Likewise.
19304 (build5_stat): Likewise.
19305 (build_decl_stat): Likewise.
19306 * tree.h (make_node_stat): Adjust prototype.
19307 (make_node): remove macro.
19309 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19311 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19312 (PPC_FEATURE2_SCV): Likewise.
19313 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19315 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19317 * config/aarch64/aarch64.c
19318 (aarch64_internal_mov_immediate): Add new special pattern.
19319 * config/aarch64/aarch64.md (*movdi_aarch64):
19320 Add reg/32bit const mov case.
19322 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19323 Richard Sandiford <richard.sandiford@linaro.org>
19325 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19326 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19327 Add integer and movi cases.
19328 (movi-split-hf-df-sf split, fp16): New.
19329 (enabled): Added TARGET_FP_F16INST.
19330 * config/aarch64/iterators.md (GPF_HF): New.
19331 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19333 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19335 * config/aarch64/aarch64.c
19336 (aarch64_simd_container_mode): Add prototype.
19337 (aarch64_expand_mov_immediate): Add HI support.
19338 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19339 (aarch64_can_const_movi_rtx_p): New.
19340 (aarch64_preferred_reload_class):
19341 Remove restrictions of using FP registers for certain SIMD operations.
19342 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19343 (aarch64_valid_floating_const): Add integer move validation.
19344 (aarch64_simd_imm_scalar_p): Remove.
19345 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19346 (aarch64_legitimate_constant_p): Expand list of supported cases.
19347 * config/aarch64/aarch64-protos.h
19348 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19349 (aarch64_reinterpret_float_as_int): New.
19350 (aarch64_simd_imm_scalar_p): Remove.
19351 * config/aarch64/constraints.md (Uvi): New.
19352 (Dd): Split into Ds and new Dd.
19353 * config/aarch64/aarch64.md (*movsi_aarch64):
19355 (*movdi_aarch64): Add SIMD mov case.
19357 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19359 * tree-predcom.c: (struct chain): Handle store-store chain in which
19360 stores for elimination only store loop invariant values.
19361 (execute_pred_commoning_chain): Ditto.
19362 (prepare_initializers_chain_store_elim): Ditto.
19363 (prepare_finalizers): Ditto.
19364 (is_inv_store_elimination_chain): New function.
19365 (initialize_root_vars_store_elim_1): New function.
19367 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19369 * tree-predcom.c: Revise general description of the pass.
19370 (enum chain_type): New enum type for store elimination.
19371 (struct chain): New field supporting store elimination.
19372 (struct component): Ditto.
19373 (dump_chain): Dump store-stores chain.
19374 (release_chain): Release resources.
19375 (split_data_refs_to_components): Compute and create component
19376 contains only stores for elimination.
19377 (get_chain_last_ref_at): New function.
19378 (make_invariant_chain): Initialization.
19379 (make_rooted_chain): Specify chain type in parameter and record it.
19380 (add_looparound_copies): Skip for store-stores chain.
19381 (determine_roots_comp): Compute type of chain and pass it to
19383 (initialize_root_vars_store_elim_2): New function.
19384 (finalize_eliminated_stores): New function.
19385 (remove_stmt): Handle store for elimination.
19386 (execute_pred_commoning_chain): Execute predictive commoning on
19387 store-store chains.
19388 (determine_unroll_factor): Skip unroll for store-stores chain.
19389 (prepare_initializers_chain_store_elim): New function.
19390 (prepare_initializers_chain): Hanlde store-store chain.
19391 (prepare_finalizers_chain, prepare_finalizers): New function.
19392 (tree_predictive_commoning_loop): Return integer value indicating
19393 if loop is unrolled or lcssa form is corrupted.
19394 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19396 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19398 * tree-predcom.c (initialize_root): Delete.
19399 (execute_pred_commoning_chain): Initialize root vars and replace
19400 reference of non-combined chain directly, rather than call above
19403 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19405 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19406 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19408 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19410 * tree-predcom.c (struct chain): New field init_seq.
19411 (release_chain): Release init_seq.
19412 (prepare_initializers_chain): Record intialization stmts in above
19414 (insert_init_seqs): New function.
19415 (tree_predictive_commoning_loop): Call insert_init_seqs.
19417 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19419 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19421 2017-07-28 Richard Biener <rguenther@suse.de>
19423 * match.pd: Remove superfluous :c.
19424 * genmatch.c (simplify::id): Add member.
19425 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19427 (current_id): New global.
19428 (dt_node::parent): Move from ...
19429 (dt_operand::parent): ... here. Add for_id member.
19430 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19431 (decision_tree::find_node): Relax order requirement when
19432 merging DT_TRUE nodes to ones inbetween the current simplify
19433 and the one we try to merge with. Add diagnostic whenever
19434 we need to enforce pattern order by not merging.
19435 (decision_tree::insert): Set current_id.
19436 (decision_tree::print_node): Dump parent node and for_id.
19437 (parser::last_id): Add member.
19438 (parser::push_simplify): Assign unique id.
19439 (parser::parser): Initialize last_id.
19441 2017-07-28 Martin Liska <mliska@suse.cz>
19444 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19445 gimple_build_debug_bind.
19447 2017-07-28 Richard Biener <rguenther@suse.de>
19449 PR tree-optimization/81502
19450 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19452 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19454 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19455 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19456 for BIT_FIELD_REF args.
19457 * fold-const.c (make_bit_field_ref): Likewise.
19458 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19460 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19463 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19464 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19465 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19466 Or it into SANITIZER_UNDEFINED.
19467 * ubsan.c: Include gimple-fold.h and varasm.h.
19468 (ubsan_expand_ptr_ifn): New function.
19469 (instrument_pointer_overflow): New function.
19470 (maybe_instrument_pointer_overflow): New function.
19471 (instrument_object_size): Formatting fix.
19472 (pass_ubsan::execute): Call instrument_pointer_overflow
19473 and maybe_instrument_pointer_overflow.
19474 * internal-fn.c (expand_UBSAN_PTR): New function.
19475 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19476 * sanitizer.def (__ubsan_handle_pointer_overflow,
19477 __ubsan_handle_pointer_overflow_abort): New builtins.
19478 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19479 * internal-fn.def (UBSAN_PTR): New internal function.
19480 * opts.c (sanitizer_opts): Add pointer-overflow.
19481 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19482 * fold-const.c (build_range_check): Compute pointer range check in
19483 integral type if pointer arithmetics would be needed. Formatting
19486 2017-07-28 Martin Liska <mliska@suse.cz>
19489 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19490 parameters that are of a variable-length.
19492 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19494 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19496 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19497 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19498 (CRT_CALL_STATIC_FUNCTION): Likewise.
19499 (ASM_DEFAULT_SPEC): New define.
19500 (ASM_SPEC32): Likewise.
19501 (ASM_SPEC64): Likewise.
19502 (ASM_SPEC_COMMON): Likewise.
19503 (ASM_SPEC): Likewise.
19504 (INVALID_64BIT): Likewise.
19505 (LINK_OS_DEFAULT_SPEC): Likewise.
19506 (LINK_OS_SPEC32): Likewise.
19507 (LINK_OS_SPEC64): Likewise.
19508 (POWERPC_LINUX): Likewise.
19509 (PTRDIFF_TYPE): Likewise.
19510 (RESTORE_FP_PREFIX): Likewise.
19511 (RESTORE_FP_SUFFIX): Likewise.
19512 (SAVE_FP_PREFIX): Likewise.
19513 (SAVE_FP_SUFFIX): Likewise.
19514 (SIZE_TYPE): Likewise.
19515 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19516 (TARGET_64BIT): Likewise.
19517 (TARGET_64BIT): Likewise.
19518 (TARGET_AIX): Likewise.
19519 (WCHAR_TYPE_SIZE): Likewise.
19520 (WCHAR_TYPE): Undefine.
19521 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19522 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19523 (CPP_OS_RTEMS_SPEC): Delete.
19524 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19525 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19527 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19529 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19531 PR tree-optimization/81578
19532 * tree-parloops.c (build_new_reduction): Bail out if
19533 reduction_code isn't one of the standard OpenMP reductions.
19534 Move the details printing after that decision.
19536 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19538 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19539 related to reload_in_progress.
19540 (splat_input_operand): Likewise.
19541 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19543 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19545 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19546 (TARGET_INSTANTIATE_DECLS): Likewise.
19547 (legitimate_indexed_address_p): Delete reload_in_progress code.
19548 (rs6000_debug_legitimate_address_p): Likewise.
19549 (rs6000_eliminate_indexed_memrefs): Likewise.
19550 (rs6000_emit_le_vsx_store): Likewise.
19551 (rs6000_emit_move_si_sf_subreg): Likewise.
19552 (rs6000_emit_move): Likewise.
19553 (register_to_reg_type): Likewise.
19554 (rs6000_pre_atomic_barrier): Likewise.
19555 (rs6000_machopic_legitimize_pic_address): Likewise.
19556 (rs6000_allocate_stack_temp): Likewise.
19557 (rs6000_address_for_fpconvert): Likewise.
19558 (rs6000_address_for_altivec): Likewise.
19559 (rs6000_secondary_memory_needed_rtx): Delete function.
19560 (rs6000_check_sdmode): Likewise.
19561 (rs6000_alloc_sdmode_stack_slot): Likewise.
19562 (rs6000_instantiate_decls): Likewise.
19563 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19564 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19565 Delete reload_in_progress.
19566 (*vec_reload_and_plus_<mptrsize>): Likewise.
19567 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19568 (vsx_div_v2di): Likewise.
19569 (vsx_udiv_v2di): Likewise.
19571 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19573 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19574 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19575 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19576 (rs6000_debug_reg_global): Delete print of LRA status.
19577 (rs6000_option_override_internal): Delete dead LRA related code.
19578 (rs6000_lra_p): Delete function.
19579 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19581 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19583 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19584 * config/riscv/rtems.h: New file.
19586 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19587 Sudakshina Das <sudi.das@arm.com>
19589 * config/aarch64/aarch64.md
19590 (define_split for and<mode>3nr_compare): Move
19591 non aarch64_logical_operand to a register.
19592 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19593 register immediate operand to a register.
19594 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19596 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19598 PR middle-end/81564
19599 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19601 2017-07-27 Richard Biener <rguenther@suse.de>
19603 PR tree-optimization/81573
19604 PR tree-optimization/81494
19605 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19606 multi defuse cycle case.
19608 2017-07-27 Richard Biener <rguenther@suse.de>
19610 PR tree-optimization/81571
19611 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19614 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19616 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19617 earlier and only if MASK_FPU is set. Adjust formatting.
19619 2017-07-27 Martin Liska <mliska@suse.cz>
19621 * opt-functions.awk: Add validation of value of Init.
19622 * optc-gen.awk: Pass new argument.
19624 2017-07-27 Martin Liska <mliska@suse.cz>
19626 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19627 Fix wrong condition.
19629 2017-07-27 Martin Liska <mliska@suse.cz>
19631 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19632 BBs and edges seen by autoFDO.
19634 2017-07-27 Richard Biener <rguenther@suse.de>
19636 PR tree-optimization/81502
19637 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19638 with incompatible but same sized type.
19639 (execute_update_addresses_taken): Likewise.
19641 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19643 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19644 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19646 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19649 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19650 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19651 Change s_operand to memory_operand.
19653 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19655 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19656 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19657 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19658 Emit instructions rather than returning an expression. Handle TFmode
19659 and KFmode by casting to TImode.
19660 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19661 (rs6000_emit_le_vsx_store): Likewise.
19662 * config/rs6000/vsx.md (VSX_TI): New iterator.
19663 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19664 (*vsx_le_undo_permute_<mode>): Likewise.
19665 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19666 emit the split sequence.
19667 (*vsx_le_perm_store_<mode>): Likewise.
19669 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19671 PR tree-optimization/81555
19672 PR tree-optimization/81556
19673 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19674 if true, force CHANGED for the recursive invocation.
19675 (reassociate_bb): Remember original length of ops array, pass
19676 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19678 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19679 attributes for noipa attribute. For naked attribute use
19680 lookup_attribute first before lookup_attribute_spec.
19681 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19683 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19684 for functions with noipa attribute.
19685 (cgraph_externally_visible_p): Return true for functions with noipa
19687 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19688 for functions with noipa attribute.
19689 * doc/extend.texi: Document noipa function attribute.
19690 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19691 also for functions with noipa attribute.
19692 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19694 2017-07-26 Andrew Pinski <apinski@cavium.com>
19696 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19697 vec_unalign_load_cost and vec_unalign_store_cost.
19699 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19701 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19702 -mvsx-small-integer option.
19703 (ISA_3_0_MASKS_IEEE): Likewise.
19704 (OTHER_VSX_VECTOR_MASKS): Likewise.
19705 (POWERPC_MASKS): Likewise.
19706 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19707 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19708 code, only testing for DImode being allowed in non-VSX floating
19710 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19711 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19713 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19714 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19715 TARGET_P8_VECTOR test.
19716 (rs6000_secondary_reload_simple_move): Likewise.
19717 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19718 since TARGET_P9_VECTOR was already tested.
19719 (rs6000_opt_masks): Remove -mvsx-small-integer.
19720 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19721 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19723 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19724 test for TARGET_VEXTRACTUB was used, and that uses
19726 (p9 extract splitter): Likewise.
19727 (vsx_extract_<mode>_di_p9): Likewise.
19728 (vsx_extract_<mode>_store_p9): Likewise.
19729 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19730 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19731 the elimination of TARGET_VSX_SMALL_INTEGER.
19732 (vsx_extract_<mode>_p8): Likewise.
19733 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19734 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19735 (vsx_set_<mode>_p9): Likewise.
19736 (vsx_set_v4sf_p9): Likewise.
19737 (vsx_set_v4sf_p9_zero): Likewise.
19738 (vsx_insert_extract_v4sf_p9): Likewise.
19739 (vsx_insert_extract_v4sf_p9_2): Likewise.
19740 * config/rs6000/rs6000.md (sign extend splitter): Change
19741 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19742 (floatsi<mode>2_lfiwax_mem): Likewise.
19743 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19744 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19745 since a test for TARGET_P9_VECTOR was used.
19746 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19747 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19748 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19749 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19750 TARGET_P8_VECTOR test.
19751 (fix_trunc<mode>si2_stfiwx): Likewise.
19752 (fix_trunc<mode>si2_internal): Likewise.
19753 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19754 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19756 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19757 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19758 TARGET_P8_VECTOR test.
19759 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19760 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19761 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19763 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19764 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19765 since a test for TARGET_P9_VECTOR was used.
19766 (splitter for loading small constants): Likewise.
19768 2017-07-26 Andrew Pinski <apinski@cavium.com>
19770 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19773 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19776 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19777 (fp_valid_at): Likewise.
19779 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19781 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19782 (qdf24xx_addrcost_table): Likewise.
19783 (cortexa57_tunings): Update to use generic_branch_cost.
19784 (cortexa72_tunings): Likewise.
19785 (cortexa73_tunings): Likewise.
19786 (qdf24xx_tunings): Likewise.
19788 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19790 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19791 (thunderx2t99_branch_cost): Likewise.
19792 (cortexa35_tunings): Update to use generic_branch_cost.
19793 (cortexa53_tunings): Likewise.
19794 (cortexa57_tunings): Likewise.
19795 (cortexa72_tunings): Likewise.
19796 (cortexa73_tunings): Likewise.
19797 (thunderx2t99_tunings): Likewise.
19799 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19801 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19802 (sparc_option_override): Honour MASK_FSMULD.
19803 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19804 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19805 * config/sparc/sparc.opt (mfsmuld): New option.
19806 * doc/invoke.texi (mfsmuld): Document option.
19808 2017-07-26 Marek Polacek <polacek@redhat.com>
19810 PR middle-end/70992
19811 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19813 2017-07-26 Richard Biener <rguenther@suse.de>
19815 * gimple-match-head.c (do_valueize): Return OP if valueize
19817 (get_def): New helper to get at the def stmt of a SSA name
19818 if valueize allows.
19819 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19820 do_valueize to get at the def stmt.
19821 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19823 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19825 PR middle-end/46932
19826 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19828 2017-07-26 Martin Liska <mliska@suse.cz>
19831 * function.c (expand_function_start): Make expansion of
19832 nonlocal_goto_save_area after parm_birth_insn.
19834 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19836 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19837 from all CPU target flags enable members.
19839 2017-07-26 Richard Biener <rguenther@suse.de>
19841 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19842 (decision_tree::gen): Make 'type' const.
19843 (write_predicate): Likewise.
19845 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19847 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19848 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19849 (rs6000_option_override_internal): Likewise.
19850 (rs6000_expand_vector_set): Likewise.
19851 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19852 (TARGET_UPPER_REGS_SF): Likewise.
19853 (TARGET_UPPER_REGS_DI): Likewise.
19854 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19855 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19856 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19857 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19858 (Splitters for DI constants in Altivec registers): Likewise.
19859 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19860 (vsx_set_v4sf_p9): Likewise.
19861 (vsx_set_v4sf_p9_zero): Likewise.
19862 (vsx_insert_extract_v4sf_p9): Likewise.
19863 (vsx_insert_extract_v4sf_p9_2): Likewise.
19865 2017-07-25 Carl Love <cel@us.ibm.com>
19867 * doc/extend.texi: Update the built-in documentation file for the
19868 existing built-in functions
19869 vector signed char vec_cnttz (vector signed char);
19870 vector unsigned char vec_cnttz (vector unsigned char);
19871 vector signed short vec_cnttz (vector signed short);
19872 vector unsigned short vec_cnttz (vector unsigned short);
19873 vector signed int vec_cnttz (vector signed int);
19874 vector unsigned int vec_cnttz (vector unsigned int);
19875 vector signed long long vec_cnttz (vector signed long long);
19876 vector unsigned long long vec_cnttz (vector unsigned long long);
19878 2017-07-25 Andrew Pinski <apinski@cavium.com>
19880 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19881 accesses where the use is for the first operand of a BIT_INSERT.
19883 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19886 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19887 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19889 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19891 * config/i386/gstabs.h: Delete.
19892 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19894 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19896 * config/i386/i386.c (ix86_decompose_address): Do not check for
19897 register RTX when looking at index_reg or base_reg.
19898 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19900 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19902 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19903 to update EH info here.
19905 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19907 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19909 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19911 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19913 2017-07-25 Torsten Duwe <duwe@suse.de>
19915 * common.opt: Introduce -fpatchable-function-entry
19916 command line option, and its variables function_entry_patch_area_size
19917 and function_entry_patch_area_start.
19918 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19919 including a two-value parser.
19920 * target.def (print_patchable_function_entry): New target hook.
19921 * targhooks.h (default_print_patchable_function_entry): New function.
19922 * targhooks.c (default_print_patchable_function_entry): Likewise.
19923 * toplev.c (process_options): Switch off IPA-RA if
19924 patchable function entries are being generated.
19925 * varasm.c (assemble_start_function): Look at the
19926 patchable-function-entry command line switch and current
19927 function attributes and maybe generate NOP instructions by
19928 calling the print_patchable_function_entry hook.
19929 * doc/extend.texi: Document patchable_function_entry attribute.
19930 * doc/invoke.texi: Document -fpatchable_function_entry
19931 command line option.
19932 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19934 * doc/tm.texi: Re-generate.
19936 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19939 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19940 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19942 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19944 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19946 2017-07-25 Richard Biener <rguenther@suse.de>
19948 PR tree-optimization/81455
19949 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19950 not walk in cycles when looking for guards.
19952 2017-07-25 Richard Biener <rguenther@suse.de>
19954 PR tree-optimization/81529
19955 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19956 when optimizing backedge uses.
19958 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19960 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19962 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19963 to dl_section_ref. On AIX, append an expression to subtract
19964 the size of the section length to dl_section_ref.
19966 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19968 * configure.ac: If any of the config.* scripts fail, exit 1.
19969 * configure: Regenerate.
19971 2017-07-25 Richard Biener <rguenther@suse.de>
19973 PR middle-end/81546
19974 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19975 of immediate uses, be more verbose on errors.
19977 2017-07-25 Richard Biener <rguenther@suse.de>
19979 PR tree-optimization/81510
19980 * tree-vect-loop.c (vect_is_simple_reduction): When the
19981 reduction stmt is not inside the loop bail out.
19983 2017-07-25 Richard Biener <rguenther@suse.de>
19985 PR tree-optimization/81303
19986 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19987 profitability check against LOOP_VINFO_NITERSM1.
19989 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19991 * domwalk.c (cmp_bb_postorder): Simplify.
19992 (sort_bbs_postorder): New function. Use it...
19993 (dom_walker::walk): ...here to optimize common cases.
19995 2017-07-25 Martin Liska <mliska@suse.cz>
19998 * ipa-visibility.c (function_and_variable_visibility): Make the
19999 redirection just on target that supports aliasing.
20000 Fix GNU coding style.
20002 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20005 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20007 * config/arm/rtems.h: Likewise.
20008 * config/bfin/rtems.h: Likewise.
20009 * config/i386/rtemself.h: Likewise.
20010 * config/lm32/rtems.h: Likewise.
20011 * config/m32c/rtems.h: Likewise.
20012 * config/m68k/rtemself.h: Likewise.
20013 * config/microblaze/rtems.h: Likewise.
20014 * config/mips/rtems.h: Likewise.
20015 * config/moxie/rtems.h: Likewise.
20016 * config/nios2/rtems.h: Likewise.
20017 * config/powerpcspe/rtems.h: Likewise.
20018 * config/rs6000/rtems.h: Likewise.
20019 * config/rtems.h: Likewise.
20020 * config/sh/rtems.h: Likewise.
20021 * config/sh/rtemself.h: Likewise.
20022 * config/sparc/rtemself.h: Likewise.
20024 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20027 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20028 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20029 * tree-ssa-structalias.c (alias_get_name): Same.
20031 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20034 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20035 instructions if no du chain is found.
20037 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20039 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20041 2017-07-25 Richard Biener <rguenther@suse.de>
20043 PR middle-end/81505
20044 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20047 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20049 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20050 upper-regs options.
20051 (ISA_2_7_MASKS_SERVER): Likewise.
20052 (ISA_3_0_MASKS_IEEE): Likewise.
20053 (OTHER_P8_VECTOR_MASKS): Likewise.
20054 (OTHER_VSX_VECTOR_MASKS): Likewise.
20055 (POWERPC_MASKS): Likewise.
20056 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20057 duplicate list of options.
20058 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20059 explicit -mupper-regs options.
20060 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20061 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20062 alias for -mupper-regs-df.
20063 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20064 (rs6000_init_hard_regno_mode_ok): Likewise.
20065 (rs6000_option_override_internal): Likewise.
20066 (rs6000_opt_masks): Likewise.
20067 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20068 options in terms of whether -mvsx or -mpower8-vector was used.
20069 (TARGET_UPPER_REGS_DI): Likewise.
20070 (TARGET_UPPER_REGS_SF): Likewise.
20071 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20072 -mupper-regs-* options.
20074 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20076 * passes.c (emergency_dump_function): Print some empty lines and a
20077 header before the RTL dump.
20079 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20081 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20083 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20086 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20087 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20089 2017-07-24 Carl Love <cel@us.ibm.com>
20091 * config/rs6000/rs6000-c.c: Add support for built-in functions
20092 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20093 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20094 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20095 vec_extract_fp_from_shortl): Add defines for the two builtins.
20096 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20097 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20099 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20100 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20101 * doc/extend.texi: Update the built-in documentation file for the
20102 new built-in function.
20104 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20107 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20109 * doc/generic.texi: Likewise.
20110 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20111 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20113 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20115 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20116 (aarch64_mls_elt_merge<mode>): Likewise.
20118 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20120 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20121 having __cxa_atexit.
20123 2017-07-23 Michael Collison <michael.collison@arm.com>
20125 * config/arm/arm.c (arm_option_override): Deprecate
20126 use of -mstructure-size-boundary.
20127 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20128 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20130 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20133 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20134 Reduce cost estimate for direct moves.
20136 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20139 * config/i386/i386.c (ix86_option_override_internal): Disable
20140 BMI, BMI2 and TBM instructions for -m16.
20142 2017-07-21 Carl Love <cel@us.ibm.com>
20144 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20145 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20146 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20147 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20148 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20149 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20150 VMULOSW): New enum "unspec" values.
20151 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20152 altivec_vmulosw): New patterns.
20153 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20154 VMULOSW): Add definitions.
20156 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20158 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20159 (qdf24xx): Likewise.
20160 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20161 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20162 (AARCH64_FL_V8_1): Renumber.
20163 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20164 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20165 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20166 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20167 rdma to feature modifiers list.
20169 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20171 PR middle-end/56727
20172 * ipa-visibility (function_and_variable_visibility): Convert
20173 recursive PLT call to direct call if appropriate.
20175 2017-07-21 Andrew Pinski <apinski@cavium.com>
20177 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20178 operand 1 to see if the types precision matches.
20179 * fold-const.c (operand_equal_p): Likewise.
20181 2017-07-21 Richard Biener <rguenther@suse.de>
20183 PR tree-optimization/81303
20184 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20185 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20186 (vect_peeling_hash_get_lowest_cost): Adjust.
20187 (vect_enhance_data_refs_alignment): Likewise. Use
20188 vect_get_peeling_costs_all_drs to compute the penalty for no
20189 peeling to match up costs.
20191 2017-07-21 Richard Biener <rguenther@suse.de>
20193 PR tree-optimization/81500
20194 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20195 we didn't identify a reduction path.
20197 2017-07-21 Tom de Vries <tom@codesourcery.com>
20198 Cesar Philippidis <cesar@codesourcery.com>
20200 PR gcov-profile/81442
20201 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20204 2017-07-21 Tom de Vries <tom@codesourcery.com>
20207 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20209 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20210 nvptx_override_options_after_change.
20212 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20214 * dwarf2out.c (output_file_names): Avoid double testing for
20215 dwarf_version >= 5.
20217 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20219 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20221 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20223 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20225 (try_crossjump_to_edge): Do not punt on partitioned functions.
20227 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20229 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20230 Put all BBs reachable only via paths crossing cold region to cold
20232 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20234 2016-07-21 Richard Biener <rguenther@suse.de>
20236 PR tree-optimization/81303
20237 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20238 into account prologue and epilogue iterations when raising
20239 min_profitable_iters to sth at least covering one vector iteration.
20241 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20243 * config/arm/arm.c (arm_test_cpu_arch_dat):
20246 2017-07-20 Nathan Sidwell <nathan@acm.org>
20248 Remove TYPE_METHODS.
20249 * tree.h (TYPE_METHODS): Delete.
20250 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20251 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20252 (dbxout_type_methods): Scan TYPE_FIELDS.
20253 (dbxout_type): Don't check TYPE_METHODS here.
20254 * function.c (use_register_for_decl): Always ignore register for
20255 class types when not optimizing.
20256 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20257 * tree.c (free_lang_data_in_type): Stitch out member functions and
20258 templates from TYPE_FIELDS.
20259 (build_distinct_type_copy, verify_type_variant,
20260 verify_type): Member fns are on TYPE_FIELDS.
20261 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20262 * tree-pretty-print.c (dump_generic_node): Likewise.
20264 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20267 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20268 V2TImode and V4TImode.
20269 (ix86_expand_vector_extract): Likewise.
20270 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20271 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20272 (ssescalarmode): Handle V4TImode and V2TImode.
20273 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20274 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20275 (VEXTRACTI128_MODE): New mode iterator.
20276 (splitter for *vec_extractv?ti first element): New.
20277 (VEC_INIT_MODE): New mode iterator.
20278 (vec_init<mode>): Consolidate 3 expanders into one using
20279 VEC_INIT_MODE mode iterator.
20281 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20283 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20284 non_spilled_static_chain_regno_p.
20286 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20288 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20290 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20292 * bb-reorder.c (connect_traces): Allow copying of blocks within
20295 2017-07-20 Richard Biener <rguenther@suse.de>
20297 * gimple.h (gimple_phi_result): Add gphi * overload.
20298 (gimple_phi_result_ptr): Likewise.
20299 (gimple_phi_arg): Likewise. Adjust index assert to only
20300 allow actual argument accesses rather than all slots available
20302 (gimple_phi_arg_def): Add gphi * overload.
20303 * tree-phinodes.c (make_phi_node): Initialize only actual
20305 (resize_phi_node): Clear memory not covered by old node,
20306 do not initialize excess argument slots.
20307 (reserve_phi_args_for_new_edge): Initialize new argument slot
20310 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20312 PR tree-optimization/81388
20314 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20316 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20317 by removing computation of may_be_zero.
20319 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20320 Tom de Vries <tom@codesourcery.com>
20322 PR middle-end/81030
20323 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20324 when gimple level profile disagrees with what RTL expander did.
20326 2017-07-20 Richard Biener <rguenther@suse.de>
20328 PR tree-optimization/61171
20329 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20330 (vect_analyze_stmt): Add slp instance parameter.
20331 (vectorizable_reduction): Likewise.
20332 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20333 (vect_is_simple_reduction): Deal with chains not detected
20334 as SLP reduction chain, specifically not properly associated
20335 chains containing a mix of plus/minus.
20336 (get_reduction_op): Remove.
20337 (get_initial_defs_for_reduction): Simplify, pass in whether
20338 this is a reduction chain, pass in the SLP node for the PHIs.
20339 (vect_create_epilog_for_reduction): Get the SLP instance as
20341 (vectorizable_reduction): Get the SLP instance as arg.
20342 During analysis remember the SLP node with the PHIs in the
20343 instance. Simplify getting at the vectorized reduction PHIs.
20344 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20345 through SLP instance.
20346 (vect_slp_analyze_operations): Likewise.
20347 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20348 (vect_transform_stmt): Likewise.
20350 2017-07-20 Tom de Vries <tom@codesourcery.com>
20352 PR tree-optimization/81489
20353 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20354 read of phi arg location to before loop that modifies phi.
20356 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20358 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20361 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20363 PR middle-end/81331
20364 * except.c (execute): Fix ordering issue.
20366 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20368 PR rtl-optimization/81423
20369 * combine.c (make_compound_operation_int): Don't try to optimize
20370 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20372 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20374 PR rtl-optimization/81423
20375 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20376 with a constant that is -1 in the truncated to mode.
20378 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20380 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20381 (determine_unlikely_bbs): ... here.
20382 * predict.h (propagate_unlikely_bbs_forward): Declare.
20383 * cfgexpand.c (pass_expand::execute): Use it.
20384 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20386 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20387 propagate_unlikely_bbs_forward.
20389 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20391 PR middle-end/81331
20392 * except.c (maybe_add_nop_after_section_switch): New function.
20395 2017-07-19 Tom de Vries <tom@codesourcery.com>
20397 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20399 2017-07-19 Tom de Vries <tom@codesourcery.com>
20401 * gimple.h (gimple_phi_arg): Make assert more strict.
20403 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20405 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20406 * config/rs6000/mmintrin.h: New file.
20407 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20409 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20411 PR tree-optimization/81346
20412 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20414 2017-07-19 Tom de Vries <tom@codesourcery.com>
20416 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20418 2017-07-19 Tom de Vries <tom@codesourcery.com>
20420 * config/nvptx/nvptx-modes.def: Add V2DImode.
20421 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20422 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20423 (nvptx_output_mov_insn): Handle lack of mov.b128.
20424 (nvptx_print_operand): Handle 'H' and 'L' codes.
20425 (nvptx_vector_mode_supported): Allow V2DImode.
20426 (nvptx_preferred_simd_mode): New function.
20427 (nvptx_data_alignment): New function.
20428 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20429 nvptx_preferred_simd_mode.
20430 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20432 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20434 2017-07-19 Tom de Vries <tom@codesourcery.com>
20436 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20437 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20438 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20439 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20440 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20441 (mov<VECIM>_insn): New define_insn.
20442 (define_expand "mov<VECIM>): New define_expand.
20444 2017-07-19 Tom de Vries <tom@codesourcery.com>
20446 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20448 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20450 PR tree-optimization/81346
20451 * fold-const.h (fold_div_compare, range_check_type): Declare.
20452 * fold-const.c (range_check_type): New function.
20453 (build_range_check): Use range_check_type.
20454 (fold_div_compare): No longer static, rewritten into
20455 a match.pd helper function.
20456 (fold_comparison): Don't call fold_div_compare here.
20457 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20458 as helper function.
20460 2017-07-19 Nathan Sidwell <nathan@acm.org>
20462 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20463 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20464 * tree.c (find_decls_types_r, verify_type): Use
20465 TYPE_{MIN,MAX}_VALUE_RAW.
20466 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20467 (hash_tree): Likewise.
20468 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20470 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20473 2017-07-18 Tom de Vries <tom@codesourcery.com>
20475 PR middle-end/81464
20476 * omp-expand.c (expand_omp_for_static_chunk): Handle
20477 equal-argument loop exit phi.
20479 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20482 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20483 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20484 operand 2 predicate.
20485 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20486 operand 2 predicate.
20487 (ror,rol -> rorx splitters): Use const_int_operand as
20488 operand 2 predicate.
20490 2017-06-18 Richard Biener <rguenther@suse.de>
20492 PR tree-optimization/81410
20493 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20494 the gap in the ! slp_perm SLP case after each group.
20496 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20498 PR middle-end/81463
20499 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20502 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20504 PR middle-end/81462
20505 * predict.c (set_even_probabilities): Cleanup; do not affect
20506 probabilities that are already known.
20507 (combine_predictions_for_bb): Call even when count is set.
20509 2017-07-18 Nathan Sidwell <nathan@acm.org>
20511 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20514 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20517 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20518 optimization for loop niter analysis.
20520 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20523 * config/avr/avr.c (avr_optimize_casesi): Don't use
20524 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20526 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20528 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20529 body_cost_vec from _vect_peel_extended_info.
20530 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20531 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20534 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20536 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20538 2017-07-18 Richard Biener <rguenther@suse.de>
20540 PR tree-optimization/80620
20541 PR tree-optimization/81403
20542 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20543 info when re-using a VN table entry.
20545 2017-07-18 Richard Biener <rguenther@suse.de>
20547 PR tree-optimization/81418
20548 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20549 vectype_in. Verify that with lane-reducing reduction operations
20550 we have a single def-use cycle.
20552 2017-07-17 Carl Love <cel@us.ibm.com>
20554 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20556 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20557 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20558 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20559 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20560 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20561 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20562 VMULOSW): New enum "unspec" values.
20563 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20564 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20565 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20566 altivec_vmulosw): New patterns.
20567 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20568 VMULOSW): Add definitions.
20570 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20572 * config/alpha/alpha.c: Include predict.h.
20574 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20576 * tree-vrp.c (compare_assert_loc): Fix comparison function
20577 to return predictable results.
20579 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20581 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20583 (subdi3): Likewise.
20584 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20585 * doc/invoke.texi (mexpand-adddi): Update text.
20587 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20589 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20590 that also clobbers the CC register. The old expand code is moved
20592 (*arc_clzsi2): ... here.
20593 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20594 the CC register. The old expand code is moved to ...
20595 (arc_ctzsi2): ... here.
20597 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20599 * config/arc/arc.opt (mindexed-loads): Use initial value
20600 TARGET_INDEXED_LOADS_DEFAULT.
20601 (mauto-modify-reg): Use initial value
20602 TARGET_AUTO_MODIFY_REG_DEFAULT.
20603 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20604 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20605 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20606 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20608 2017-07-17 Martin Liska <mliska@suse.cz>
20611 * opts.c (finish_options): Do not allow -fgnu-tm
20612 w/ -fsanitize={kernel-,}address. Say sorry.
20614 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20617 * tree-loop-distribution.c (classify_partition): Only assert on
20618 numer of iterations.
20619 (merge_dep_scc_partitions): Delete prameter. Update function call.
20620 (distribute_loop): Remove code handling loop with unknown niters.
20621 (pass_loop_distribution::execute): Skip loop with unknown niters.
20623 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20626 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20627 function sort_partitions_by_post_order.
20629 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20631 PR tree-optimization/81374
20632 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20633 the max index of basic blocks, rather than number of basic blocks.
20635 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20637 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20639 (arc_legitimate_pic_operand_p): Likewise.
20640 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20642 (arc_needs_pcl_p): Likewise.
20643 (arc_legitimate_pc_offset_p): Likewise.
20644 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20645 function is also used in constrains.md.
20646 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20647 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20648 PLUS. Only return true/false in known cases, otherwise assert.
20649 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20650 is already called in arc_legitimate_constant_p.
20651 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20653 (LEGITIMATE_PIC_OPERAND_P): Use
20654 arc_raw_symbolic_reference_mentioned_p function.
20655 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20660 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20661 Andrew Burgess <andrew.burgess@embecosm.com>
20663 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20664 (arc_return_address_register): New function.
20665 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20666 (arc_handle_fndecl_attribute): Add naked attribute.
20667 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20668 (TARGET_WARN_FUNC_RETURN): Likewise.
20669 (arc_allocate_stack_slots_for_args): New function.
20670 (arc_warn_func_return): Likewise.
20671 (machine_function): Change type fn_type.
20672 (arc_compute_function_type): Consider new naked function type,
20673 change function return type.
20674 (arc_must_save_register): Adapt to handle new
20675 arc_compute_function_type's return type.
20676 (arc_expand_prologue): Likewise.
20677 (arc_expand_epilogue): Likewise.
20678 (arc_return_address_regs): Delete.
20679 (arc_return_address_register): New function.
20680 (arc_epilogue_uses): Use above function.
20681 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20682 (arc_function_type): Change encoding, add naked type.
20683 (ARC_INTERRUPT_P): Change to handle the new encoding.
20684 (ARC_FAST_INTERRUPT_P): Likewise.
20685 (ARC_NORMAL_P): Define.
20686 (ARC_NAKED_P): Likewise.
20687 (arc_compute_function_type): Delete prototype.
20688 * config/arc/arc.md (in_ret_delay_slot): Use
20689 arc_return_address_register function.
20690 (simple_return): Likewise.
20691 (p_return_i): Likewise.
20693 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20695 PR tree-optimization/81428
20696 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20697 can't be built for those types.
20699 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20701 Remove stuff dead since r239246.
20703 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20704 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20705 (avr_inform_devices): Remove dead stuff.
20707 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20709 * config/arm/arm_neon.h: Fix softp typo.
20711 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20713 PR tree-optimization/81365
20714 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20715 aggregate moves onto bb predecessor edges, make sure there are no
20716 loads that could alias the lhs in between the start of bb and the
20719 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20722 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20723 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20724 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20726 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20728 PR tree-optimization/81396
20729 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20730 (init_symbolic_number): Initialize it to 1.
20731 (perform_symbolic_merge): Add n_ops from both operands into the new
20733 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20734 without base_addr as useless if they need more than one operation.
20735 (bswap_replace): Handle !bswap case for NULL base_addr.
20737 2017-07-17 Tom de Vries <tom@codesourcery.com>
20740 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20743 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20745 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20746 conditional builtin define __FIX_LEON3FT_B2BST.
20748 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20750 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20751 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20754 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20756 PR rtl-optimization/81424
20757 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20758 to remove potential trapping from operands if -fnon-call-exceptions.
20760 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20762 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20763 profile_proability for scalling.
20764 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20766 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20768 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20770 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20772 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20773 fixpoint arithmetics.
20775 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20777 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20778 fixpoint arithmetics.
20780 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20782 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20783 fixpoint arithmetics.
20785 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20787 * profile-count.h (profile_probability::from_reg_br_prob_note,
20788 profile_probability::to_reg_br_prob_note): New functions.
20789 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20790 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20791 * predict.c (probability_reliable_p): Update.
20792 (edge_probability_reliable_p): Update.
20793 (br_prob_note_reliable_p): Update.
20794 (invert_br_probabilities): Update.
20795 (add_reg_br_prob_note): New function.
20796 (combine_predictions_for_insn): Update.
20797 * asan.c (asan_clear_shadow): Update.
20798 * cfgbuild.c (compute_outgoing_frequencies): Update.
20799 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20800 (update_br_prob_note): Update.
20801 (rtl_verify_edges): Update.
20802 (purge_dead_edges): Update.
20803 (fixup_reorder_chain): Update.
20804 * emit-rtl.c (try_split): Update.
20805 * ifcvt.c (cond_exec_process_insns): Update.
20806 (cond_exec_process_if_block): Update.
20807 (dead_or_predicable): Update.
20808 * internal-fn.c (expand_addsub_overflow): Update.
20809 (expand_neg_overflow): Update.
20810 (expand_mul_overflow): Update.
20811 * loop-doloop.c (doloop_modify): Update.
20812 * loop-unroll.c (compare_and_jump_seq): Update.
20813 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20814 * predict.h: Update.
20815 * reorg.c (mostly_true_jump): Update.
20817 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20818 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20819 * config/arc/arc.c: (emit_unlikely_jump): Update.
20820 * config/arm/arm.c: (emit_unlikely_jump): Update.
20821 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20822 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20823 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20824 (ix86_print_operand): Update.
20825 (ix86_split_fp_branch): Update.
20826 (predict_jump): Update.
20827 * config/ia64/ia64.c (ia64_print_operand): Update.
20828 * config/mmix/mmix.c (mmix_print_operand): Update.
20829 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20830 (rs6000_expand_split_stack_prologue): Update.
20831 * config/rs6000/rs6000.c: Update.
20832 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20833 (s390_expand_vec_movstr): Update.
20834 (s390_expand_cs_tdsi): Update.
20835 (s390_expand_split_stack_prologue): Update.
20836 * config/sh/sh.c (sh_print_operand): Update.
20837 (expand_cbranchsi4): Update.
20838 (expand_cbranchdi4): Update.
20839 * config/sparc/sparc.c (output_v9branch): Update.
20840 * config/spu/spu.c (get_branch_target): Update.
20841 (ea_load_store_inline): Update.
20842 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20843 * config/tilepro/tilepro.c: Update.
20845 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20847 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20848 (gimplify_save_expr): Likewise.
20850 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20852 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20854 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20856 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20858 * timevar.def (TV_IPA_FNSUMMARY): Define.
20860 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20862 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20863 to back store errata sensitive sequence from being generated.
20864 (sqrtdf2_fix): Likewise.
20866 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20868 * tree-ssa-threadupdate.c (compute_path_counts,
20869 update_joiner_offpath_counts): Use profile_probability.
20871 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20874 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20876 * config/arm/arm-c.c (arm_cpu_builtins): Define
20877 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20879 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20881 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20882 array entries to represent __ieee128 versions of the
20883 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20884 scalar_extract_sig, and scalar_insert_exp built-in functions.
20885 (altivec_resolve_overloaded_builtin): Add special case handling
20886 for the __builtin_scalar_insert_exp function, as represented by
20887 the P9V_BUILTIN_VEC_VSIEDP constant.
20888 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20889 exponent support for __ieee128 argument.
20890 (VSESQP): Add scalar extract signature support for __ieee128
20892 (VSTDCNQP): Add scalar test negative support for __ieee128
20894 (VSIEQP): Add scalar insert exponent support for __int128 argument
20895 with __ieee128 result.
20896 (VSIEQPF): Add scalar insert exponent support for __ieee128
20897 argument with __ieee128 result.
20898 (VSTDCQP): Add scalar test data class support for __ieee128
20900 (VSTDCNQP): Add overload support for scalar test negative with
20901 __ieee128 argument.
20902 (VSTDCQP): Add overload support for scalar test data class
20903 __ieee128 argument.
20904 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20905 UNSPEC_VSX_SXSIGDP.
20906 (UNSPEC_VSX_SIEXPQP): New constant.
20907 (xsxexpqp): New insn for VSX scalar extract exponent quad
20909 (xsxsigqp): New insn for VSX scalar extract significand quad
20911 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20912 precision with floating point argument.
20913 (xststdcqp): New expand for VSX scalar test data class quad
20915 (xststdcnegqp): New expand for VSX scalar test negative quad
20917 (xststdcqp): New insn to match expansions for VSX scalar test data
20918 class quad precision and VSX scalar test negative quad precision.
20919 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20920 special case operand checking to enforce that second operand of
20921 VSX scalar test data class with quad precision argument is a 7-bit
20923 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20924 prototypes and descriptions of __ieee128 versions of
20925 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20926 scalar_test_data_class, and scalar_test_neg built-in functions.
20928 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20930 PR tree-optimization/81162
20931 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20932 replace a negate with an add.
20934 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20936 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20938 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20940 * config/arm/arm-c.c (arm_cpu_builtins): Define
20941 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20943 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20945 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20946 (armv8-r): Set ARM Cortex-R52 as default CPU.
20947 * config/arm/arm-tables.opt: Regenerate.
20948 * config/arm/arm-tune.md: Regenerate.
20949 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20951 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20952 extension for -mcpu=cortex-r52.
20954 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20956 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20957 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20958 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20959 (fp-armv8): Define it as FP_ARMv8 only.
20960 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20961 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20963 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20964 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20965 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20966 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20967 than TARGET_FPU_ARMV8.
20968 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20969 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20970 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20972 * config/arm/neon.md (neon_vrint): Likewise.
20973 (neon_vcvt): Likewise.
20974 (neon_<fmaxmin_op><mode>): Likewise.
20975 (<fmaxmin><mode>3): Likewise.
20976 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20977 * config/arm/predicates.md (arm_cond_move_operator): Check against
20978 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20980 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20982 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20983 to top of function.
20985 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20987 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20988 loop in comment with memset.
20990 2017-07-14 Martin Liska <mliska@suse.cz>
20992 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20993 * dwarf2out.c (is_java): Remove the function.
20994 (output_pubname): Remove usage of the function.
20995 (lower_bound_default): Remove usage of DW_LANG_Java.
20996 (gen_compile_unit_die): Likewise.
20997 * gcc.c: Remove compiler defaults for .java and .zip files.
20998 * gimple-expr.c (remove_suffix): Change as there's no longer
20999 extension than 4-letter one.
21000 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21001 (gimplify_save_expr): Likewise.
21002 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21003 as it's possible even for other languages than Java.
21004 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21005 * lto-opts.c (lto_write_options): Remove reference to Java.
21006 * opts.c (strip_off_ending): Update file extension handling.
21007 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21008 * tree-eh.c (lower_resx): Likewise.
21009 * tree.c (free_lang_data_in_type): Remove dead code.
21010 (find_decls_types_r): Likewise.
21011 (build_common_builtin_nodes): Remove Java from a comment.
21012 (verify_type): Remove dead code.
21013 * varasm.c (assemble_external): Remove Java from a comment.
21015 2017-07-14 Martin Liska <mliska@suse.cz>
21017 * opts.c (finish_options): Add quotes.
21018 (common_handle_option): Likewise.
21020 2017-07-14 Martin Liska <mliska@suse.cz>
21022 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21023 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21024 Remove N_SO_PASCAL.
21025 * dwarf2out.c (lower_bound_default): Do not handle
21027 (gen_compile_unit_die): Likewise.
21028 * gcc.c: Remove default extension binding for GNU Pascal.
21029 * stmt.c: Remove Pascal language from a comment.
21030 * xcoffout.c: Likewise.
21032 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21035 * diagnostic-show-locus.c (fixit_cmp): New function.
21036 (layout::layout): Sort m_fixit_hints.
21037 (column_range::column_range): Assert that the values are valid.
21038 (struct char_span): New struct.
21039 (correction::overwrite): New method.
21040 (struct source_line): New struct.
21041 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21042 calls in terms of classes source_line and char_span, and
21043 correction::overwrite.
21044 (selftest::test_overlapped_fixit_printing_2): New function.
21045 (selftest::diagnostic_show_locus_c_tests): Call it.
21047 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21049 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21050 early if there is no lhs.
21052 2017-07-13 Martin Liska <mliska@suse.cz>
21054 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21055 (gen_reference_type_die): Likewise.
21056 * stor-layout.c: Remove Pascal-related comment.
21058 2017-07-13 Martin Liska <mliska@suse.cz>
21060 * opts.c (finish_options): Add quotes to error messages.
21061 (parse_sanitizer_options): Likewise.
21063 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21065 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21067 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21069 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21071 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21073 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21075 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21077 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21080 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21081 provides the hardware capability bits, define the macro
21082 __BUILTIN_CPU_SUPPORTS__.
21083 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21084 if GLIBC does not provide the hardware capability bits. Add a
21085 gcc_unreachable call if the built-in cpu function is neither
21086 __builtin_cpu_is nor __builtin_cpu_supports.
21087 (rs6000_get_function_versions_dispatcher): Change the warning
21088 that an old GLIBC is used which does not export the capability
21089 bits to be an error.
21090 * doc/extend.texi (target_clones attribute): Document the
21091 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21092 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21093 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21094 the macros defined by GCC if the newer GLIBC is available.
21096 2017-07-12 Jeff Law <law@redhat.com>
21098 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21099 remaining includes slightly.
21100 * config/riscv/riscv-builtins.c: Include profile-count.h.
21102 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21105 * config/avr/avr.c (avr_set_current_function): In diagnostic
21106 messages: Quote keywords and (parts of) identifiers.
21107 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21110 2017-07-12 Carl Love <cel@us.ibm.com>
21112 * config/rs6000/rs6000-c.c: Add support for built-in functions
21113 vector bool char vec_revb (vector bool char);
21114 vector bool short vec_revb (vector short char);
21115 vector bool int vec_revb (vector bool int);
21116 vector bool long long vec_revb (vector bool long long);
21117 * doc/extend.texi: Update the built-in documentation file for the
21118 new built-in functions.
21120 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21122 * config/s390/s390.md: Remove movcc splitter.
21124 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21126 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21127 load/store on condition.
21129 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21132 * config/avr/avr.c (avr_encode_section_info)
21133 [progmem && !TREE_READONLY]: Error if progmem object needs
21136 2017-07-11 Michael Collison <michael.collison@arm.com>
21138 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21141 2017-07-11 Carl Love <cel@us.ibm.com>
21143 * config/rs6000/rs6000-c.c: Add support for builtins
21144 vector unsigned int vec_parity_lsbb (vector signed int);
21145 vector unsigned int vec_parity_lsbb (vector unsigned int);
21146 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21147 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21148 vector unsigned long long vec_parity_lsbb (vector signed long long);
21149 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21150 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21151 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21152 * doc/extend.texi: Update the built-in documentation file for the
21153 new built-in functions.
21155 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21157 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21158 (layout::m_primary_loc): New field.
21159 (layout::layout): Initialize new field. Move location filtering
21160 logic from here to...
21161 (layout::maybe_add_location_range): ...this new method. Add
21162 support for filtering to just the lines already specified by other
21164 (layout::will_show_line_p): New method.
21165 (gcc_rich_location::add_location_if_nearby): New method.
21166 (selftest::test_add_location_if_nearby): New test function.
21167 (selftest::diagnostic_show_locus_c_tests): Call it.
21168 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21171 2017-07-11 Tom de Vries <tom@codesourcery.com>
21173 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21174 (bb_first_real_insn): New function.
21175 (nvptx_single): Add extra initialization of broadcasted condition
21178 2017-07-11 Nathan Sidwell <nathan@acm.org>
21180 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21182 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21184 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21185 Binutils doc as TEXI will mess them up.
21186 * doc/invoke.texi (AVR Options): Same here.
21188 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21190 * config/sparc/sparc.opt (mfix-ut700): New option.
21191 (mfix-gr712rc): Likewise.
21192 (sparc_fix_b2bst): New variable.
21193 * doc/invoke.texi (SPARC options): Document them.
21194 (ARM options): Fix warnings.
21195 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21196 instructions to prevent sequences that can trigger the store-store
21197 errata for certain LEON3FT processors.
21198 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21199 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21200 * config/sparc/sparc.md (fix_b2bst): New attribute.
21201 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21203 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21206 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21208 (*rsqrtsf2_sse): Ditto.
21210 (div<mode>3): Macroize insn from divdf3 and divsf3
21211 using MODEF mode iterator.
21213 2017-07-10 Martin Sebor <msebor@redhat.com>
21215 PR tree-optimization/80397
21216 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21217 instead of testing for equality to INTEGER_TYPE.
21219 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21221 * config.gcc: Remove uclibc from arc target spec.
21223 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21225 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21227 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21230 * lto-wrapper.c (remove_option): New function.
21231 (merge_and_complain): Merge PIC/PIE options more realistically.
21233 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21235 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21239 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21240 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21241 * config.in: Regenerate.
21242 * configure: Regenerate.
21243 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21244 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21245 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21246 (TARGET_GASISR_PROLOGUES): ...target mask.
21247 * common/config/avr/avr-common.c
21248 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21249 Set -mgas-isr-prologues.
21250 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21251 INSERT_PASS_BEFORE for it.
21252 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21253 * config/avr/avr.c (avr_option_override)
21254 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21255 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21256 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21257 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21258 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21260 (make_avr_pass_pre_proep): New function.
21261 (emit_push_sfr) <treg>: Add argument to function and use it
21262 instead of TMP_REG.
21263 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21264 and set machine->gasisr.yes.
21265 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21266 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21267 __gcc_isr.n_pushed to .L__stack_usage.
21268 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21269 (avr_asm_final_postscan_insn): ...this new static function.
21270 * config/avr/avr.h (machine_function)
21271 <is_no_gccisr, use_L__stack_usage>: New fields.
21272 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21273 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21274 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21275 (gasisr, *gasisr): New expander and insn.
21276 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21277 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21278 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21280 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21282 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21285 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21287 Move jump-tables out of .text again.
21290 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21291 (ASM_OUTPUT_ADDR_VEC): New function.
21292 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21293 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21294 INSN_ADDRESSes as asm comment.
21295 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21296 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21297 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21298 * config/avr/avr.md (*tablejump): Adjust comment.
21299 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21300 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21302 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21303 (avr_output_addr_vec): New proto.
21304 (avr_log_t) <insn_addresses>: New field.
21306 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21309 * config/i386/i386.c (ix86_function_arg_advance): Set
21310 outgoing_args_on_stack to true if there are outgoing arguments
21312 (ix86_function_arg): Likewise.
21313 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21314 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21315 * config/i386/i386.h (machine_function): Add
21316 outgoing_args_on_stack.
21318 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21320 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21321 supporting pthreds.
21322 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21324 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21326 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21327 (REAL_H): Remove $(MACHMODE_H).
21328 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21330 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21331 $(MACHMODE_H) and double-int.h.
21332 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21334 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21337 2017-07-07 Andrew Pinski <apinski@cavium.com>
21339 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21340 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21342 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21344 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21345 Add warning if GCC was not configured to link against a GLIBC that
21346 exports the hardware capability bits.
21347 (make_resolver_func): Make resolver function private and not a
21348 COMDAT function. Create the name with clone_function_name instead
21349 of make_unique_name.
21352 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21353 correct operand in doing the split.
21355 2017-07-07 Carl Love <cel@us.ibm.com>
21357 * config/rs6000/rs6000-c: Add support for built-in function
21358 vector unsigned short vec_pack_to_short_fp32 (vector float,
21360 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21361 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21362 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21363 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21364 (convert_4f32_8i16): Add define_expand.
21365 * doc/extend.texi: Update the built-in documentation file for the
21366 new built-in function.
21368 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21370 * config/sparc/m8.md: New file.
21371 * config/sparc/sparc.md: Include m8.md.
21373 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21375 * config/sparc/sparc.opt: New option -mvis4b.
21376 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21377 (sparc_option_override): Handle VIS4B.
21378 (enum sparc_builtins): Define
21379 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21380 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21381 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21382 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21383 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21384 (check_constant_argument): New function.
21385 (sparc_vis_init_builtins): Define builtins
21386 __builtin_vis_dictunpack{8,16,32},
21387 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21388 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21389 __builtin_vis_fpcmpde{8,16,32}shl and
21390 __builtin_vis_fpcmpur{8,16,32}shl.
21391 (sparc_expand_builtin): Check that the constant operands to
21392 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21393 constant and in range.
21394 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21396 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21397 (SPARC_IMM5_P): Likewise.
21398 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21399 (enabled): Handle vis4b.
21400 (UNSPEC_DICTUNPACK): New unspec.
21401 (UNSPEC_FPCMPSHL): Likewise.
21402 (UNSPEC_FPUCMPSHL): Likewise.
21403 (UNSPEC_FPCMPDESHL): Likewise.
21404 (UNSPEC_FPCMPURSHL): Likewise.
21405 (cpu_feature): New CPU feature `vis4b'.
21406 (dictunpack{8,16,32}): New insns.
21407 (FPCSMODE): New mode iterator.
21408 (fpcscond): New code iterator.
21409 (fpcsucond): Likewise.
21410 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21411 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21412 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21413 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21414 * config/sparc/constraints.md: Define constraints `q' for unsigned
21415 2-bit integer constants and `t' for unsigned 5-bit integer
21417 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21419 (imm5_operand_dictunpack16): Likewise.
21420 (imm5_operand_dictunpack32): Likewise.
21421 (imm2_operand): Likewise.
21422 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21423 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21424 ditunpack* and fpcmp*shl builtins.
21426 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21428 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21429 * config.in: Add HAVE_AS_SPARC6 define.
21430 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21432 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21434 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21435 (CPP_CPU_SPEC): Handle m8.
21436 (ASM_CPU_SPEC): Likewise.
21437 * config/sparc/sparc-opts.h (enum processor_type): Add
21439 * config/sparc/sparc.c (m8_costs): New struct.
21440 (sparc_option_override): Handle TARGET_CPU_m8.
21441 (sparc32_initialize_trampoline): Likewise.
21442 (sparc64_initialize_trampoline): Likewise.
21443 (sparc_issue_rate): Likewise.
21444 (sparc_register_move_cost): Likewise.
21445 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21446 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21447 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21448 (CPP_CPU_SPEC): Handle M8.
21449 (ASM_CPU_SPEC): Likewise.
21450 (AS_M8_FLAG): Define.
21451 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21452 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21453 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21455 * configure: Regenerate.
21456 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21459 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21461 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21463 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21464 ("*movdi_insn_sp32"): Do not set v3pipe.
21465 ("*movsi_insn"): Likewise.
21466 ("*movdi_insn_sp64"): Likewise.
21467 ("*movsf_insn"): Likewise.
21468 ("*movdf_insn_sp32"): Likewise.
21469 ("*movdf_insn_sp64"): Likewise.
21470 ("*zero_extendsidi2_insn_sp64"): Likewise.
21471 ("*sign_extendsidi2_insn"): Likewise.
21472 ("*mov<VM32:mode>_insn"): Likewise.
21473 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21474 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21475 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21476 ("<vlop:code><VL:mode>3"): Likewise.
21477 ("*not_<vlop:code><VL:mode>3"): Likewise.
21478 ("*nand<VL:mode>_vis"): Likewise.
21479 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21480 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21481 ("one_cmpl<VL:mode>2"): Likewise.
21482 ("faligndata<VM64:mode>_vis"): Likewise.
21483 ("alignaddrsi_vis"): Likewise.
21484 ("alignaddrdi_vis"): Likweise.
21485 ("alignaddrlsi_vis"): Likewise.
21486 ("alignaddrldi_vis"): Likewise.
21487 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21488 ("bmaskdi_vis"): Likewise.
21489 ("bmasksi_vis"): Likewise.
21490 ("bshuffle<VM64:mode>_vis"): Likewise.
21491 ("cmask8<P:mode>_vis"): Likewise.
21492 ("cmask16<P:mode>_vis"): Likewise.
21493 ("cmask32<P:mode>_vis"): Likewise.
21494 ("pdistn<P:mode>_vis"): Likewise.
21495 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21497 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21499 * config/sparc/sparc.md ("subtype"): New insn attribute.
21500 ("*wrgsr_sp64"): Set insn subtype.
21501 ("*rdgsr_sp64"): Likewise.
21502 ("alignaddrsi_vis"): Likewise.
21503 ("alignaddrdi_vis"): Likewise.
21504 ("alignaddrlsi_vis"): Likewise.
21505 ("alignaddrldi_vis"): Likewise.
21506 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21507 ("fexpand_vis"): Likewise.
21508 ("fpmerge_vis"): Likewise.
21509 ("faligndata<VM64:mode>_vis"): Likewise.
21510 ("bshuffle<VM64:mode>_vis"): Likewise.
21511 ("cmask8<P:mode>_vis"): Likewise.
21512 ("cmask16<P:mode>_vis"): Likewise.
21513 ("cmask32<P:mode>_vis"): Likewise.
21514 ("fchksm16_vis"): Likewise.
21515 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21516 ("fmean16_vis"): Likewise.
21517 ("fp<plusminus_insn>64_vis"): Likewise.
21518 ("<plusminus_insn>v8qi3"): Likewise.
21519 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21520 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21521 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21522 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21523 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21524 ("*movqi_insn"): Likewise.
21525 ("*movhi_insn"): Likewise.
21526 ("*movsi_insn"): Likewise.
21527 ("movsi_pic_gotdata_op"): Likewise.
21528 ("*movdi_insn_sp32"): Likewise.
21529 ("*movdi_insn_sp64"): Likewise.
21530 ("movdi_pic_gotdata_op"): Likewise.
21531 ("*movsf_insn"): Likewise.
21532 ("*movdf_insn_sp32"): Likewise.
21533 ("*movdf_insn_sp64"): Likewise.
21534 ("*zero_extendhisi2_insn"): Likewise.
21535 ("*zero_extendqihi2_insn"): Likewise.
21536 ("*zero_extendqisi2_insn"): Likewise.
21537 ("*zero_extendqidi2_insn"): Likewise.
21538 ("*zero_extendhidi2_insn"): Likewise.
21539 ("*zero_extendsidi2_insn_sp64"): Likewise.
21540 ("ldfsr"): Likewise.
21541 ("prefetch_64"): Likewise.
21542 ("prefetch_32"): Likewise.
21543 ("tie_ld32"): Likewise.
21544 ("tie_ld64"): Likewise.
21545 ("*tldo_ldub_sp32"): Likewise.
21546 ("*tldo_ldub1_sp32"): Likewise.
21547 ("*tldo_ldub2_sp32"): Likewise.
21548 ("*tldo_ldub_sp64"): Likewise.
21549 ("*tldo_ldub1_sp64"): Likewise.
21550 ("*tldo_ldub2_sp64"): Likewise.
21551 ("*tldo_ldub3_sp64"): Likewise.
21552 ("*tldo_lduh_sp32"): Likewise.
21553 ("*tldo_lduh1_sp32"): Likewise.
21554 ("*tldo_lduh_sp64"): Likewise.
21555 ("*tldo_lduh1_sp64"): Likewise.
21556 ("*tldo_lduh2_sp64"): Likewise.
21557 ("*tldo_lduw_sp32"): Likewise.
21558 ("*tldo_lduw_sp64"): Likewise.
21559 ("*tldo_lduw1_sp64"): Likewise.
21560 ("*tldo_ldx_sp64"): Likewise.
21561 ("*mov<VM32:mode>_insn"): Likewise.
21562 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21563 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21565 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21567 * config/sparc/sparc.md ("type"): New insn type viscmp.
21568 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21570 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21571 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21572 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21573 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21575 ("n7_vis_logical_11cycle"): Likewise.
21576 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21577 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21578 * config/sparc/niagara.md ("niag_vis"): Likewise.
21579 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21580 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21582 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21584 * config/sparc/sparc.md: New instruction type `bmask'.
21585 (bmaskdi_vis): Use the `bmask' type.
21586 (bmasksi_vis): Likewise.
21587 * config/sparc/ultra3.md (us3_array): Likewise.
21588 * config/sparc/niagara7.md (n7_array): Likewise.
21589 * config/sparc/niagara4.md (n4_array): Likewise.
21590 * config/sparc/niagara2.md (niag2_vis): Likewise.
21591 (niag3_vis): Likewise.
21592 * config/sparc/niagara.md (niag_vis): Likewise.
21594 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21596 * ipa-comdats.c: Remove optimize check from gate.
21597 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21598 for functions not optimized.
21599 (ipa_fn_summary_read): Skip optimize check.
21600 (ipa_fn_summary_write): Likewise.
21601 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21603 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21605 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21606 (check_callers): Give up on caller which is not optimized.
21607 (inline_small_functions): Likewise.
21608 (ipa_inline): Do not give up when not optimizing.
21609 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21610 away unoptimizes cdtors.
21611 (whole_program_function_and_variable_visibility): Do
21612 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21613 * ipa.c (process_references): Do not check optimize.
21614 (symbol_table::remove_unreachable_nodes): Update optimize check.
21615 (set_writeonly_bit): Update optimize check.
21616 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21617 (pass_ipa_single_use::gate): Remove.
21619 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21621 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21622 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21623 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21624 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21625 permute_load, permute_store, adjust_extract, adjust_splat,
21626 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21627 replace_swap_with_copy, dump_swap_insn_table,
21628 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21629 recombine_lvx_pattern, recombine_stvx_pattern,
21630 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21631 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21632 to file rs6000-p8swap.c.
21633 * config/rs6000/rs6000-p8swap.c: New file.
21634 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21635 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21636 and rs6000*-*-* targets.
21638 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21640 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21642 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21644 * lto-wrapper.c (merge_and_complain): Do not merge
21645 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21646 fsigned_zeros, ftrapping_math, fwrapv.
21647 (append_compiler_options): Do not track these options.
21648 (append_linker_options): Likewie
21650 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21652 * cgraphunit.c (cgraph_node::finalize_function): When
21653 !flag_toplevel_reorde set no_reorder flag.
21654 (varpool_node::finalize_decl): Likewise.
21655 (symbol_table::compile): Drop no toplevel reorder path.
21657 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21659 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21660 edges; zero probability is not better than uninitialized.
21662 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21664 * asan.h (asan_sanitize_allocas_p): Declare.
21665 * asan.c (asan_sanitize_allocas_p): New function.
21666 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21667 (handle_builtin_alloca): Likewise.
21668 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21669 if !asan_sanitize_allocas_p.
21670 * params.def (asan-instrument-allocas): Add new option.
21671 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21672 * opts.c (common_handle_option): Disable allocas sanitization for
21675 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21677 * asan.c: Include gimple-fold.h.
21678 (get_last_alloca_addr): New function.
21679 (handle_builtin_stackrestore): Likewise.
21680 (handle_builtin_alloca): Likewise.
21681 (asan_emit_allocas_unpoison): Likewise.
21682 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21683 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21684 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21685 (instrument_builtin_call): Pass gimple iterator to
21686 get_mem_refs_of_builtin_call.
21687 (last_alloca_addr): New global.
21688 * asan.h (asan_emit_allocas_unpoison): Declare.
21689 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21690 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21691 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21692 if function calls alloca.
21693 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21694 * gimple-fold.h (replace_call_with_value): Declare.
21695 * internal-fn.c: Include asan.h.
21696 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21697 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21699 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21701 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21702 (C_SELFTEST_FLAGS): New.
21703 (CPP_SELFTEST_FLAGS): New.
21704 (SELFTEST_DEPS): New, from deps of s-selftest.
21705 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21706 (CPP_SELFTEST_DEPS): New.
21707 (selftest): Add dependency on s-selftest-c++.
21708 (s-selftest): Rename to...
21709 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21710 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21711 than SELFTEST_FLAGS.
21712 (selftest-gdb): Rename to...
21713 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21715 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21716 (selftest-valgrind): Rename to...
21717 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21719 (selftest-valgrind): Reintroduce as an alias for
21720 selftest-c-valgrind.
21721 (s-selftest-c++): New.
21722 (selftest-c++-gdb): New.
21723 (selftest-c++-valgrind): New.
21725 2017-07-06 Olivier Hainque <hainque@adacore.com>
21727 * gcc.c (process_command): When deciding if undefined variables
21728 should be ignored when processing specs, accept "gcc -v" as well.
21730 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21732 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21733 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21735 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21737 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21738 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21739 * config/arm/arm-tables.opt: Regenerate.
21740 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21742 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21744 2017-07-06 Carl Love <cel@us.ibm.com>
21746 * ChangeLog: Clean up from mid air collision
21748 2017-07-06 Carl Love <cel@us.ibm.com>
21750 * config/rs6000/rs6000-c.c: Add support for built-in functions
21751 vector signed int vec_subc (vector signed int, vector signed int);
21752 vector signed __int128 vec_subc (vector signed __int128,
21753 vector signed __int128);
21754 vector unsigned __int128 vec_subc (vector unsigned __int128,
21755 vector unsigned __int128);
21756 vector signed int vec_sube (vector signed int, vector signed int,
21757 vector signed int);
21758 vector unsigned int vec_sube (vector unsigned int,
21759 vector unsigned int,
21760 vector unsigned int);
21761 vector signed __int128 vec_sube (vector signed __int128,
21762 vector signed __int128,
21763 vector signed__int128);
21764 vector unsigned __int128 vec_sube (vector unsigned __int128,
21765 vector unsigned __int128,
21766 vector unsigned __int128);
21767 vector signed int vec_subec (vector signed int, vector signed int,
21768 vector signed int);
21769 vector unsigned int vec_subec (vector unsigned int,
21770 vector unsigned int,
21771 vector unsigned int);
21772 vector signed __int128 vec_subec (vector signed __int128,
21773 vector signed __int128,
21774 vector signed__int128);
21775 vector unsigned __int128 vec_subec (vector unsigned __int128,
21776 vector unsigned __int128,
21777 vector unsigned __int128);
21778 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21779 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21780 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21781 BU_ALTIVEC_OVERLOAD_X definitions.
21782 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21783 * doc/extend.texi: Update the built-in documentation file for the new
21784 built-in functions.
21786 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21789 * diagnostic-show-locus.c (layout::layout): Use start and finish
21790 spelling location for the start and finish of each range.
21791 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21792 Add unused aspect param.
21793 * input.c (expand_location_1): Add "aspect" param, and use it
21794 to access the correct part of the location.
21795 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21797 (expand_location_to_spelling_point): Likewise.
21798 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21799 param, and pass it to expand_location_1.
21801 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21803 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21804 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21805 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21806 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21807 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21808 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21809 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21810 _mm_maskz_getmant_ss): New intrinsics.
21811 (__builtin_ia32_getexpss128_mask): Changed to ...
21812 __builtin_ia32_getexpss128_round ... this.
21813 (__builtin_ia32_getexpsd128_mask): Changed to ...
21814 __builtin_ia32_getexpsd128_round ... this.
21815 * config/i386/i386-builtin-types.def
21816 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21817 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21818 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21819 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21820 __builtin_ia32_getmantss_mask_round): New builtins.
21821 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21822 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21823 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21824 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21825 * config/i386/sse.md
21826 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21827 avx512f_sgetexp<mode><mask_scalar_name>
21828 <round_saeonly_scalar_name> ... this.
21829 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21830 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21831 vgetexp<ssescalarmodesuffix>
21832 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21833 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21834 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21835 avx512f_vgetmant<mode><mask_scalar_name>
21836 <round_saeonly_scalar_name> ... this.
21837 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21838 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21839 vgetmant<ssescalarmodesuffix>
21840 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21841 %0<mask_scalar_operand4>, %1, %2
21842 <round_saeonly_scalar_mask_op4>, %3} ... this.
21843 * config/i386/subst.md (mask_scalar_operand4,
21844 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21845 round_saeonly_scalar_nimm_predicate): New subst attributes.
21847 2017-07-06 Julia Koval <julia.koval@intel.com>
21849 * config/i386/i386.c (ix86_erase_embedded_rounding):
21850 Remove code for old rounding pattern.
21852 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21854 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21856 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21858 * doc/sourcebuild.texi (Test Directives, Variants of
21859 dg-require-support): Add documentation for dg-require-stack-check.
21861 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21863 * config/i386/subst.md (mask_scalar, round_scalar,
21864 round_saeonly_scalar): New meta-templates.
21865 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21866 round_scalar_mask_operand3, round_scalar_mask_op3,
21867 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21868 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21869 round_saeonly_scalar_constraint,
21870 round_saeonly_scalar_prefix): New subst attribute.
21871 * config/i386/sse.md
21872 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21873 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21874 <round_scalar_name> ... this.
21875 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21876 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21877 <round_scalar_name> ... this.
21878 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21879 <sse>_vm<code><mode>3<mask_scalar_name>
21880 <round_saeonly_scalar_name> ... this.
21881 (v<plusminus_mnemonic><ssescalarmodesuffix>
21882 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21883 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21884 v<plusminus_mnemonic><ssescalarmodesuffix>
21885 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21886 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21887 (v<multdiv_mnemonic><ssescalarmodesuffix>
21888 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21889 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21890 v<multdiv_mnemonic><ssescalarmodesuffix>
21891 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21892 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21893 (v<maxmin_float><ssescalarmodesuffix>
21894 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21895 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21896 v<maxmin_float><ssescalarmodesuffix>
21897 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21898 %0<mask_scalar_operand3>, %1, %<iptr>2
21899 <round_saeonly_scalar_mask_op3>} ... this.
21901 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21903 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21904 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21906 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21907 Alan Hayward <alan.hayward@arm.com>
21908 David Sherwood <david.sherwood@arm.com>
21910 * combine.c (simplify_if_then_else): Remove "enum" before
21912 * compare-elim.c (can_eliminate_compare): Likewise.
21913 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21915 (aarch64_lookup_simd_builtin_type): Likewise.
21916 (aarch64_simd_builtin_type): Likewise.
21917 (aarch64_init_simd_builtin_types): Likewise.
21918 (aarch64_simd_expand_args): Likewise.
21919 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21921 (aarch64_reverse_mask): Likewise.
21922 (aarch64_simd_emit_reg_reg_move): Likewise.
21923 (aarch64_gen_adjusted_ldpstp): Likewise.
21924 (aarch64_ccmp_mode_to_code): Likewise.
21925 (aarch64_operands_ok_for_ldpstp): Likewise.
21926 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21927 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21929 (aarch64_min_divisions_for_recip_mul): Likewise.
21930 (aarch64_reassociation_width): Likewise.
21931 (aarch64_get_condition_code_1): Likewise.
21932 (aarch64_simd_emit_reg_reg_move): Likewise.
21933 (aarch64_simd_attr_length_rglist): Likewise.
21934 (aarch64_reverse_mask): Likewise.
21935 (aarch64_operands_ok_for_ldpstp): Likewise.
21936 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21937 (aarch64_gen_adjusted_ldpstp): Likewise.
21938 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21940 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21941 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21942 (arm_lookup_simd_builtin_type): Likewise.
21943 (arm_simd_builtin_type): Likewise.
21944 (arm_init_simd_builtin_types): Likewise.
21945 (arm_expand_builtin_args): Likewise.
21946 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21947 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21948 (ft32_setup_incoming_varargs): Likewise.
21949 (ft32_function_arg): Likewise.
21950 (ft32_function_arg_advance): Likewise.
21951 (ft32_pass_by_reference): Likewise.
21952 (ft32_arg_partial_bytes): Likewise.
21953 (ft32_valid_pointer_mode): Likewise.
21954 (ft32_addr_space_pointer_mode): Likewise.
21955 (ft32_addr_space_legitimate_address_p): Likewise.
21956 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21958 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21959 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21960 (iamcu_alignment): Likewise.
21961 (canonicalize_vector_int_perm): Likewise.
21962 (ix86_noce_conversion_profitable_p): Likewise.
21963 (ix86_mpx_bound_mode): Likewise.
21964 (ix86_operands_ok_for_move_multiple): Likewise.
21965 * config/microblaze/microblaze-protos.h
21966 (microblaze_expand_conditional_branch_reg): Likewise.
21967 * config/microblaze/microblaze.c
21968 (microblaze_expand_conditional_branch_reg): Likewise.
21969 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21971 (rs6000_reassociation_width): Likewise.
21972 (rs6000_invalid_binary_op): Likewise.
21973 (fusion_p9_p): Likewise.
21974 (emit_fusion_p9_load): Likewise.
21975 (emit_fusion_p9_store): Likewise.
21976 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21978 (riscv_hard_regno_mode_ok_p): Likewise.
21979 (riscv_address_insns): Likewise.
21980 (riscv_split_symbol): Likewise.
21981 (riscv_legitimize_move): Likewise.
21982 (riscv_function_value): Likewise.
21983 (riscv_hard_regno_nregs): Likewise.
21984 (riscv_expand_builtin): Likewise.
21985 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21986 (riscv_build_integer): Likewise.
21987 (riscv_split_integer): Likewise.
21988 (riscv_legitimate_constant_p): Likewise.
21989 (riscv_cannot_force_const_mem): Likewise.
21990 (riscv_regno_mode_ok_for_base_p): Likewise.
21991 (riscv_valid_base_register_p): Likewise.
21992 (riscv_valid_offset_p): Likewise.
21993 (riscv_valid_lo_sum_p): Likewise.
21994 (riscv_classify_address): Likewise.
21995 (riscv_legitimate_address_p): Likewise.
21996 (riscv_address_insns): Likewise.
21997 (riscv_load_store_insns): Likewise.
21998 (riscv_force_binary): Likewise.
21999 (riscv_split_symbol): Likewise.
22000 (riscv_force_address): Likewise.
22001 (riscv_legitimize_address): Likewise.
22002 (riscv_move_integer): Likewise.
22003 (riscv_legitimize_const_move): Likewise.
22004 (riscv_legitimize_move): Likewise.
22005 (riscv_address_cost): Likewise.
22006 (riscv_subword): Likewise.
22007 (riscv_output_move): Likewise.
22008 (riscv_canonicalize_int_order_test): Likewise.
22009 (riscv_emit_int_order_test): Likewise.
22010 (riscv_function_arg_boundary): Likewise.
22011 (riscv_pass_mode_in_fpr_p): Likewise.
22012 (riscv_pass_fpr_single): Likewise.
22013 (riscv_pass_fpr_pair): Likewise.
22014 (riscv_get_arg_info): Likewise.
22015 (riscv_function_arg): Likewise.
22016 (riscv_function_arg_advance): Likewise.
22017 (riscv_arg_partial_bytes): Likewise.
22018 (riscv_function_value): Likewise.
22019 (riscv_pass_by_reference): Likewise.
22020 (riscv_setup_incoming_varargs): Likewise.
22021 (riscv_print_operand): Likewise.
22022 (riscv_elf_select_rtx_section): Likewise.
22023 (riscv_save_restore_reg): Likewise.
22024 (riscv_for_each_saved_reg): Likewise.
22025 (riscv_register_move_cost): Likewise.
22026 (riscv_hard_regno_mode_ok_p): Likewise.
22027 (riscv_hard_regno_nregs): Likewise.
22028 (riscv_class_max_nregs): Likewise.
22029 (riscv_memory_move_cost): Likewise.
22030 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22031 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22032 (rl78_addr_space_address_mode): Likewise.
22033 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22035 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22036 (rs6000_reassociation_width): Likewise.
22037 (rs6000_invalid_binary_op): Likewise.
22038 (fusion_p9_p): Likewise.
22039 (emit_fusion_p9_load): Likewise.
22040 (emit_fusion_p9_store): Likewise.
22041 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22042 (ok_for_simple_move_operands): Likewise.
22043 (ok_for_simple_move_strict_operands): Likewise.
22044 (ok_for_simple_arith_logic_operands): Likewise.
22045 (visium_legitimize_reload_address): Likewise.
22046 (visium_select_cc_mode): Likewise.
22047 (output_cbranch): Likewise.
22048 (visium_split_double_move): Likewise.
22049 (visium_expand_copysign): Likewise.
22050 (visium_expand_int_cstore): Likewise.
22051 (visium_expand_fp_cstore): Likewise.
22052 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22053 (visium_function_arg): Likewise.
22054 (visium_function_arg_advance): Likewise.
22055 (visium_libcall_value): Likewise.
22056 (visium_setup_incoming_varargs): Likewise.
22057 (visium_legitimate_constant_p): Likewise.
22058 (visium_legitimate_address_p): Likewise.
22059 (visium_legitimize_address): Likewise.
22060 (visium_secondary_reload): Likewise.
22061 (visium_register_move_cost): Likewise.
22062 (visium_memory_move_cost): Likewise.
22063 (prepare_move_operands): Likewise.
22064 (ok_for_simple_move_operands): Likewise.
22065 (ok_for_simple_move_strict_operands): Likewise.
22066 (ok_for_simple_arith_logic_operands): Likewise.
22067 (visium_function_value_1): Likewise.
22068 (rtx_ok_for_offset_p): Likewise.
22069 (visium_legitimize_reload_address): Likewise.
22070 (visium_split_double_move): Likewise.
22071 (visium_expand_copysign): Likewise.
22072 (visium_expand_int_cstore): Likewise.
22073 (visium_expand_fp_cstore): Likewise.
22074 (visium_split_cstore): Likewise.
22075 (visium_select_cc_mode): Likewise.
22076 (visium_split_cbranch): Likewise.
22077 (output_cbranch): Likewise.
22078 (visium_print_operand_address): Likewise.
22079 * expmed.c (flip_storage_order): Likewise.
22080 * expmed.h (emit_cstore): Likewise.
22081 (flip_storage_order): Likewise.
22082 * genrecog.c (validate_pattern): Likewise.
22083 * hsa-gen.c (gen_hsa_addr): Likewise.
22084 * internal-fn.c (expand_arith_overflow): Likewise.
22085 * ira-color.c (allocno_copy_cost_saving): Likewise.
22086 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22087 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22088 (process_invariant_for_inheritance): Likewise.
22089 * lra-eliminations.c (move_plus_up): Likewise.
22090 * omp-low.c (lower_oacc_reductions): Likewise.
22091 * simplify-rtx.c (simplify_subreg): Likewise.
22092 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22093 (TARGET_CHKP_BOUND_MODE): Likewise..
22094 * targhooks.c (default_chkp_bound_mode): Likewise.
22095 (default_setup_incoming_vararg_bounds): Likewise.
22096 * targhooks.h (default_chkp_bound_mode): Likewise.
22097 (default_setup_incoming_vararg_bounds): Likewise.
22098 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22099 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22100 (have_whole_vector_shift): Likewise.
22101 * tree-vect-stmts.c (vectorizable_load): Likewise.
22102 * doc/tm.texi: Regenerate.
22104 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22106 Graceful degrade if Binutils PR21472 is not available.
22109 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22110 .rodata in flash test fails.
22111 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22112 * confgure: Regenerate.
22113 * config.in: Regenerate.
22114 * config/avr/avr.c (avr_asm_named_section)
22115 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22116 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22117 (avr_asm_init_sections): Same.
22119 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22121 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22122 (fma<VH:mode>4_intrinsic): Likewise.
22123 (*fmsub<VCVTF:mode>4): Likewise.
22124 (*fmsub<VH:mode>4_intrinsic): Likewise.
22126 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22129 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22130 Don't depend on "optimize > 0".
22131 (out_movhi_r_mr, out_movqi_mr_r): Same.
22132 (out_movhi_mr_r, out_movqi_r_mr): Same.
22133 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22134 io_address_operand on "optimize > 0".
22136 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22138 * tree-loop-distribution.c: Add general explanantion on the pass.
22139 (generate_loops_for_partition): Mark distributed loop.
22140 (pg_add_dependence_edges): New parameter. Handle alias data
22141 dependence specially and record it in the parameter if asked.
22142 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22143 (init_partition_graph_vertices, add_partition_graph_edge): New.
22144 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22145 (free_partition_graph_vdata, build_partition_graph): New.
22146 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22147 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22148 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22149 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22150 (version_for_distribution_p, finalize_partitions): New.
22151 (distribute_loop): Handle alias data dependence specially. Factor
22152 out loop fusion code as functions and call these functions.
22154 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22156 * tree-loop-distribution.c (classify_partition): New parameter and
22157 better handle reduction statement.
22158 (rdg_build_partitions): Revise comment.
22159 (distribute_loop): Compute statements in all partitions and pass it
22160 to classify_partition.
22162 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22164 * tree-loop-distribution.c (enum partition_type): New.
22165 (struct partition): New field type.
22166 (partition_merge_into): Add parameter. Update partition type.
22167 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22168 (build_rdg_partition_for_vertex): Compute partition type.
22169 (rdg_build_partitions): Dump partition type.
22170 (distribute_loop): Update calls to partition_merge_into.
22172 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22174 * tree-loop-distribution.c (struct ddr_hasher): New.
22175 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22177 (classify_partition): Call get_data_dependence.
22178 (pg_add_dependence_edges): Ditto.
22179 (distribute_loop): Release data dependence hash table.
22181 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22183 * tree-loop-distribution.c (ref_base_address): Delete.
22184 (similar_memory_accesses): Rename ...
22185 (share_memory_accesses): ... to this. Check if partitions access
22186 the same memory reference.
22187 (distribute_loop): Call share_memory_accesses.
22189 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22191 * tree-loop-distribution.c (struct partition): New field recording
22192 its data reference.
22193 (partition_alloc, partition_free): Init and release data refs.
22194 (partition_merge_into): Merge data refs.
22195 (build_rdg_partition_for_vertex): Collect data refs for partition.
22196 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22198 (distribute_loop): Remve data refs from vertice data of partition
22201 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22203 * tree-loop-distribution.c (params.h): Include header file.
22204 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22205 (datarefs_vec): New global var.
22206 (create_rdg_vertices): Use datarefs_vec directly.
22207 (free_rdg): Don't free data references.
22208 (build_rdg): Update use. Don't free data references.
22209 (distribute_loop): Compute global variable for data references.
22210 Bail out if there are too many data references.
22212 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22214 * tree-loop-distribution.c (loop_nest): New global var.
22215 (build_rdg): Use loop directly, rather than loop nest.
22216 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22218 (distribute_loop): Compute global variable loop nest. Update use.
22220 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22222 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22223 (partition_merge_into): New parameter. Dump reason for fusion.
22224 (distribute_loop): Update use of partition_merge_into.
22226 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22228 * tree-loop-distribution.c (bb_top_order_index): New.
22229 (bb_top_order_index_size, bb_top_order_cmp): New.
22230 (stmts_from_loop): Use topological order.
22231 (pass_loop_distribution::execute): Compute and release topological
22232 order for basic blocks.
22234 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22236 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22239 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22241 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22242 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22243 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22244 * internal-fn.def (LOOP_DIST_ALIAS): New.
22245 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22246 (fold_loop_internal_call): ... this.
22247 (vect_loop_dist_alias_call): New function.
22248 (set_uid_loop_bbs): Call fold_loop_internal_call.
22249 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22252 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22255 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22256 Require dead FLAGS_REG at the beginning of a peephole.
22258 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22261 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22262 arguments in the call to __builtin_ia32_sbb_u32.
22263 (_subborrow_u64): Swap _X and _Y arguments in the call to
22264 __builtin_ia32_sbb_u64.
22266 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22269 * tree-vrp.c (compare_assert_loc): Turn into a function template
22270 with stable template parameter. Only test if a->e is NULL,
22271 !a->e == !b->e has been verified already. Use e == NULL or
22272 e != NULL instead of e or ! e tests. If stable is true, don't use
22273 iterative_hash_expr, on the other side allow a or b or both NULL
22274 and sort the NULLs last.
22275 (process_assert_insertions): Sort using compare_assert_loc<false>
22276 instead of compare_assert_loc, later sort using
22277 compare_assert_loc<true> before calling process_assert_insertions_for
22278 in a loop. Use break instead of continue once seen NULL pointer.
22280 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22282 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22283 Cortex-R7 and Cortex-R8 processors.
22285 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22287 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22288 uninitialized while src is not.
22290 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22292 * common/config/arm/arm-common.c: Adjust include path for
22294 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22295 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22296 (arm-cpu-data.h): Likewise.
22297 (arm-cpu-cdata.h): Likewise.
22298 * config/arm/arm-cpu.h: Delete.
22299 * config/arm/arm-cpu-cdata.h: Delete.
22300 * config/arm/arm-cpu-data.h: Delete.
22302 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22304 * config/arm/arm-cpus.in (cortex-a55): New.
22305 (cortex-a75): Likewise.
22306 (cortex-a75.cortex-a55): Likewise.
22307 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22309 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22310 * config/arm/arm-cpu-cdata.h: Regenerate.
22311 * config/arm/arm-cpu-data.h: Regenerate.
22312 * config/arm/arm-cpu.h: Regenerate.
22313 * config/arm/arm-tables.opt: Regenerate.
22314 * config/arm/arm-tune.md: Regenerate.
22316 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22318 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22320 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22322 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22325 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22327 PR tree-optimization/81292
22328 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22329 full_string_p, also call adjust_related_strinfos if the adjustment
22330 is simple, otherwise invalidate related strinfos.
22332 2017-07-04 Martin Liska <mliska@suse.cz>
22335 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22336 newly created variable as DECL_IGNORED_P.
22338 2017-07-04 Martin Liska <mliska@suse.cz>
22341 * ipa-inline.c (inline_small_functions):
22342 Use xstrdup_for_dump.
22344 2017-07-04 Tom de Vries <tom@codesourcery.com>
22346 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22348 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22351 * config/darwin.c (darwin_function_switched_text_sections):
22354 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22356 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22358 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22360 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22362 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22364 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22365 min_profitable_iters, and th as inclusive lower bounds.
22366 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22367 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22368 for min_profitable_iters and min_profitable_estimate.
22369 (vect_transform_loop): Treat th as an inclusive lower bound.
22370 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22372 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22375 * config/darwin.c (darwin_function_switched_text_sections):
22376 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22377 in two pieces, and suppress the use of buf.
22379 2017-07-03 Nathan Sidwell <nathan@acm.org>
22381 * hash-table.h (hash_table_mod1): Fix indentation.
22383 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22385 PR middle-end/81290
22386 * predict.c (force_edge_cold): Be more careful about propagation
22388 * profile-count.h (profile_probability::guessed,
22389 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22391 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22393 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22395 * doc/invoke.texi (rcpc architecture extension): Document it.
22397 2017-07-03 Richard Biener <rguenther@suse.de>
22399 PR tree-optimization/60510
22400 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22401 the scalar reduction PHI and use it.
22402 (vectorizable_reduction): Properly guard the single_defuse_cycle
22403 path for non-SLP reduction chains where we cannot use it.
22404 Rework reduc_def/index and vector type deduction. Rework
22405 vector operand gathering during reduction op code-gen.
22406 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22407 chains dissolve the chain and leave it to non-SLP reduction
22410 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22412 * tree-data-ref.h (dr_alignment): Declare.
22413 * tree-data-ref.c (dr_alignment): New function.
22414 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22415 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22417 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22419 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22421 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22422 and base_misalignment fields.
22423 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22424 * tree-data-ref.c: Include builtins.h.
22425 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22426 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22427 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22428 * tree-vect-data-refs.c: Include tree-cfg.h.
22429 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22430 fields instead of calculating an alignment here.
22431 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22432 innermost_loop_behavior fields.
22434 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22436 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22438 (DR_STEP_ALIGNMENT): New macro.
22439 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22440 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22441 (create_data_ref): Print it.
22442 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22443 to tell whether the step preserves vector (mis)alignment.
22444 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22445 Move the check for an integer step and generalise to all INTEGER_CST.
22446 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22447 Print the outer step alignment.
22449 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22451 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22452 with offset_alignment.
22453 (DR_ALIGNED_TO): Delete.
22454 (DR_OFFSET_ALIGNMENT): New macro.
22455 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22456 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22457 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22458 (create_data_ref): Likewise.
22459 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22460 (vect_analyze_data_refs): Likewise.
22461 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22462 creating dummy innermost behavior.
22464 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22466 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22467 with a "innermost_loop_behavior *" and refeence tree.
22468 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22469 (create_data_ref): Update call accordingly.
22470 * tree-predcom.c (find_looparound_phi): Likewise.
22472 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22474 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22475 fields with dr_wrt_vec_loop.
22476 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22477 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22478 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22479 (vect_dr_behavior): New function.
22480 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22481 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22482 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22483 track whether the step preserves the misalignment.
22484 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22485 Use vect_dr_behavior.
22486 (vect_setup_realignment): Update call accordingly.
22487 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22488 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22489 call to vect_create_addr_base_for_vector_ref.
22490 (vect_create_cond_for_align_checks): Likewise.
22491 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22492 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22493 (vect_recog_mask_conversion_pattern): Likewise.
22494 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22495 (new_stmt_vec_info): Remove redundant zeroing.
22497 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22499 * common/config/arm/arm-common.c (arm_be8_option): New function.
22500 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22501 (ISA_ARMv6): Add isa_bit_be8.
22502 * config/arm/arm.h (arm_be8_option): Add prototype.
22503 (BE8_SPEC_FUNCTION): New define.
22504 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22505 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22506 (mlittle-endian): Similarly.
22507 (mbe8, mbe32): New options.
22508 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22509 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22511 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22513 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22515 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22517 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22518 (cleanup_tree_cfg_bb): Use it.
22519 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22521 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22523 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22526 * loop-doloop.c (add_test): Update profile.
22528 2017-07-03 Martin Liska <mliska@suse.cz>
22531 * sanopt.c (rewrite_usage_of_param): New function.
22532 (sanitize_rewrite_addressable_params): Likewise.
22533 (pass_sanopt::execute): Call rewrite_usage_of_param.
22535 2017-07-03 Richard Biener <rguenther@suse.de>
22537 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22538 back to using VIEW_CONVERT_EXPR.
22540 2017-07-03 Martin Liska <mliska@suse.cz>
22543 * doc/extend.texi: Document when a resolver function is
22544 generated for target_clones.
22546 2017-07-03 Martin Liska <mliska@suse.cz>
22548 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22549 and shadow memory of auto variables which are subject of
22550 use-after-scope sanitization.
22551 (asan_expand_mark_ifn): Add do set only when is_poison.
22553 2016-07-03 Richard Biener <rguenther@suse.de>
22555 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22557 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22558 (vectorizable_reduction): Perform reduction PHI creation when
22559 visiting a reduction PHI and adjust and simplify code generation
22560 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22561 (vect_transform_loop): Visit reduction PHIs.
22562 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22563 defs into the SLP tree.
22564 (vect_build_slp_tree): Reduction defs terminate the recursion.
22565 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22567 (vect_get_vec_defs_for_stmt_copy): Export.
22568 (vect_get_vec_defs): Likewise.
22569 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22571 (vect_get_vec_defs_for_stmt_copy): Declare.
22572 (vect_get_vec_defs): Likewise.
22574 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22576 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22577 parameter with a "loop" parameter and use it instead of the
22578 loop containing DR_STMT. Don't check simple_iv when doing
22579 BB analysis. Describe the two analysis modes in the comment.
22581 2017-07-03 Tom de Vries <tom@codesourcery.com>
22583 PR tree-optimization/69468
22584 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22585 (find_same_succ_bb): Handle ignore_edge_flags.
22587 2017-07-03 Tom de Vries <tom@codesourcery.com>
22589 PR tree-optimization/81192
22590 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22592 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22594 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22596 2017-07-03 Tom de Vries <tom@codesourcery.com>
22598 PR tree-optimization/81192
22599 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22600 BB_SAME_SUCC (bb) == NULL.
22602 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22604 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22607 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22609 * dumpfile.c: Include profile-count.h
22610 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22612 (insert_cond_bb): Update profile.
22613 * tree-cfg.h (insert_cond_bb): Update prototype.
22614 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22615 * tree-dump.c: Do not include tree-cfg.
22617 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22619 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22621 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22623 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22626 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22628 * tree-complex.c (expand_complex_div_wide): update profile.
22630 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22631 Alan Hayward <alan.hayward@arm.com>
22632 David Sherwood <david.sherwood@arm.com>
22634 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22635 (CORETYPES_H): New define.
22636 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22637 (insn-modes-inline.h, s-modes-inline-h): New rules.
22638 (generated_files): Add insn-modes-inline.h.
22639 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22640 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22641 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22642 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22643 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22644 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22645 (build/gencodes.o, build/genconditions.o): Likewise.
22646 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22647 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22648 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22649 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22650 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22651 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22652 * coretypes.h: Include everything up to real.h for generators.
22653 Include insn-modes.h first. Include wide-int-print.h after
22654 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22655 * machmode.h: Don't include insn-modes.h here.
22656 * function-tests.c: Remove includes of signop.h, machmode.h,
22657 double-int.h and wide-int.h.
22659 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22661 * optc-save-gen.awk: Likewise.
22662 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22663 * godump.c: Remove include of wide-int-print.h.
22664 * pretty-print.h: Likewise.
22665 * wide-int-print.cc: Likewise.
22666 * wide-int.cc: Likewise.
22667 * hash-map-tests.c: Remove include of signop.h.
22668 * hash-set-tests.c: Likewise.
22669 * rtl-tests.c: Likewise.
22670 * mkconfig.sh: Remove include of machmode.h.
22671 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22673 (emit_insn_modes_inline_h): ...this new function. Emit the code
22674 into an insn-modes-inline.h header file, adding appropriate
22675 include guards and end comments.
22676 (emit_insn_modes_c_header): Remove include of machmode.h.
22677 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22679 (main): Handle -i flag and call emit_insn_modes_inline_h when
22682 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22684 * tree-ssa-strlen.c (strinfo): Rename the length field to
22685 nonzero_chars. Add a full_string_p field.
22686 (compare_nonzero_chars, zero_length_string_p): New functions.
22687 (get_addr_stridx): Add an offset_out parameter.
22688 Use compare_nonzero_chars.
22689 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22690 (new_strinfo): Update after above changes to strinfo.
22691 (set_endptr_and_length): Set full_string_p.
22692 (get_string_length): Update after above changes to strinfo.
22693 (unshare_strinfo): Update call to new_strinfo.
22694 (maybe_invalidate): Likewise.
22695 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22696 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22697 as a uhwi instead of an shwi. Update after above changes to
22698 strinfo and new_strinfo.
22699 (zero_length_string): Assert that chainsi contains full strings.
22700 Use zero_length_string_p. Update call to new_strinfo.
22701 (adjust_related_strinfos): Update after above changes to strinfo.
22702 Copy full_string_p from origsi.
22703 (adjust_last_stmt): Use zero_length_string_p.
22704 (handle_builtin_strlen): Update after above changes to strinfo and
22705 new_strinfo. Install the lhs as the string length if the previous
22706 entry didn't describe a full string.
22707 (handle_builtin_strchr): Update after above changes to strinfo
22709 (handle_builtin_strcpy): Likewise.
22710 (handle_builtin_strcat): Likewise.
22711 (handle_builtin_malloc): Likewise.
22712 (handle_pointer_plus): Likewise.
22713 (handle_builtin_memcpy): Likewise. Track nonzero characters
22714 that aren't necessarily followed by a nul terminator.
22715 (handle_char_store): Likewise.
22717 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22719 PR tree-optimization/80769
22720 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22721 for malloc and calloc. Document the new invariant that all related
22722 strinfos have delayed lengths or none do.
22723 (verify_related_strinfos): Move earlier in file.
22724 (set_endptr_and_length): New function, split out from...
22725 (get_string_length): ...here. Also set the lengths of related
22727 (zero_length_string): Assert that chainsi has known (rather than
22729 (adjust_related_strinfos): Likewise.
22731 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22733 PR tree-optimization/81136
22734 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22735 assert that two references with the same misalignment have the same
22736 compile-time misalignment if those compile-time misalignments
22739 2017-07-01 Andi Kleen <ak@linux.intel.com>
22741 * print-tree.c (print_node): Print all attributes.
22743 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22745 * cfg.c (scale_bbs_frequencies): New function.
22746 * cfg.h (scale_bbs_frequencies): Declare it.
22747 * cfgloopanal.c (single_likely_exit): Cleanup.
22748 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22750 (scale_loop_profile): Likewise.
22751 (loop_version): Likewise.
22752 (create_empty_loop_on_edge): Update.
22753 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22754 scale_loop_frequencies, scale_loop_profile, loopify,
22755 loop_version): Update prototypes.
22756 * modulo-sched.c (sms_schedule): Update.
22757 * predict.c (unlikely_executed_edge_p): Also check probability.
22758 (probably_never_executed_edge_p): Fix typo.
22759 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22760 * tree-parloops.c (gen_parallel_loop): Update.
22761 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22762 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22763 * tree-ssa-loop-split.c (split_loop): Update.
22764 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22765 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22766 (vect_loop_versioning): Update.
22767 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22769 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22771 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22773 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22775 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22776 to keep profile consistent.
22778 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22780 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22781 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22782 * profile-count.h (max_safe_multiplier): Make unsigned.
22783 (profile_count::guessed_zero): New.
22785 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22787 * bb-reorder.c (fix_up_crossing_landing_pad,
22788 fix_crossing_conditional_branches): Use make_single_succ_edge
22789 to keep profile consistent.
22791 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22793 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22796 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22799 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22800 the right scopes, make sure cond_jump isn't preserved between multiple
22801 iterations. Search for fallthru edge whenever there are 3+ edges and
22802 use find_fallthru_edge for it.
22804 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22806 Patch by Alexander Monakov <amonakov@ispras.ru>
22807 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22808 probabilities consistently.
22810 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22812 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22813 emit_cmp_and_jump_insns.
22815 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22818 * tree-inline.c (expand_call_inline): Combine profile statuses.
22820 2017-06-30 Andrew Pinski <apinski@cavium.com>
22822 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22823 fold_stmt returned true.
22825 2017-06-30 Nathan Sidwell <nathan@acm.org>
22827 * ggc.h (empty_string): Delete.
22828 * cfgexpand.c (expand_asm_stmt): Use plain "".
22829 * optabs.c (expand_asm_memory_barrier): Likewise.
22830 * stringpool.c (empty_string): Delete.
22831 (digit_vector, digit_string): Delete.
22832 (ggc_alloc_string): Use plain "", don't optimize single digit
22833 strings. Use ggc_alloc_atomic.
22835 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22837 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22838 comparison set and one other set, use the cost of the non-comparison
22841 2017-06-30 Nathan Sidwell <nathan@acm.org>
22843 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22846 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22848 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22849 loops. Remove now unneeded calls to gimple_switch_set_label() that
22850 just set removed labels to NULL_TREE.
22852 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22854 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22855 (set_range_info): ...here. Only call set_range_info_raw if domain
22857 (set_nonzero_bits): Call set_range_info_raw.
22858 * tree-ssanames.h (set_range_info_raw): New.
22860 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22863 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22864 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22865 of nonimmediate_operand and <store_mask_constraint> instead of m
22866 for the input operand. For V8FI iterator, always split if input
22867 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22868 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22869 <store_mask_predicate> instead of register_operand and
22870 <store_mask_constraint> instead of v for the input operand. Make
22871 sure both operands aren't MEMs for if not <mask_applied>.
22873 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22875 * lto-wrapper.c (copy_file) Close both file descriptors before
22878 2017-06-30 Martin Liska <mliska@suse.cz>
22881 * multiple_target.c (create_dispatcher_calls): Make ifunc
22882 also for function that don't have calls or are not referenced.
22884 2017-06-30 Richard Biener <rguenther@suse.de>
22886 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22887 analyze the first scalar stmt. Move vector type computation
22888 for the BB case here from ...
22889 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22890 live operation processing in the SLP case properly.
22892 2017-06-30 Richard Biener <rguenther@suse.de>
22894 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22896 2017-06-30 Martin Liska <mliska@suse.cz>
22899 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22900 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22902 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22904 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22907 2017-06-30 Richard Biener <rguenther@suse.de>
22909 PR tree-optimization/81249
22910 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22911 condition reduction result to original scalar type.
22913 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22915 * profile-count.h (enum profile_quality): Fix typos and whitespace
22918 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22920 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22921 type for branch probabilities.
22923 2017-06-29 Julian Brown <julian@codesourcery.com>
22924 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22926 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22927 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22928 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22929 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22931 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22933 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22934 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22935 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22936 CC usage from generic code to here.
22937 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22938 CC usage into the target macros.
22940 2017-06-29 Maya Rashish <coypu@sdf.org>
22942 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22945 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22947 * arm/arm-builtins.c: Include profile-count.h
22948 * except.c (sjlj_emit_function_enter): Use
22949 profile_probability::unlikely.
22951 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22953 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22954 and tocrel_offset be pointer args rather than implicitly using
22956 (legitimate_constant_pool_address_p, rs6000_emit_move,
22957 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22958 tocrel_offset and use in toc_relative_expr_p call.
22959 (print_operand, print_operand_address): Use static tocrel_base_oac
22960 and tocrel_offset_oac.
22961 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22964 2017-06-29 Maya Rashish <coypu@sdf.org>
22966 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22968 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22970 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22971 objects, take into account only the alignment of 'op0' and 'mode1' if
22974 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22976 * ccmp.c (ccmp_tree_comparison_p): New function.
22977 (ccmp_candidate_p): Update to use above function.
22978 (get_compare_parts): New function.
22979 (expand_ccmp_next): Update to use new functions.
22980 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22982 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22983 take mode as argument.
22984 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22985 * expr.c (expand_expr_real_1): Pass mode as argument.
22987 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22989 * combine.c (combine_instructions): Print insns to dump_file, together
22992 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22994 * asan.c (asan_emit_stack_protection): Update.
22995 (create_cond_insert_point): Update.
22996 * auto-profile.c (afdo_propagate_circuit): Update.
22997 * basic-block.h (struct edge_def): Turn probability to
22998 profile_probability.
22999 (EDGE_FREQUENCY): Update.
23000 * bb-reorder.c (find_traces_1_round): Update.
23001 (better_edge_p): Update.
23002 (sanitize_hot_paths): Update.
23003 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23004 (make_single_succ_edge): Update.
23005 (check_bb_profile): Update.
23006 (dump_edge_info): Update.
23007 (update_bb_profile_for_threading): Update.
23008 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23009 probabilitycount to 0.
23010 * cfgbuild.c (compute_outgoing_frequencies): Update.
23011 * cfgcleanup.c (try_forward_edges): Update.
23012 (outgoing_edges_match): Update.
23013 (try_crossjump_to_edge): Update.
23014 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23015 (expand_gimple_tailcall): Update.
23016 (construct_init_block): Use make_single_succ_edge.
23017 (construct_exit_block): Use make_single_succ_edge.
23018 * cfghooks.c (verify_flow_info): Update.
23019 (redirect_edge_succ_nodup): Update.
23020 (split_edge): Update.
23021 (account_profile_record): Update.
23022 * cfgloopanal.c (single_likely_exit): Update.
23023 * cfgloopmanip.c (scale_loop_profile): Update.
23024 (set_zero_probability): Remove.
23025 (duplicate_loop_to_header_edge): Update.
23026 * cfgloopmanip.h (loop_version): Update prototype.
23027 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23028 (force_nonfallthru_and_redirect): Update.
23029 (update_br_prob_note): Update.
23030 (rtl_verify_edges): Update.
23031 (purge_dead_edges): Update.
23032 (rtl_lv_add_condition_to_bb): Update.
23033 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23034 * cgraphunit.c (init_lowered_empty_function): Update.
23035 (cgraph_node::expand_thunk): Update.
23036 * cilk-common.c: Include profile-count.h
23037 * dojump.c (inv): Remove.
23038 (jumpifnot): Update.
23039 (jumpifnot_1): Update.
23040 (do_jump_1): Update.
23042 (do_jump_by_parts_greater_rtx): Update.
23043 (do_compare_rtx_and_jump): Update.
23044 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23045 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23046 * dwarf2cfi.c: Include profile-count.h
23047 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23048 (sjlj_emit_dispatch_table): Likewise.
23049 * explow.c: Include profile-count.h
23050 * expmed.c (emit_store_flag_force): Update.
23051 (do_cmp_and_jump): Update.
23052 * expr.c (compare_by_pieces_d::generate): Update.
23053 (compare_by_pieces_d::finish_mode): Update.
23054 (emit_block_move_via_loop): Update.
23055 (store_expr_with_bounds): Update.
23056 (store_constructor): Update.
23057 (expand_expr_real_2): Update.
23058 (expand_expr_real_1): Update.
23059 * expr.h (try_casesi, try_tablejump): Update prototypes.
23060 * gimple-pretty-print.c (dump_probability): Update.
23061 (dump_profile): New.
23062 (dump_gimple_label): Update.
23063 (dump_gimple_bb_header): Update.
23064 * graph.c (draw_cfg_node_succ_edges): Update.
23065 * hsa-gen.c (convert_switch_statements): Update.
23066 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23067 (find_if_case_1): Update.
23068 (find_if_case_2): Update.
23069 * internal-fn.c (expand_arith_overflow_result_store): Update.
23070 (expand_addsub_overflow): Update.
23071 (expand_neg_overflow): Update.
23072 (expand_mul_overflow): Update.
23073 (expand_vector_ubsan_overflow): Update.
23074 * ipa-cp.c (good_cloning_opportunity_p): Update.
23075 * ipa-split.c (split_function): Use make_single_succ_edge.
23076 * ipa-utils.c (ipa_merge_profiles): Update.
23077 * loop-doloop.c (add_test): Update.
23078 (doloop_modify): Update.
23079 * loop-unroll.c (compare_and_jump_seq): Update.
23080 (unroll_loop_runtime_iterations): Update.
23081 * lra-constraints.c (lra_inheritance): Update.
23082 * lto-streamer-in.c (input_cfg): Update.
23083 * lto-streamer-out.c (output_cfg): Update.
23084 * mcf.c (adjust_cfg_counts): Update.
23085 * modulo-sched.c (sms_schedule): Update.
23086 * omp-expand.c (expand_omp_for_init_counts): Update.
23087 (extract_omp_for_update_vars): Update.
23088 (expand_omp_ordered_sink): Update.
23089 (expand_omp_for_ordered_loops): Update.
23090 (expand_omp_for_generic): Update.
23091 (expand_omp_for_static_nochunk): Update.
23092 (expand_omp_for_static_chunk): Update.
23093 (expand_cilk_for): Update.
23094 (expand_omp_simd): Update.
23095 (expand_omp_taskloop_for_outer): Update.
23096 (expand_omp_taskloop_for_inner): Update.
23097 * omp-simd-clone.c (simd_clone_adjust): Update.
23098 * optabs.c (expand_doubleword_shift): Update.
23099 (expand_abs): Update.
23100 (emit_cmp_and_jump_insn_1): Update.
23101 (expand_compare_and_swap_loop): Update.
23102 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23103 * predict.c (predictable_edge_p): Update.
23104 (edge_probability_reliable_p): Update.
23105 (set_even_probabilities): Update.
23106 (combine_predictions_for_insn): Update.
23107 (combine_predictions_for_bb): Update.
23108 (propagate_freq): Update.
23109 (estimate_bb_frequencies): Update.
23110 (force_edge_cold): Update.
23111 * profile-count.c (profile_count::dump): Add missing space into dump.
23112 (profile_count::debug): Add newline.
23113 (profile_count::differs_from_p): Explicitly convert to unsigned.
23114 (profile_count::stream_in): Update.
23115 (profile_probability::dump): New member function.
23116 (profile_probability::debug): New member function.
23117 (profile_probability::differs_from_p): New member function.
23118 (profile_probability::differs_lot_from_p): New member function.
23119 (profile_probability::stream_in): New member function.
23120 (profile_probability::stream_out): New member function.
23121 * profile-count.h (profile_count_quality): Rename to ...
23122 (profile_quality): ... this one.
23123 (profile_probability): New.
23124 (profile_count): Update.
23125 * profile.c (compute_branch_probabilities): Update.
23126 * recog.c (peep2_attempt): Update.
23127 * sched-ebb.c (schedule_ebbs): Update.
23128 * sched-rgn.c (find_single_block_region): Update.
23129 (compute_dom_prob_ps): Update.
23130 (schedule_region): Update.
23131 * sel-sched-ir.c (compute_succs_info): Update.
23132 * stmt.c (struct case_node): Update.
23133 (do_jump_if_equal): Update.
23134 (get_outgoing_edge_probs): Update.
23135 (conditional_probability): Update.
23136 (emit_case_dispatch_table): Update.
23137 (expand_case): Update.
23138 (expand_sjlj_dispatch_table): Update.
23139 (emit_case_nodes): Update.
23140 * targhooks.c: Update.
23141 * tracer.c (better_p): Update.
23142 (find_best_successor): Update.
23143 * trans-mem.c (expand_transaction): Update.
23144 * tree-call-cdce.c: Update.
23145 * tree-cfg.c (gimple_split_edge): Upate.
23146 (move_sese_region_to_fn): Upate.
23147 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23148 * tree-eh.c (lower_resx): Upate.
23149 (cleanup_empty_eh_move_lp): Upate.
23150 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23151 * tree-inline.c (copy_edges_for_bb): Update.
23152 (copy_cfg_body): Update.
23153 * tree-parloops.c (gen_parallel_loop): Update.
23154 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23155 (gimple_gen_time_profiler): Update.
23156 * tree-ssa-dce.c (remove_dead_stmt): Update.
23157 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23158 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23159 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23160 (unloop_loops): Update.
23161 (try_peel_loop): Update.
23162 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23163 * tree-ssa-loop-split.c (connect_loops): Update.
23164 (split_loop): Update.
23165 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23166 (hoist_guard): Update.
23167 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23168 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23169 (value_replacement): Update.
23170 * tree-ssa-reassoc.c (branch_fixup): Update.
23171 * tree-ssa-tail-merge.c (replace_block_by): Update.
23172 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23173 (create_edge_and_update_destination_phis): Update.
23174 (compute_path_counts): Update.
23175 (recompute_probabilities): Update.
23176 (update_joiner_offpath_counts): Update.
23177 (freqs_to_counts_path): Update.
23178 (duplicate_thread_path): Update.
23179 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23180 (struct switch_conv_info): Update.
23181 (gen_inbound_check): Update.
23182 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23183 (vect_do_peeling): Update.
23184 (vect_loop_versioning): Update.
23185 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23186 (optimize_mask_stores): Update.
23187 * ubsan.c (ubsan_expand_null_ifn): Update.
23188 * value-prof.c (gimple_divmod_fixed_value): Update.
23189 (gimple_divmod_fixed_value_transform): Update.
23190 (gimple_mod_pow2): Update.
23191 (gimple_mod_pow2_value_transform): Update.
23192 (gimple_mod_subtract): Update.
23193 (gimple_mod_subtract_transform): Update.
23194 (gimple_ic): Update.
23195 (gimple_stringop_fixed_value): Update.
23196 (gimple_stringops_transform): Update.
23197 * value-prof.h: Update.
23199 2017-06-29 Carl Love <cel@us.ibm.com>
23201 * config/rs6000/rs6000-c.c: Add support for built-in functions
23202 vector signed int vec_signed (vector float);
23203 vector signed long long vec_signed (vector double);
23204 vector signed int vec_signed2 (vector double, vector double);
23205 vector signed int vec_signede (vector double);
23206 vector signed int vec_signedo (vector double);
23207 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23208 instruction generator.
23209 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23210 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23211 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23213 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23214 vunsignede_v2df): Add define_expands.
23215 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23216 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23217 VEC_UNSIGNEDO): Add definitions.
23218 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23219 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23220 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23221 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23222 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23223 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23224 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23225 vec_unsignede, vec_unsignedo): Add builtin defines.
23226 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23228 * doc/extend.texi: Update the built-in documentation file for the
23229 new built-in functions.
23231 2017-06-29 Richard Biener <rguenther@suse.de>
23233 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23234 reduction chains to LOOP_VINFO_REDUCTIONS.
23235 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23236 SLP reductions after processing reduction chains.
23238 2017-06-29 Nathan Sidwell <nathan@acm.org>
23240 * builtins.c (fold_builtin_FUNCTION): Use
23241 lang_hooks.decl_printable_name.
23243 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23245 PR middle-end/81194
23246 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23247 with only one label.
23248 * stmt.c (expand_case): Assert NCASES is greater than one.
23250 2017-06-29 Richard Biener <rguenther@suse.de>
23252 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23254 (group_case_labels): Likewise.
23255 (find_taken_edge): Push sanity checking on val to workers...
23256 (find_taken_edge_cond_expr): ... here
23257 (find_taken_edge_switch_expr): ... and here, handle cases
23258 with just a default label.
23259 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23260 (group_case_labels): Likewise.
23261 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23262 group_case_labels does anything cleanup the CFG again.
23264 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23266 PR tree-optimization/81196
23267 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23268 exit condition comparing two IVs.
23270 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23272 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23273 profile to the dummy entry at the end of the list of architectures.
23274 * config/arm/arm-cpu-cdata.h: Regenerated.
23276 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23277 Michael Collison <michael.collison@arm.com>
23280 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23282 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23283 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23284 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23285 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23286 of shift when the shift amount is masked with constant equal to
23287 the size of the mode.
23288 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23291 2017-06-29 Martin Liska <mliska@suse.cz>
23293 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23295 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23298 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23301 (edge_growth_cache_entry::edge_growth_cache_entry): New
23303 (reset_edge_growth_cache): Update to use constructor.
23305 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23307 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23308 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23309 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23311 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23313 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23314 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23316 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23318 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23319 (*-linux-uclibc*): Add t-uclibc tmake_file.
23320 * config/t-musl: New.
23321 * config/t-uclibc: New.
23323 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23325 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23327 (gen_comm_data): Emit architectural setting of arch_prof.
23328 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23330 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23331 (armv8-m.base, armv8-m.main): Likewise.
23332 * arm-protos.h (arm_build_target): Add profile field.
23333 (arch_option): Likewise.
23334 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23336 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23337 arm_active_target.profile.
23339 2017-06-28 Richard Biener <rguenther@suse.de>
23341 PR middle-end/81227
23342 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23343 TYPE_OVERFLOW_WRAPS.
23344 * match.pd (negate_expr_p): Likewise.
23345 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23346 fold_build2, not fold_binary.
23348 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23350 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23351 Convert memory address to Pmode.
23352 (aarch64_print_operand): Assert MEM operands are always Pmode.
23354 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23357 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23358 Remove redundant if.
23359 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23360 * config/arm/aarch-common-protos.h
23361 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23362 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23364 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23367 * multiple_target.c (create_dispatcher_calls): Set the default
23368 clone to be static, not public.
23370 2017-06-28 Richard Biener <rguenther@suse.de>
23372 * tree-vect-loop.c (vectorizable_reduction): Move special
23373 cond reduction IV var creation ...
23374 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23375 parameter. Use STMT_VINFO_VECTYPE.
23376 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23379 2017-06-28 Martin Liska <mliska@suse.cz>
23382 * ipa-visibility.c (non_local_p): Handle visibility.
23384 2017-06-28 Martin Liska <mliska@suse.cz>
23387 * common.opt: Add IntegerRange to various options.
23388 * opt-functions.awk (integer_range_info): New function.
23389 * optc-gen.awk: Add integer_range_info to cl_options struct.
23390 * opts-common.c (decode_cmdline_option): Handle
23391 CL_ERR_INT_RANGE_ARG.
23392 (cmdline_handle_error): Likewise.
23393 * opts.c (print_filtered_help): Show valid interval in
23394 when --help is provided.
23395 * opts.h (struct cl_option): Add range_min and range_max fields.
23396 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23398 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23400 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23401 (x * C EQ/NE y * C): New transformation.
23403 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23405 * genmultilib (combination_space): Accept '+' in option names.
23407 2017-06-28 Martin Liska <mliska@suse.cz>
23410 * asan.c (instrument_derefs): Bail out inner references
23411 that are hard register variables.
23413 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23416 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23417 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23419 2017-06-28 Richard Biener <rguenther@suse.de>
23421 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23422 (vect_get_slp_defs): Adjust.
23423 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23425 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23427 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23428 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23429 (vectorizable_reduction): Adjust.
23430 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23432 (vect_get_slp_defs): Likewise.
23433 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23434 (vectorizable_bswap): Adjust.
23435 (vectorizable_call): Likewise.
23436 (vectorizable_conversion): Likewise.
23437 (vectorizable_assignment): Likewise.
23438 (vectorizable_shift): Likewise.
23439 (vectorizable_operation): Likewise.
23440 (vectorizable_store): Likewise.
23441 (vectorizable_condition): Likewise.
23442 (vectorizable_comparison): Likewise.
23444 2017-06-28 Michael Collison <michael.collison@arm.com>
23447 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23449 (arm_gen_movmemqi): Removed unused variable 'i'.
23450 Convert 'for' loop into 'while' loop.
23451 (arm_expand_prologue): Remove last unnecessary set of insn.
23452 (thumb_pop): Remove unused variable 'pushed_words'.
23453 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23455 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23457 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23458 * config/s390/s390-protos.h: Add prototype of
23459 s390_rel_address_ok_p.
23460 * config/s390/s390.c (s390_got_symbol): New function.
23461 (s390_rel_address_ok_p): New function.
23462 (legitimize_pic_address): Use s390_rel_address_ok_p.
23463 (s390_load_got): Use s390_got_symbol.
23464 (s390_option_override): Issue error if
23465 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23466 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23468 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23470 2017-06-27 Andrew Pinski <apinski@cavium.com>
23472 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23473 (X * copysign (1.0, X)): New pattern.
23474 (X * copysign (1.0, -X)): New pattern.
23475 (copysign (-1.0, CST)): New pattern.
23477 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23479 * genmultilib (combination_space): Remove variable.
23480 Validate reuse rules against regular expression for any sequence
23481 of multilib options in any order.
23483 2017-06-27 Michael Collison <michael.collison@arm.com>
23485 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23486 call aarch64_split_simd_combine.
23487 * (aarch64_combine_internal<mode>): Delete pattern.
23488 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23489 Allow register and subreg operands.
23491 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23493 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23494 specific need, just fallback on defaults.
23495 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23497 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23498 Olivier Hainque <hainque@adacore.com>
23500 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23502 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23503 targets. Pick a default if no particular attempt applied.
23504 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23507 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23509 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23510 (x86_64-wrs-vxworks7): Likewise.
23512 2017-06-27 Marek Polacek <polacek@redhat.com>
23515 * ubsan.c (instrument_null): Check get_base_address's result for null.
23517 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23519 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23521 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23523 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23524 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23525 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23526 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23527 New function types.
23528 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23529 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23530 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23531 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23532 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23533 BUILT_IN_FEUPDATEENV): New builtins.
23534 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23535 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23536 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23537 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23539 (builtin_structptr_types): Adjust size.
23540 * tree.c (builtin_structptr_types): Add four entries.
23542 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23543 Olivier Hainque <hainque@adacore.com>
23545 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23546 (TLS_SYM): New local macro, forcing reference to __tls__ on
23547 link command lines for VxWorks 7 RTPs, triggering initialization
23549 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23550 OS features TLS support, true for RTPs on VxWorks 7.
23551 * config/vxworks.c (vxworks_override_options): Setup emutls
23554 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23556 * predict.c (test_prediction_value_range): Use -1U instead of -1
23557 to avoid narrowing conversion warning.
23558 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23559 to avoid narrowing conversion warning.
23560 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23562 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23564 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23566 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23567 64bit configurations.
23568 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23569 (SIZE_TYPE): Likewise.
23570 * config/vxworks.c (vxworks_emutls_var_fields): Use
23571 long_unsigned_type_node instead of unsigned_type_node as the offset
23572 field type, which is "pointer" mode in emutls.c.
23574 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23577 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23579 PR middle-end/81207
23580 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23581 gimple_vuse copying separately from gimple_vdef copying.
23583 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23585 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23588 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23589 Olivier Hainque <hainque@adacore.com>
23591 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23592 for all vxworks7 targets.
23593 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23594 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23595 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23596 variations for VX6/VX7 and 32/64bits later on in ...
23597 (VXWORKS_LIB_SPEC): Leverage new macros.
23598 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23599 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23601 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23603 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23604 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23606 2017-06-26 Carl Love <cel@us.ibm.com>
23608 * config/rs6000/rs6000-c.c: Add support for built-in functions
23609 vector bool char vec_reve (vector bool char);
23610 vector signed char vec_reve (vector signed char);
23611 vector unsigned char vec_reve (vector unsigned char);
23612 vector bool int vec_reve (vector bool int);
23613 vector signed int vec_reve (vector signed int);
23614 vector unsigned int vec_reve (vector unsigned int);
23615 vector bool long long vec_reve (vector bool long long);
23616 vector signed long long vec_reve (vector signed long long);
23617 vector unsigned long long vec_reve (vector unsigned long long);
23618 vector bool short vec_reve (vector bool short);
23619 vector signed short vec_reve (vector signed short);
23620 vector double vec_reve (vector double);
23621 vector float vec_reve (vector float);
23622 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23623 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23624 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23625 (altivec_vreve): New pattern.
23626 * config/rs6000/altivec.h (vec_reve): New define.
23627 * doc/extend.texi (vec_rev): Update the built-in documentation file
23628 for the new built-in functions.
23630 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23632 PR tree-optimization/71815
23633 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23635 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23637 (slsr_process_phi): Likewise.
23638 (replace_uncond_cands_and_profitable_phis): Don't replace a
23639 multiply candidate with a stride of 1 (copy or cast).
23640 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23642 (lowest_cost_path): Likewise.
23643 (total_savings): Likewise.
23645 2017-06-26 Richard Biener <rguenther@suse.de>
23648 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23649 Use def_builtin_pure for all gather builtins.
23651 2017-06-26 Richard Biener <rguenther@suse.de>
23653 PR tree-optimization/81203
23654 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23655 non-dominating BBs.
23657 2017-06-26 Marek Polacek <polacek@redhat.com>
23660 * doc/invoke.texi: Document -Wmultistatement-macros.
23662 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23664 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23665 arm_neon_ok_no_float_abi effective target.
23667 2017-06-26 Richard Biener <rguenther@suse.de>
23669 PR tree-optimization/80928
23670 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23671 (copy_bbs): Set BB_DUPLICATED flag early.
23672 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23674 (execute_on_shrinking_pred): Likewise.
23675 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23676 BB_DUPLICATED blocks.
23677 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23678 iterate over all PHIs considering removal of *gsi.
23680 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23682 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23685 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23687 * config/rs6000/rs6000-string.c: (expand_block_clear,
23688 do_load_for_compare, select_block_compare_mode,
23689 compute_current_alignment, expand_block_compare,
23690 expand_strncmp_align_check, expand_strn_compare,
23691 expand_block_move, rs6000_output_load_multiple)
23692 Move functions related to string/block move/compare
23693 to a separate file.
23694 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23695 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23696 for this function which is now used in two files.
23697 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23698 * config.gcc: Add rs6000-string.o to extra_objs for
23699 targets powerpc*-*-* and rs6000*-*-*.
23701 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23704 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23705 32-bit, since indexed is not valid for DImode.
23706 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23707 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23708 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23709 (define_peephole2 for Altivec d-form store): Likewise.
23712 * multiple_target.c (create_dispatcher_calls): Only create the
23713 dispatcher call if the function is the default clone of a
23714 versioned function.
23716 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23718 PR middle-end/80902
23719 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23720 a call, force the call to not be a tail call.
23722 2017-06-23 Jeff Law <law@redhat.com>
23724 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23727 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23729 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23730 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23731 handling for early expansion of vector shifts (sl,sr,sra,rl).
23732 (builtin_function_type): Add vector shift right instructions
23733 to the unsigned argument list.
23735 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23737 rtl-optimizatoin/79286
23738 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23739 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23740 trap. PIC register plus a const unspec without offset can never trap.
23742 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23744 * tree.h (builtin_structptr_type): New type.
23745 (builtin_structptr_types): Declare new array.
23746 * tree.c (builtin_structptr_types): New array.
23747 (free_lang_data, build_common_tree_nodes): Use it.
23749 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23752 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23755 2017-06-22 Matt Turner <mattst88@gmail.com>
23757 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23758 Lake models to skylake case. Assume skylake for unknown
23759 models with clflushopt.
23761 2017-06-22 Jeff Law <law@redhat.com>
23763 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23764 frame sizes that do not satisfy aarch64_uimm12_shift.
23766 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23768 * profile-count.h (apply_probability,
23769 apply_scale, probability_in): Fix checks for zero.
23771 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23773 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23774 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23776 2016-06-22 Richard Biener <rguenther@suse.de>
23778 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23779 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23780 REDUC_MAX_EXPR support.
23781 (vectorizable_reduction): Likewise.
23782 (vect_create_epilog_for_reduction): Likewise.
23784 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23786 * match.pd (A / (1 << B) -> A >> B): New.
23787 * generic-match-head.c: Include optabs-tree.h.
23788 * gimple-match-head.c: Likewise.
23789 * optabs-tree.h (target_supports_op_p): New.
23790 * optabs-tree.c (target_supports_op_p): New.
23792 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23794 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23795 $gcc_cv_ld --help output.
23796 (gcc_cv_ld_demangle): Likewise.
23797 (gcc_cv_ld_eh_frame_hdr): Likewise.
23798 (gcc_cv_ld_pie): Likewise.
23799 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23800 (gcc_cv_ld_buildid): Likewise.
23801 (gcc_cv_ld_sysroot): Likewise.
23802 (ld_bndplt_support): Likewise.
23803 (ld_pushpopstate_support): Likewise.
23804 * configure: Regenerate.
23805 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23807 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23810 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23811 operands indexes to avoid gap between operands and match_dups.
23813 2017-06-21 Andrew Pinski <apinski@cavium.com>
23815 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23816 Increment Arith_shift and Arith_shift_reg by 1.
23817 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23819 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23820 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23821 (aarch64_strip_extend): Add new argument and test for it.
23822 (aarch64_cheap_mult_shift_p): New function.
23823 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23824 add a cost if it is true.
23825 Update calls to aarch64_strip_extend.
23826 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23828 2017-06-21 Andrew Pinski <apinski@cavium.com>
23830 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23832 (thunderxt88): Likewise.
23833 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23834 (thunderx_prefetch_tune): New variable.
23835 (thunderx2t99_prefetch_tune): Update for the correct values.
23836 (thunderxt88_tunings): New variable.
23837 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23838 generic_prefetch_tune.
23839 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23841 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23843 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23844 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23845 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23846 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23847 (aarch64_atomic_cas<mode>, GPI): Likewise.
23849 2017-06-21 Martin Liska <mliska@suse.cz>
23851 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23852 statements on cold and hot labels.
23853 * predict.c (tree_estimate_probability_bb): Remove the
23854 prediction from this place.
23856 2017-06-21 Martin Liska <mliska@suse.cz>
23858 PR tree-optimization/79489
23859 * gimplify.c (maybe_add_early_return_predict_stmt): New
23861 (gimplify_return_expr): Call the function.
23862 * predict.c (tree_estimate_probability_bb): Remove handling
23864 * predict.def: Update comment about early return predictor.
23865 * gimple-predict.h (is_gimple_predict): New function.
23866 * predict.def: Change default value of early return to 66.
23867 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23869 * passes.def: Put pass_strip_predict_hints to the beginning of
23872 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23874 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23875 FUNCTION_DECL declarations.
23876 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23878 (dwaf2out_decl): Likewise.
23879 * godump.c (go_early_global_decl): Skip call to the real debug hook
23880 for FUNCTION_DECL declarations.
23881 * passes.c (rest_of_decl_compilation): Skip call to the
23882 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23883 -fdump-go-spec is passed.
23885 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23887 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23888 Reorder for compactness.
23889 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23890 (def_builtin_pure, def_builtin_pure2): New functions.
23891 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23893 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23895 * match.pd (nop_convert): New predicate.
23896 ((A +- CST1) +- CST2): Allow some NOP conversions.
23898 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23901 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23902 with ctors/dtors if GOVD_SHARED is set.
23904 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23906 * config/aarch64/aarch64.md (movti_aarch64):
23907 Emit mov rather than orr.
23908 (movtf_aarch64): Likewise.
23909 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23910 Emit mov rather than orr.
23912 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23914 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23915 Swap alternatives, make integer dup more expensive.
23917 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23919 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23920 Return true for non-tls symbols.
23922 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23924 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23925 (cortex-a75): Likewise.
23926 (cortex-a75.cortex-a55): Likewise.
23927 * config/aarch64/aarch64-tune.md: Regenerate.
23928 * doc/invoke.texi (-mtune): Document new values for -mtune.
23930 2017-06-21 Tom de Vries <tom@codesourcery.com>
23932 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23933 stack_size feature.
23934 (Effective-Target Keywords, Other attributes): Suggest using
23935 dg-add-options stack_size feature to get stack limit in stack_size
23936 effective target documentation.
23938 2017-06-21 Julian Brown <julian@codesourcery.com>
23939 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23941 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23942 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23943 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23945 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23946 attribute type list for neon_multiply.
23947 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23948 attribute type list for neon_multiply.
23949 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23950 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23951 attribute type list for neon_multiply.
23952 * config/arm/types.md (crypto_pmull): Add.
23953 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23954 attribute type list.
23956 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23958 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23961 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23963 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23964 to make sure not to dereference a NULL cost_classes_ptr pointer.
23966 2017-06-20 Carl Love <cel@us.ibm.com>
23968 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23969 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23970 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23971 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23972 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23973 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23974 VMULOSW): New enum "unspec" values.
23975 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23976 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23977 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23978 altivec_vmulosw): New patterns.
23979 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23980 VMULOSW): Add definitions.
23982 2017-06-20 Julia Koval <julia.koval@intel.com>
23984 * config/i386/i386.c: Fix rounding expand for new pattern.
23985 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23987 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23989 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23990 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23992 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23994 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23997 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23999 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24000 architecture, then by implementer ID.
24001 * config/aarch64/aarch64-tune.md: Regenerate.
24003 2017-06-20 Richard Biener <rguenther@suse.de>
24005 PR middle-end/81097
24006 * fold-const.c (split_tree): Fold to type before negating.
24008 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24010 * diagnostic-show-locus.c
24011 (selftest::test_fixit_deletion_affecting_newline): New function.
24012 (selftest::diagnostic_show_locus_c_tests): Call it.
24014 2017-06-20 Andreas Schwab <schwab@suse.de>
24017 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24020 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24022 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24023 __ARM_FEATURE_COPROC according to support.
24025 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24027 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24028 Rewritten to avoid overflow for > 32-bit pointers.
24031 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24032 by removing enum keyword.
24033 (ubsan_type_descriptor): Likewise. Formatting fix.
24036 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24037 splitter): Require TARGET_SSE2 in the condition.
24039 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24042 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24043 for doing vector set of SFmode on ISA 3.0.
24044 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24045 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24047 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24048 SFmode value into a V4SF variable that was extracted from another
24049 V4SF variable without converting the element to double precision
24050 and back to single precision vector format.
24051 (vsx_insert_extract_v4sf_p9_2): Likewise.
24053 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24055 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24056 in UWHI to avoid undefined overflow.
24059 * ubsan.h (enum ubsan_encode_value_phase): New.
24060 (ubsan_encode_value): Change second argument to
24061 enum ubsan_encode_value_phase with default value of
24062 UBSAN_ENCODE_VALUE_GENERIC.
24063 * ubsan.c (ubsan_encode_value): Change second argument to
24064 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24065 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24066 create_tmp_var_raw instead of create_tmp_var and use a
24068 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24069 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24070 ubsan_encode_value callers.
24073 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24074 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24075 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24077 2017-06-19 Richard Biener <rguenther@suse.de>
24079 PR middle-end/81118
24080 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24081 estimates if we changed anything.
24083 2017-06-19 Richard Biener <rguenther@suse.de>
24085 PR tree-optimization/80887
24086 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24087 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24088 simplified lookups, then reset mprts_hook.
24089 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24091 (try_to_simplify): Likewise.
24093 2017-06-19 Martin Liska <mliska@suse.cz>
24096 * gimplify.c (gimplify_switch_expr):
24097 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24099 2017-06-19 Martin Liska <mliska@suse.cz>
24101 * doc/install.texi: Document that PGO runs in 4 stages.
24103 2017-06-19 Martin Liska <mliska@suse.cz>
24106 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24107 to dispatcher function name.
24108 * multiple_target.c (replace_function_decl): New function.
24109 (create_dispatcher_calls): Redirect both edges and references.
24111 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24113 * profile-count.c (profile_count::dump): Dump quality.
24114 (profile_count::differs_from_p): Update for unsigned val.
24115 * profile-count.h (profile_count_quality): New enum.
24116 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24118 2017-06-19 Richard Biener <rguenther@suse.de>
24120 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24121 struct function as arg.
24122 (estimate_numbers_of_iterations): Export overload with loop arg.
24123 (free_numbers_of_iterations_estimates_loop): Use an overload of
24124 free_numbers_of_iterations_estimates instead.
24125 * tree-cfg.c (remove_bb): Adjust.
24126 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24127 * tree-parloops.c (gen_parallel_loop): Likewise.
24128 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24130 (tree_unroll_loops_completely): Likewise.
24131 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24132 Use an overload instead and export.
24133 (estimated_loop_iterations): Adjust.
24134 (max_loop_iterations): Likewise.
24135 (likely_max_loop_iterations): Likewise.
24136 (estimate_numbers_of_iterations): Take struct function as arg
24138 (loop_exits_before_overflow): Adjust.
24139 (free_numbers_of_iterations_estimates_loop): Use an overload.
24140 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24141 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24143 2017-06-19 Richard Biener <rguenther@suse.de>
24146 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24147 RANGE_EXPR conservatively.
24149 2017-06-16 Carl Love <cel@us.ibm.com>
24151 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24152 definitions for vec_float, vec_float2, vec_floato,
24153 vec_floate built-ins.
24154 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24155 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24157 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24158 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24159 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24160 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24161 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24162 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24163 vec_floato): Add builtin defines.
24164 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24165 Update the built-in documentation file for the new built-in
24168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24170 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24171 (mthumb): Mark as the negative of -marm.
24173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24175 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24177 (ARM Options, -mtune): Document that this accepts the same
24178 extension options as -mcpu.
24179 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24181 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24183 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24184 permitted extensions.
24186 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24188 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24189 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24190 (armv8-m.main): Add option +nodsp.
24191 * config/arm/arm-cpu-cdata.h: Regenerated.
24193 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24195 * config/arm/t-fuchsia: New file.
24196 * config.gcc (arm*-*-fuchsia*): Use it.
24198 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24200 * config/arm/t-symbian: Rewrite for new option infrastructure.
24202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24204 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24205 (MULTILIB_REQUIRED): Likewise.
24207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24209 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24210 (MULTILIB_RESUE): Likewise.
24211 (MULTILIB_MATCHES): Likewise.
24212 (MULTLIB_REQUIRED): Likewise.
24214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24216 * config/arm/t-rtems: Rewrite for new option framework.
24218 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24220 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24221 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24222 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24223 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24224 * config/arm/t-multilib: ... here.
24225 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24226 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24227 armv7-a and armv8*-a when A-profile libraries have not been built.
24228 * config/arm/t-rmprofile: Rewrite.
24230 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24232 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24233 with a backslash. Remove the backslash after substituting unescaped
24235 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24237 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24239 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24240 the driver to pass through the default setting of -mfloat-abi.
24241 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24243 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24244 (all_feat_combs): New rule.
24245 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24247 * config/arm/t-aprofile: Rewrite.
24249 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24251 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24252 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24253 fpu is not specified by the user/command-line.
24254 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24255 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24256 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24257 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24258 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24259 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24261 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24263 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24264 * config/arm/t-arm-elf: Rewritten.
24266 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24268 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24269 have some floating-point instructions.
24270 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24271 (TARGET_MAYBE_HARD_FLOAT): New macro.
24272 * config/arm/arm-builtins.c (arm_init_builtins): Use
24273 TARGET_MAYBE_HARD_FLOAT.
24274 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24278 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24279 (configargs.h): Include it.
24280 (arm_print_hint_for_fpu_option): New function.
24281 (arm_parse_fpu_option): New function.
24282 (candidate_extension): New class.
24283 (arm_canon_for_multilib): New function.
24284 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24285 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24286 (ARCH_CANONICAL_SPECS): New macro.
24287 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24291 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24292 are set after handling multilib fragments. Set target_cpu_default2
24295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24297 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24299 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24300 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24302 (check_arch): Likewise.
24303 * config/arm/arm.c (arm_configure_build_target): Handle
24304 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24305 options in the default.
24307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24309 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24310 when an option is an alias of another.
24311 * config/arm/parsecpu.awk (optalias): New parser token.
24312 (gen_comm_data): Mark non-alias options as such. Emit entries
24313 for extension aliases.
24314 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24315 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24316 (armv6kz, armv6zk, armv6t2): Likewise.
24317 (armv7): Make vfpv3-d16 an alias.
24318 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24320 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24321 Sort in canonical order.
24322 (armv8-a): Sort in canonical order.
24323 (armv8.1-a, armv8.2-a): Likewise.
24324 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24326 (cortex-a9): Sort in canonical order.
24327 * config/arm/arm.c (selftests.h): Include it.
24328 (arm_test_cpu_arch_data): New function.
24329 (arm_run_self_tests): New function.
24330 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24331 (targetm): Move declaration to the end of the file.
24332 * arm-cpu-cdata.h: Regenerated.
24334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24336 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24337 call to target_mode_check describing the type of option passed.
24338 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24339 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24340 arm_parse_cpu_option_name to match parameters against list of
24342 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24343 arm_arch_core_flags data structure.
24344 * config/arm/arm-cpu_cdata.h: Regenerated.
24346 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24348 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24350 (arm_print_hint_for_cpu_option): Likewise.
24351 (arm_print_hint_for_arch_option): Likewise.
24352 (arm_parse_cpu_option_name): Likewise.
24353 (arm_parse_arch_option_name): Likewise.
24354 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24355 of entries in the all_fpus list.
24356 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24357 (arm_parse_cpu_option_name): Declare.
24358 (arm_parse_arch_option_name): Declare.
24359 (arm_parse_option_features): Declare.
24360 (arm_intialize_isa): Declare.
24361 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24363 (gen_comm_data): ... here. Make definitions non-static.
24364 * config/arm/arm-cpu-data.h: Regenerated.
24365 * config/arm/arm-cpu-cdata.h: Regenerated.
24367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24369 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24370 (cpu_arch_extension): New structure.
24371 (cpu_arch_option, arch_option, cpu_option): New structures.
24372 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24373 architecture types.
24374 (gen_data): Generate new format data tables.
24375 * config/arm/arm.c (cpu_tune): New structure.
24376 (cpu_option, processors): Delete.
24377 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24378 (arm_print_hint_for_cpu_option): ... this and ...
24379 (arm_print_hint_for_arch_option): ... this.
24380 (arm_parse_arch_cpu_name): Delete. Replace with ...
24381 (arm_parse_cpu_option_name): ... this and ...
24382 (arm_parse_arch_option_name): ... this.
24383 (arm_unrecognized_feature): Change type of target parameter to
24385 (arm_parse_arch_cpu_features): Delete. Replace with ...
24386 (arm_parse_option_features): ... this.
24387 (arm_configure_build_target): Rework to use new configuration data
24389 (arm_print_tune_info): Rework for new configuration data tables.
24390 * config/arm/arm-cpu-data.h: Regenerated.
24391 * config/arm/arm-cpu.h: Regenerated.
24393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24395 * Makefile.in (OBJS): Move sbitmap.o from here ...
24396 (OBJS-libcommon): ... to here.
24398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24400 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24401 (ISA_ALL_CRYPTO): New macro.
24402 (ISA_ALL_SIMD): New macro
24403 (ISA_ALL_FP): New macro.
24404 * config/arm/arm.c (fpu_bitlist): Update initializer.
24405 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24407 (arm9e): Add fpu. Add option for nofp
24408 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24409 (arm926ej-s, arm1026ej-s): Likewise.
24410 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24411 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24412 neon-fp16, neon-vfpv4, nofp and nosimd.
24413 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24414 (cortex-a8): Add fpu. Add option for nofp.
24415 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24416 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24417 (cortex-r4f): Add fpu.
24418 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24419 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24421 (cortex-r8): Likewise.
24422 (cortex-m4): Add fpu. Add option for nofp.
24423 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24424 (cortex-a17.cortex-a7): Likewise.
24425 (cortex-a32): Add fpu. Add options for crypto and nofp.
24426 (cortex-a35, cortex-a53): Likewise.
24427 (cortex-a57): Add fpu. Add option for crypto.
24428 (cortex-a72, cortex-a73): Likewise.
24429 (exynos-m1): Likewise.
24430 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24431 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24432 (cortex-m33): Add fpu. Add option for nofp.
24433 * config/arm/arm-cpu-cdata.h: Regenerated
24434 * config/arm/arm-cpu-data.h: Regenerated.
24436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24438 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24439 (armv5te, armv5tej): Likewise.
24440 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24441 (armv7): Add options fp and vfpv3-d16.
24442 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24443 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24445 (armv7ve): Likewise.
24446 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24447 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24448 (armv8-a): Add nocrypto option.
24449 (armv8.1-a, armv8.2-a): Likewise.
24450 (armv8-m.main): add options fp, fp.dp and nofp.
24452 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24454 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24456 (armv8-a+crc): Delete.
24457 (armv8.1-a): Add options simd, crypto and nofp.
24458 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24459 (armv8.2-a+fp16): Delete.
24460 (armv8-m.main): Add option dsp.
24461 (armv8-m.main+dsp): Delete.
24462 (cortex-a8): Add fpu. Add nofp option.
24463 (cortex-a9): Add fpu. Add nofp and nosimd options.
24464 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24465 link to main cpu and architecture data structures.
24466 (gen_comm_data): Only put isa attributes from the main architecture
24468 (option): New statement for architecture and CPU entries.
24469 * arm.c (struct cpu_option): New structure.
24470 (struct processors): Add entry for options.
24471 (arm_unrecognized_feature): New function.
24472 (arm_parse_arch_cpu_name): Ignore any characters after the first
24474 (arm_parse_arch_cpu_feature): New function.
24475 (arm_configure_build_target): Separate out any CPU and architecture
24476 features and parse separately. Don't error out if -mfpu=auto is
24477 used with only an architecture string.
24478 (arm_print_asm_arch_directives): New function.
24479 (arm_file_start): Call it.
24480 * config/arm/arm-cpu-cdata.h: Regenerated.
24481 * config/arm/arm-cpu-data.h: Likewise.
24482 * config/arm/arm-tables.opt: Likewise.
24484 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24486 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24487 assembler when it is not -mfpu=auto.
24489 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24491 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24492 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24493 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24494 (ASM_CPU_SPEC): Rewrite.
24495 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24496 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24497 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24498 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24499 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24500 copied string is NUL-terminated. Also strip any characters prefixed
24502 (arm_rewrite_selected_arch): New function.
24503 (arm_rewrite_march): New function.
24505 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24507 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24508 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24509 (march, mcpu, mtune): Convert to string-based options.
24510 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24511 (arm_parse_arch_cpu_name): New function.
24512 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24513 identify selected architecture or CPU.
24514 (arm_option_save): New function.
24515 (TARGET_OPTION_SAVE): Redefine.
24516 (arm_option_restore): Restore string options.
24517 (arm_option_print): Print string options.
24519 2017-06-16 Martin Sebor <msebor@redhat.com>
24521 PR tree-optimization/80933
24522 PR tree-optimization/80934
24523 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24524 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24525 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24526 (gimple_fold_builtin): Call them.
24528 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24530 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24531 as unlikely; update profile.
24533 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24535 * predict.c (force_edge_cold): Handle declaring edges impossible
24538 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24540 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24542 (try_unroll_loop_completely): Fix reporting.
24544 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24546 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24548 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24551 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24552 if given a non-constant argument for an intrinsic which requires a
24555 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24557 * profile.c (compare_freqs): New function.
24558 (branch_prob): Sort edge list.
24559 (find_spanning_tree): Assume that the list is priority sorted.
24561 2017-06-16 Richard Biener <rguenther@suse.de>
24563 PR tree-optimization/81090
24564 * passes.def (pass_record_bounds): Remove.
24565 * tree-pass.h (make_pass_record_bounds): Likewise.
24566 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24567 make_pass_record_bounds): Likewise.
24568 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24569 not free niter estimates at the beginning but at the end.
24570 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24572 2017-06-16 Richard Biener <rguenther@suse.de>
24574 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24575 initializer to workaround ICE in host GCC 4.8.
24577 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24579 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24581 (clone_inlined_nodes): Update.
24583 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24585 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24586 prefetch settings, and enable prefetching by default at -O3.
24588 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24590 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24591 Set flag_prefetch_loop_arrays according to tuning data.
24593 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24595 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24596 New tune structure.
24597 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24598 [Unrelated to main purpose of the patch] Place the pointer field last
24599 to enable type checking errors when tune structure are wrongly merged.
24600 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24601 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24602 (thunderx2t99_prefetch_tune): New tune constants.
24603 (tune_params *_tunings): Update all tunings (no functional change).
24604 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24605 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24606 from tunings structures.
24608 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24611 * ubsan.c (instrument_null): Add T argument, use it instead
24612 of computing it based on IS_LHS.
24613 (instrument_object_size): Likewise.
24614 (pass_ubsan::execute): Adjust instrument_null and
24615 instrument_object_size callers to pass gimple_get_lhs or
24616 gimple_assign_rhs1 result to it. Use instrument_null instead of
24617 calling get_base_address and instrument_mem_ref. Handle
24618 aggregate call arguments for object-size sanitization.
24620 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24622 PR tree-optimization/81089
24623 * tree-vrp.c (is_masked_range_test): Validate operands of
24626 2017-06-15 Martin Sebor <msebor@redhat.com>
24629 * dumpfile.c (dump_register): Avoid calling memset to initialize
24630 a class with a default ctor.
24631 * gcc.c (struct compiler): Remove const qualification.
24632 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24633 * hash-table.h: Ditto.
24634 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24636 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24637 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24639 * params.h (struct param_info): Make struct members non-const.
24640 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24641 with default initialization.
24642 * vec.h (vec_copy_construct, vec_default_construct): New helper
24644 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24645 with vec_copy_construct.
24646 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24647 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24648 * doc/invoke.texi (-Wclass-memaccess): Document.
24650 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24652 * emit-rtl.h (is_leaf): Update comment about local
24653 register allocator.
24655 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24658 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24659 for a variable to have a section before checking if the section has a
24661 Set section to.persistent if persistent attribute is set.
24662 Warn if .persistent attribute is used on an automatic variable.
24664 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24666 PR rtl-optimization/80474
24667 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24669 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24671 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24674 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24676 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24677 (casesi): Emit bounds checking as RTL.
24678 (casesi_internal_mips16_<mode>): Remove bounds checking.
24680 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24682 * config/xtensa/xtensa.c (xtensa_option_override): Append
24683 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24684 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24685 xtensa_doloop_hooks): Define unconditionally.
24686 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24688 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24689 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24690 for it in xtensa_option_override.
24691 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24692 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24694 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24696 * doc/cppopts.texi: Document '-' special value to -MF.
24698 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24700 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24701 (cortex_a53_fconst): Likewise.
24702 (cortex_a53_fpmul): Likewise.
24703 (cortex_a53_f_load_64): Likewise.
24704 (cortex_a53_f_load_many): Likewise.
24705 (cortex_a53_advsimd_alu): Likewise.
24706 (cortex_a53_advsimd_alu_q): Likewise.
24707 (cortex_a53_advsimd_mul): Likewise.
24708 (cortex_a53_advsimd_mul_q): Likewise.
24709 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24710 Add missing fmul, r2f_cvt and fconst cases.
24712 2017-06-14 Richard Biener <rguenther@suse.de>
24714 PR middle-end/81088
24715 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24717 (fold_binary_loc): When associating do not treat pre-existing
24718 TREE_OVERFLOW on literal constants as a reason to allow
24719 TREE_OVERFLOW on associated literal constants.
24721 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24723 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24724 (MASK_FEATURES): New macro.
24725 * config/sparc/sparc.c (sparc_option_override): Remove the special
24726 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24728 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24730 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24731 a division of 0 if non-call exceptions are enabled.
24733 2017-06-14 Andrew Pinski <apinski@cavium.com>
24734 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24737 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24738 Improve vector initialization code gen for only variable case.
24740 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24742 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24744 2017-06-14 Richard Biener <rguenther@suse.de>
24746 PR tree-optimization/81083
24747 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24750 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24752 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24753 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24754 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24755 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24756 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24757 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24759 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24761 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24762 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24764 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24766 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24768 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24770 * config/rs6000/t-rtems: Don't handle SPE.
24772 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24774 * config/rs6000/t-linux: Don't handle SPE.
24776 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24778 * config/rs6000/eabispe.h: Delete file.
24780 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24782 * config/rs6000/t-spe: Delete file.
24784 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24786 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24787 (rs6000_legitimate_offset_address_p): Return false for anything in
24788 V2SImode or V2SFmode.
24790 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24792 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24793 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24794 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24796 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24797 (rs6000_legitimate_offset_address_p): Ditto.
24798 (rs6000_emit_move): Ditto.
24799 (rs6000_init_builtins): Remove V4HI_type_node.
24801 2017-06-13 Martin Liska <mliska@suse.cz>
24804 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24805 (gate_asan): Likewise.
24806 * asan.h (asan_no_sanitize_address_p): Remove the function.
24807 (sanitize_flags_p): New function.
24808 * builtins.def: Fix coding style.
24809 * common.opt: Use renamed enum value.
24810 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24811 * doc/extend.texi: Document no_sanitize attribute.
24812 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24813 to SANITIZE_UNDEFINED_NONDEFAULT.
24814 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24815 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24816 Use sanitize_flags_p.
24817 * gimplify.c (gimplify_function_tree): Likewise.
24818 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24819 * opts.c (parse_no_sanitize_attribute): New function.
24820 (common_handle_option): Use renamed enum value.
24821 * opts.h (parse_no_sanitize_attribute): Declare.
24822 * tree.c (sanitize_flags_p): New function.
24823 * tree.h: Declared here.
24824 * tsan.c: Use sanitize_flags_p.
24825 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24826 (instrument_mem_ref): Likewise.
24827 (instrument_bool_enum_load): Likewise.
24828 (do_ubsan_in_current_function): Remove the function.
24829 (pass_ubsan::execute): Use sanitize_flags_p.
24830 * ubsan.h: Remove do_ubsan_in_current_function
24831 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24832 (dump_function_to_file): Use it here.
24834 2017-06-13 Martin Jambor <mjambor@suse.cz>
24836 PR tree-optimization/80803
24837 PR tree-optimization/81063
24838 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24839 (propagate_subaccesses_across_link): Enqueue subtree whenever
24840 necessary instead of relying on the caller.
24842 2017-06-13 Martin Jambor <mjambor@suse.cz>
24844 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24845 that have a first_link.
24846 (sort_and_splice_var_accesses): Do not check first_link before
24848 (subtree_mark_written_and_enqueue): Likewise.
24849 (propagate_all_subaccesses): Likewise and do not stop at first
24850 parent with a first_link.
24852 2017-06-13 Martin Jambor <mjambor@suse.cz>
24854 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24857 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24859 * match.pd: New pattern.
24861 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24863 * tree-vrp.c (is_masked_range_test): New function.
24864 (register_edge_assert_for): Determine ranges for
24867 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24869 PR tree-optimization/67328
24870 * fold-const.c (maskable_range_p): New function.
24871 (build_range_check): Generate bittests if possible.
24873 2017-06-13 Martin Liska <mliska@suse.cz>
24875 * gimple-pretty-print.c (dump_probability): Add new argument.
24876 (dump_edge_probability): Dump both probability and count.
24877 (dump_gimple_label): Likewise.
24878 (dump_gimple_bb_header): Likewise.
24880 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24883 * config/avr/avr-devices.c: Fix indentation.
24884 * config/avr/gen-avr-mmcu-specs.c: Dito.
24886 2017-06-13 Richard Biener <rguenther@suse.de>
24888 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24889 instead get vector type from stmt_info.
24890 (vectorizable_reduction): Adjust. Remove dead code.
24892 2017-06-13 Richard Biener <rguenther@suse.de>
24894 PR middle-end/81065
24895 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24896 case of C * (x * C2 + C3).
24897 (fold_addr_of_array_ref_difference): Properly fold index difference.
24899 2017-06-12 David S. Miller <davem@davemloft.net>
24902 * config/sparc/sparc.md (return expander): Emit frame blockage if
24903 function uses alloca.
24905 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24907 * combine.c (make_field_assignment): Check len rather than the mode
24908 precision when calling force_to_mode.
24910 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24912 Support multilibs and devices that see flash in RAM address range.
24915 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24916 (avr_mcu_t) <flash_pm_offset>: New field.
24917 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24918 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24919 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24920 (AVR_TINY_PM_OFFSET): Remove macro.
24921 * config/avr/avr.opt (-mshort-calls): New option.
24922 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24923 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24924 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24925 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24926 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24927 instead of avr_arch->have_jmp_call.
24928 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24929 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24930 avr_arch->flash_pm_offset to define.
24931 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24932 new field flash_pm_offset. Add entry for avrxmega3.
24933 (avr_texinfo): Add entry for avrxmega3.
24934 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24935 attiny212, attiny214,
24936 attiny412, attiny414, attiny416, attiny417,
24937 attiny814, attiny816, attiny817,
24938 attiny1614, attiny1616, attiny1617,
24939 attiny3214, attiny3216, attiny3217.
24940 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24941 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24942 (avr_print_operand_address) [AVR_TINY]: Same.
24943 (avr_asm_init_sections) <readonly_data_section>: Only patch
24944 callback if avr_arch->flash_pm_offset = 0.
24945 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24946 for rodata if avr_arch->flash_pm_offset != 0.
24947 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24948 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24949 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24950 (m_options): Append opt_rcall.
24951 (m_dirnames): Append dir_rcall.
24952 * config/avr/t-multilib: Regenerate.
24954 * configure.ac [target=avr]: Check whether avrxmega3 default
24955 linker description file works as needed.
24956 * configure: Regenerate.
24957 * doc/avr-mmcu.texi: Regenerate.
24958 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24959 <__AVR_ARCH__>: Document avrxmega3 and 103.
24960 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24961 <__AVR_SHORT_CALLS__>: Document it.
24962 <__AVR_PM_BASE_ADDRESS__>: Document it.
24963 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24964 (AVR Variable Attributes) <progmem>: Document this is
24965 not needed for avrxmega3.
24966 (AVR Named Address Spaces) <__flash>: Dito.
24968 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24970 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24972 2017-06-12 Doug Rupp <rupp@adacore.com>
24974 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24975 Append vxworks-stdint.h to the tm_file list.
24976 * config/vxworks-stdint.h: New file.
24978 2017-06-12 Martin Liska <mliska@suse.cz>
24980 PR tree-optimization/81041
24981 * tree-profile.c (gimple_gen_ic_func_profiler):
24982 Create an extra BB in profile-generate
24983 (gimple_gen_time_profiler): Likewise.
24985 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24987 PR tree-optimization/81003
24988 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24989 (update_range_test): Use it instead of force_gimple_operand_gsi.
24991 2017-06-12 Richard Biener <rguenther@suse.de>
24993 PR tree-optimization/81053
24994 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24995 with backedge value not defined in loop. Simplify def stmt
24998 2017-06-11 Tom de Vries <tom@codesourcery.com>
25001 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25003 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25004 nvptx_cannot_force_const_mem.
25006 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25008 * opts.c (finish_options): Move test for flag_split_stack after
25009 it has been initialized.
25011 2017-06-11 Jason Merrill <jason@redhat.com>
25013 * tree.h (id_equal): New.
25014 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25015 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25016 instead of strcmp of IDENTIFIER_POINTER.
25018 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25020 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25021 (mark_all_inlined_calls_cdtor): Fix formating.
25022 (inline_transform): Rescale profile before inlining.
25024 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25026 * cgraph.h (cgraph_edge::clone): Update prototype.
25027 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25028 (cgraph_node::create_clone): Update.
25029 (cgraph_node::create_version_clone): Update.
25030 * tree-inline.c (copy_bb): Update.
25031 (expand_call_inline): Update.
25033 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25035 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25036 factored out from ...
25037 (rs6000_emit_prologue): ... here.
25039 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25041 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25042 factored out from ...
25043 (rs6000_emit_prologue): ... here.
25045 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25047 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25049 (handle_missing_profiles): Fix computation of tp_first_run.
25050 (counts_to_freqs): Do not touch freqs when count is 0.
25052 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25054 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25057 2017-06-10 Tom de Vries <tom@codesourcery.com>
25059 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25060 attributes): Document signal effective target.
25062 2017-06-10 Tom de Vries <tom@codesourcery.com>
25064 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25065 Document effective target stack_size.
25067 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25069 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25070 to the edit_context if they can be auto-applied.
25072 2017-06-9 Ian Lance Taylor <iant@golang.org>
25074 * opts.c (finish_options): If -fsplit-stack, disable implicit
25075 -forder-blocks-and-partition.
25076 * doc/invoke.texi (Optimize Options): Document that when using
25077 -fsplit-stack -forder-blocks-and-partition is not implicitly
25080 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25082 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25083 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25084 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25085 * builtins.def (abort, trap, unreachable): Declare cold.
25086 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25087 * tree-core.h (ECF_COLD): New.
25088 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25089 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25091 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25093 * predict.c (unlikely_executed_stmt_p): Cleanup.
25095 2017-06-09 Richard Biener <rguenther@suse.de>
25097 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25098 model if the ref is always written to.
25100 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25102 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25104 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25106 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25109 2017-06-09 Tom de Vries <tom@codesourcery.com>
25112 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25113 "target cannot support label values" when encountering LABEL_REF.
25115 2017-06-09 Martin Liska <mliska@suse.cz>
25117 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25118 (gimple_gen_ic_func_profiler): Emit direct comparison
25119 of __gcov_indirect_call_callee with NULL.
25120 (gimple_gen_time_profiler): Change probability from
25121 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25123 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25125 * profile.c (edge_gcov_counts): Turn to pointer.
25126 (compute_branch_probabilities, compute_branch_probabilities): Update.
25127 (branch_prob): Do not clear edge_gcov_count.
25128 * profile.h (edge_gcov_counts): Turn to pointer.
25129 (edge_gcov_count): Update.
25131 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25133 * gimple.h (gimple_check_failed): Mark cold.
25135 2017-06-09 Richard Biener <rguenther@suse.de>
25137 PR tree-optimization/66623
25138 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25139 refactor check_reduction into two parts, properly computing
25140 whether we have to check reduction validity for outer loop
25143 2017-06-09 Richard Biener <rguenther@suse.de>
25145 PR tree-optimization/79483
25146 * graphite-scop-detection.c (order): New global.
25147 (get_order): Compute bb to order mapping that satisfies code
25148 generation constraints.
25149 (cmp_pbbs): New helper.
25150 (build_scops): Start domwalk at entry block, sort generated
25153 2017-06-09 Richard Biener <rguenther@suse.de>
25155 PR middle-end/81007
25156 * ipa-polymorphic-call.c
25157 (ipa_polymorphic_call_context::restrict_to_inner_class):
25158 Skip FIELD_DECLs with error_mark_node type.
25159 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25162 2017-06-09 Martin Liska <mliska@suse.cz>
25164 * predict.c (struct branch_predictor): New struct.
25165 (test_prediction_value_range): New test.
25166 (predict_c_tests): New function.
25167 * selftest-run-tests.c (selftest::run_tests): Run the function.
25168 * selftest.h: Declare new tests.
25170 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25173 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25174 gen_add3_insn did not fail.
25175 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25176 r0, construct that number in a temporary reg and add that reg to r0.
25177 If asked to put the result in r0 as well, fail.
25179 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25181 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25182 for early expansion of vec_eqv.
25184 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25186 PR middle-end/81005
25187 * ubsan.c (instrument_null): Avoid pointless code temporary.
25188 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25190 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25194 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25197 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25200 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25202 * predict.c (unlikely_executed_edge_p): Move ahead.
25203 (probably_never_executed_edge_p): Use it.
25205 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25207 PR middle-end/79988
25208 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25209 gimple_call_builtin_p call.
25211 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25213 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25214 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25215 rtl_check_failed_type2, rtl_check_failed_code1,
25216 rtl_check_failed_code2, rtl_check_failed_code_mode,
25217 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25218 rtvec_check_failed_bounds, rtl_check_failed_flag,
25219 _fatal_insn_not_found, _fatal_insn): Likewise.
25220 * tree.h (tree_contains_struct_check_failed,
25221 tree_check_failed, tree_not_check_failed,
25222 tree_class_check_failed, tree_range_check_failed,
25223 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25224 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25225 tree_operand_check_failed, omp_clause_check_failed,
25226 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25229 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25231 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25232 flag_branch_probabilities.
25233 * ipa-inline.c (edge_badness): Likewise.
25234 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25235 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25236 * predict.c (maybe_hot_frequency_p): Likewise.
25237 (probably_never_executed): Likewise.
25238 * sched-ebb.c (schedule_ebbs): Likewise.
25239 * sched-rgn.c (find_single_block_region): Likewise.
25240 * tracer.c (tail_duplicate): Likewise.
25242 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25244 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25245 longer requires x_flag_profile_use.
25247 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25249 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25250 instead of flag_reorder_blocks_and_partition.
25251 * dbxout.c (dbxout_function_end): Likewise.
25252 * dwarf2out.c (gen_subprogram_die): Likewise.
25253 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25254 * hw-doloop.c (reorg_loops): Likewise.
25255 * varasm.c (assemble_start_function,
25256 assemble_end_function): Likewise.
25257 (decide_function_section): Do not check for
25258 flag_reorder_blocks_and_partition.
25260 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25262 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25264 (chkp_get_hard_register_fake_addr_expr): Ditto.
25265 (chkp_build_addr_expr): Add check for hard reg case.
25266 (chkp_parse_array_and_component_ref): Ditto.
25267 (chkp_find_bounds_1): Ditto.
25268 (chkp_process_stmt): Don't generate bounds store for
25271 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25273 * predict.c (maybe_hot_bb_p): Do not check profile status.
25274 (maybe_hot_edge_p): Likewise.
25275 (probably_never_executed): Check for zero counts even if profile
25277 (unlikely_executed_edge_p): New function.
25278 (unlikely_executed_stmt_p): New function.
25279 (unlikely_executed_bb_p): New function.
25280 (set_even_probabilities): Use unlikely predicates.
25281 (combine_predictions_for_bb): Likewise.
25282 (predict_paths_for_bb): Likewise.
25283 (predict_paths_leading_to_edge): Likewise.
25284 (determine_unlikely_bbs): New function.
25285 (estimate_bb_frequencies): Use it.
25286 (compute_function_frequency): Use zero counts even if profile is
25288 * profile-count.h: Fix typo.
25290 2017-08-08 Julia Koval <julia.koval@intel.com>
25292 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25293 _mm512_mask_cvtsepi16_storeu_epi8,
25294 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25295 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25296 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25297 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25298 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25299 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25300 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25301 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25302 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25303 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25304 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25305 __builtin_ia32_pmovuswb256mem_mask,
25306 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25307 __builtin_ia32_pmovwb512mem_mask): New builtins.
25309 2017-08-08 Julia Koval <julia.koval@intel.com>
25311 PR target/73350,80862
25312 * config/i386/subst.md (round): Fix round pattern.
25313 * config/i386/i386.c (ix86_erase_embedded_rounding):
25314 Fix erasing rounding for the fixed pattern.
25316 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25318 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25320 2017-06-08 Martin Liska <mliska@suse.cz>
25322 PR gcov-profile/80911
25323 * gcov.c (block_info::block_info): New constructor.
25325 2017-06-07 Carl Love <cel@us.ibm.com>
25327 * config/rs6000/rs6000-c: The return type of the following
25328 built-in functions was implemented as int not long long. Fix sign
25329 of return value for the unsigned version of vec_mulo and vec_mule.
25330 vector unsigned long long vec_bperm (vector unsigned long long,
25331 vector unsigned char)
25332 vector signed long long vec_mule (vector signed int,
25334 vector unsigned long long vec_mule (vector unsigned int,
25335 vector unsigned int)
25336 vector signed long long vec_mulo (vector signed int,
25338 vector unsigned long long vec_mulo (vector unsigned int,
25339 vector unsigned int)
25340 * doc/extend.texi: Fix the documentation for the built-in
25343 2017-06-07 Carl Love <cel@us.ibm.com>
25346 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25349 2017-06-07 Carl Love <cel@us.ibm.com>
25351 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25352 support, Generate doublehv for signed int/float for BE case only.
25354 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25356 * doc/invoke.texi (mcx16): Rewrite.
25358 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25360 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25361 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25362 *mov<mode>_softfloat, and an anonymous splitter): Use
25363 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25365 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25367 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25369 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25370 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25371 (rs6000_debug_reg_global): Adjust.
25372 (rs6000_init_hard_regno_mode_ok): Adjust.
25373 (rs6000_dbx_register_number): Adjust.
25374 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25375 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25376 Remove SPE_ACC and SPEFSCR.
25377 (REG_ALLOC_ORDER): Ditto.
25378 (FRAME_POINTER_REGNUM): Change to 111.
25379 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25380 (REG_CLASS_NAMES): Ditto.
25381 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25382 (REGISTER_NAMES): Ditto.
25383 (ADDITIONAL_REG_NAMES): Ditto.
25384 (rs6000_reg_names): Ditto.
25385 * config/rs6000/rs6000.md: Renumber some register number
25388 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25390 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25392 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25393 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25395 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25396 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25397 Delete the SPE high registers.
25398 (REG_ALLOC_ORDER): Ditto.
25399 (enum reg_class): Remove SPE_HIGH_REGS.
25400 (REG_CLASS_NAMES): Ditto.
25401 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25402 (REGISTER_NAMES): Ditto.
25403 (rs6000_reg_names): Ditto.
25404 * doc/tm.texi.in: Remove SPE as example.
25405 * doc/tm.texi: Regenerate.
25407 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25409 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25410 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25411 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25412 * config/rs6000/rs6000.md (type): Remove "brinc".
25414 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25416 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25417 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25418 * config/rs6000/linuxspe.h: Delete file.
25419 * config/rs6000/rs6000.md: Don't include spe.md.
25420 * config/rs6000/spe.h: Delete file.
25421 * config/rs6000/spe.md: Delete file.
25422 * config/rs6000/t-rs6000: Remove spe.md.
25424 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25426 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25427 (reg_or_none500mem_operand): Delete.
25428 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25429 instead of reg_or_none500mem_operand.
25431 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25433 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25434 handling of SPE flags.
25435 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25437 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25439 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25441 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25442 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25443 paired_divv2sf3): Similar.
25444 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25445 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25446 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25448 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25449 Rename the paired_* instruction patterns.
25450 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25452 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25453 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25454 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25455 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25456 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25457 PAIRED_VECTOR_MODE.
25458 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25459 (spe_func_has_64bit_regs_p): Delete.
25460 (spe_expand_predicate_builtin): Delete.
25461 (spe_expand_evsel_builtin): Delete.
25462 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25463 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25464 (invalid_e500_subreg): Delete.
25465 (rs6000_legitimize_address): Always force_reg op2 as well, for
25466 paired single memory accesses.
25467 (rs6000_member_type_forces_blk): Delete.
25468 (rs6000_spe_function_arg): Delete.
25469 (rs6000_expand_unop_builtin): Delete SPE handling.
25470 (rs6000_expand_binop_builtin): Ditto.
25471 (spe_expand_stv_builtin): Delete.
25472 (bdesc_2arg_spe): Delete.
25473 (spe_expand_builtin): Delete.
25474 (spe_expand_predicate_builtin): Delete.
25475 (spe_expand_evsel_builtin): Delete.
25476 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25477 (spe_init_builtins): Delete.
25478 (spe_func_has_64bit_regs_p): Delete.
25479 (savres_routine_name): Delete "info" parameter. Adjust callers.
25480 (rs6000_emit_stack_reset): Ditto.
25481 (rs6000_dwarf_register_span): Delete.
25482 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25483 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25484 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25486 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25488 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25489 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25490 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25491 mulv2sf3, divv2sf3): Delete expanders.
25493 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25495 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25497 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25499 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25500 * config/rs6000/rs6000.c: Ditto.
25502 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25504 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25505 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25506 comparison_operator.
25508 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25510 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25511 * config/rs6000/rs6000.opt: Ditto.
25512 * config/rs6000/t-rtems: Ditto.
25514 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25516 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25517 TARGET_E500_SINGLE by 0, simplify.
25518 * config/rs6000/rs6000.c: Ditto.
25519 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25520 (spe_build_register_parallel): Delete.
25521 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25522 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25523 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25524 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25525 (E500_CONVERT): Delete.
25526 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25528 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25530 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25531 * config/rs6000/dfp.md: Ditto.
25532 (negdd2, *negdd2_fpr): Merge.
25533 (absdd2, *absdd2_fpr): Merge.
25534 (negtd2, *negtd2_fpr): Merge.
25535 (abstd2, *abstd2_fpr): Merge.
25536 * config/rs6000/e500.h: Delete file.
25537 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25538 TARGET_FPRS by 1 and simplify.
25539 * config/rs6000/rs6000-c.c: Ditto.
25540 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25541 TARGET_DF_SPE by 0.
25542 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25544 * config/rs6000/rs6000.md: Ditto.
25545 (floatdidf2, *floatdidf2_fpr): Merge.
25546 (move_from_CR_gt_bit): Delete.
25547 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25548 (E500_CR_IOR_COMPARE): Delete.
25549 (All patterns that require !TARGET_FPRS): Delete.
25550 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25552 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25554 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25556 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25558 * graphds.c (add_edge): Intitialize edge's attached data.
25559 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25560 pointer parameter. Call pointed function on each edge during
25561 graph traversing. Skip traversing the edge when the function
25563 (graphds_dfs, graphds_scc): Ditto.
25564 (for_each_edge): New parameter. Pass the new parameter to callback
25566 * graphds.h (skip_edge_callback): New function pointer type.
25567 (graphds_dfs, graphds_scc): New function pointer parameter.
25568 (graphds_edge_callback, for_each_edge): New parameter.
25570 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25572 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25573 out code checking if runtime alias check is possible to below ...
25574 Call the new function.
25575 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25576 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25578 2017-06-07 Marek Polacek <polacek@redhat.com>
25581 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25582 TYPE_OVERFLOW_WRAPS check.
25584 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25586 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25587 if versioning is required.
25588 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25589 peeling with the check for versioning.
25591 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25593 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25594 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25595 Set true to new parameter if new ssa variable is defined.
25596 (vect_gen_vector_loop_niters): Refactor. Set range information
25597 for the new vector loop bound variable.
25598 (vect_do_peeling): Ditto.
25600 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25602 * tree-affine.c (ssa.h): Include header file.
25603 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25604 has wrapping overflow behavior.
25606 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25608 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25610 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25612 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25613 (tree_to_aff_combination): ... here.
25615 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25617 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25618 reg_pressure model function.
25619 (ivopts_global_cost_for_size): Delete.
25620 (determine_set_costs, iv_ca_recount_cost): Call new model function
25621 ivopts_estimate_reg_pressure.
25623 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25625 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25626 expensive than udiv. Remove floating point cases from mod.
25628 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25630 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25631 Increase idiv cost.
25633 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25635 * config/aarch64/aarch64.md
25636 (copysignsf3): Fix mask generation.
25638 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25640 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25642 (class dump_manager): Add register_dumps method.
25643 * dumpfile.c: Include langhooks.h.
25644 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25645 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25646 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25647 (dump_manager::dump_register): Start with 512 entries instead of 32.
25648 (dump_manager::register_dumps): New method.
25649 * toplev.c (general_init): Instead of invoking register_dumps
25650 langhook, invoke register_dumps method on the dump manager.
25651 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25654 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25656 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25657 Say that # requires an associated define_split to exist, and that
25658 the define_split must be suitable for use after register allocation.
25660 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25662 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25663 (compute_outgoing_frequencies): Also initialize zero counts.
25664 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25665 around loops; preserve more of profile when nothing changes.
25667 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25669 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25671 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25672 * config/arm/arm-cpu-cdata.h: Regenerate.
25673 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25674 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25675 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25676 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25677 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25679 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25681 * config/arm/t-rmprofile: Likewise.
25682 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25684 2017-06-06 David S. Miller <davem@davemloft.net>
25687 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25688 blockage if function uses alloca.
25690 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25692 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25693 New "uid" fields to hold pretty-print IDs of group and ref.
25694 Memory references are now identified as <group_id>:<ref_id>
25695 instead of using [random] addresses.
25696 (dump_mem_details): Simplify, no functional change.
25697 (dump_mem_ref): Simplify and make output more concise.
25698 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25699 (find_or_create_group): Initialize group uid.
25700 (record_ref): Initialize ref uid. Improve debug output.
25701 (prune_group_by_reuse, should_issue_prefetch_p,)
25702 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25703 (mark_nontemporal_store, determine_loop_nest_reuse):
25704 Improve debug output.
25706 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25708 * dbgcnt.def (prefetch): New debug counter.
25709 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25710 (schedule_prefetches): Stop issueing prefetches if debug counter
25713 2017-06-06 Tom de Vries <tom@codesourcery.com>
25715 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25716 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25718 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25720 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25721 Use aarch64_reg_or_zero predicate for operand 4.
25722 (aarch64_compare_and_swap<mode> define_insn_and_split):
25723 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25724 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25726 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25728 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25729 (arm_compute_save_core_reg_mask): This.
25730 (thumb1_compute_save_reg_mask): Rename into ...
25731 (thumb1_compute_save_core_reg_mask): This.
25732 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25733 (arm_compute_frame_layout): Likewise.
25735 2017-06-06 Richard Biener <rguenther@suse.de>
25737 PR tree-optimization/80974
25738 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25739 keep or clear leaders SSA info.
25741 2017-06-06 Tom de Vries <tom@codesourcery.com>
25743 * config/nvptx/nvptx.c (split_mode_p): New function.
25744 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25746 2017-06-06 Tom de Vries <tom@codesourcery.com>
25748 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25750 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25753 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25756 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25758 * shrink-wrap.c (handle_simple_exit): Update profile.
25759 (try_shrink_wrapping): Upate profile.
25761 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25763 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25764 (tree_guess_outgoing_edge_probabilities): New.
25765 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25766 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25768 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25770 * ipa-split.c (split_function): Initialize return bb profile.
25772 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25774 * profile.c (compute_branch_probabilities): Also initialize
25775 EXIT_BLOCK profile.
25777 2017-06-06 Richard Biener <rguenther@suse.de>
25779 PR tree-optimization/80928
25780 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25781 (vect_analyze_loop_operations): Properly guard analysis for
25783 (vect_transform_loop): Likewise.
25784 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25785 (vect_model_induction_cost): Do not cost for pure SLP.
25786 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25787 of induction in inner loop vectorization.
25788 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25789 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25790 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25792 (vect_analyze_slp_cost_1): Cost induction.
25793 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25794 (vect_get_slp_vect_defs): Likewise.
25795 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25796 (vect_transform_stmt): Handle SLP reductions.
25797 * tree-vectorizer.h (vectorizable_induction): Adjust.
25799 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25801 * config/rs6000/rs6000.c (make_resolver_func): Update
25802 init_lowered_empty_function call.
25804 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25806 * doc/invoke.texi: Document the -fprofile-abs-path option.
25807 * common.opt (fprofile-abs-path): New option.
25808 * gcov-io.h (gcov_write_filename): Declare.
25809 * gcov-io.c (gcov_write_filename): New function.
25810 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25811 * profile.c (output_location): Likewise.
25813 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25815 * shring-wrap.c: Revert accidental commit.
25817 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25819 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25821 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25823 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25825 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25826 profile in callgraph edge.
25827 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25828 (apply_scale): Likewise.
25829 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25830 Also scale profile when inlining function with zero profile.
25831 (initialize_cfun): Update exit block profile even when it is zero.
25832 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25833 when profile is read.
25835 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25837 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25838 (CLONE_*): New constants to define the processors we can generate
25839 code for with the target_clone attribute.
25840 (rs6000_clone_map): New array to identify which clone processors
25841 the current program is running on.
25842 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25843 target_clone attribute.
25844 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25845 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25846 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25847 (cpu_expand_builtin): Add support for target_clone attribute.
25848 (rs6000_valid_attribute_p): Allow "default" attribute.
25849 (get_decl_name): New debug function to simplify printing the
25850 current function name in debugging statements.
25851 (rs6000_clone_priority): New functions to support the target_clone
25852 attribute, and be able to generate code to switch between ISA 2.05
25853 through ISA 3.0 (power6 through power9).
25854 (rs6000_compare_version_priority): Likewise.
25855 (rs6000_get_function_versions_dispatcher): Likewise.
25856 (make_resolver_func): Likewise.
25857 (add_condition_to_bb): Likewise.
25858 (dispatch_function_versions): Likewise.
25859 (rs6000_generate_version_dispatcher_body): Likewise.
25860 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25861 (fusion_gpr_load_p): Fix a spacing issue.
25862 * doc/extend.texi (Common Function Attributes): Document that the
25863 PowerPC supports the target_clone attribute.
25865 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25867 * config/arm/arm.h: explain F symbol found in description of ARM
25868 register allocation in its legend.
25870 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25872 * config/mips/frame-header-opt.c: Include profile-count.h.
25873 * config/riscv/riscv.c: Include profile-count.h
25875 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25877 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25879 (sm_set_flag_if_changed): Add bbs field.
25880 (execute_sm_if_changed_flag_set): Pass BBS.
25881 (execute_sm): Update.
25883 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25885 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25888 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25890 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25891 (peephole2): New peephole2 to emit the above.
25892 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25894 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25896 * config/aarch64/aarch64.c (define_peephole2 above
25897 *sub_<shift>_<mode>): New peephole.
25899 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25901 * config/i386/i386.c (make_resolver_func): Update.
25902 * Makefile.in: Add profile-count.h and profile-count.o
25903 * auto-profile.c (afdo_indirect_call): Update to new API.
25904 (afdo_set_bb_count): Update.
25905 (afdo_propagate_edge): Update.
25906 (afdo_propagate_circuit): Update.
25907 (afdo_calculate_branch_prob): Update.
25908 (afdo_annotate_cfg): Update.
25909 * basic-block.h: Include profile-count.h
25910 (struct edge_def): Turn count to profile_count.
25911 (struct basic_block_def): Likewie.
25912 (REG_BR_PROB_BASE): Move to profile-count.h
25913 (RDIV): Move to profile-count.h
25914 * bb-reorder.c (max_entry_count): Turn to profile_count.
25915 (find_traces): Update.
25916 (rotate_loop):Update.
25917 (connect_traces):Update.
25918 (sanitize_hot_paths):Update.
25919 * bt-load.c (migrate_btr_defs): Update.
25920 * cfg.c (RDIV): Remove.
25921 (init_flow): Use alloc_block.
25922 (alloc_block): Uninitialize count.
25923 (unchecked_make_edge): Uninitialize count.
25924 (check_bb_profile): Update.
25925 (dump_edge_info): Update.
25926 (dump_bb_info): Update.
25927 (update_bb_profile_for_threading): Update.
25928 (scale_bbs_frequencies_int): Update.
25929 (scale_bbs_frequencies_gcov_type): Update.
25930 (scale_bbs_frequencies_profile_count): New.
25931 * cfg.h (update_bb_profile_for_threading): Update.
25932 (scale_bbs_frequencies_profile_count): Declare.
25933 * cfgbuild.c (compute_outgoing_frequencies): Update.
25934 (find_many_sub_basic_blocks): Update.
25935 * cfgcleanup.c (try_forward_edges): Update.
25936 (try_crossjump_to_edge): Update.
25937 * cfgexpand.c (expand_gimple_tailcall): Update.
25938 (construct_exit_block): Update.
25939 * cfghooks.c (verify_flow_info): Update.
25940 (dump_bb_for_graph): Update.
25941 (split_edge): Update.
25942 (make_forwarder_block): Update.
25943 (duplicate_block): Update.
25944 (account_profile_record): Update.
25945 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25946 (get_estimated_loop_iterations): Update.
25947 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25948 (single_likely_exit): Update.
25949 * cfgloopmanip.c (scale_loop_profile): Update.
25951 (set_zero_probability): Update.
25952 (lv_adjust_loop_entry_edge): Update.
25953 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25954 (purge_dead_edges): Update.
25955 (rtl_account_profile_record): Update.
25956 * cgraph.c (cgraph_node::create): Uninitialize count.
25957 (symbol_table::create_edge): Uninitialize count.
25958 (cgraph_update_edges_for_call_stmt_node): Update.
25959 (cgraph_edge::dump_edge_flags): Update.
25960 (cgraph_node::dump): Update.
25961 (cgraph_edge::maybe_hot_p): Update.
25962 * cgraph.h: Include profile-count.h
25963 (create_clone), create_edge, create_indirect_edge): Update.
25964 (cgraph_node): Turn count to profile_count.
25965 (cgraph_edge0: Likewise.
25966 (make_speculative, clone): Update.
25967 (create_edge): Update.
25968 (init_lowered_empty_function): Update.
25969 * cgraphclones.c (cgraph_edge::clone): Update.
25970 (duplicate_thunk_for_node): Update.
25971 (cgraph_node::create_clone): Update.
25972 * cgraphunit.c (cgraph_node::analyze): Update.
25973 (cgraph_node::expand_thunk): Update.
25974 * final.c (dump_basic_block_info): Update.
25975 * gimple-streamer-in.c (input_bb): Update.
25976 * gimple-streamer-out.c (output_bb): Update.
25977 * graphite.c (print_global_statistics): Update.
25978 (print_graphite_scop_statistics): Update.
25979 * hsa-brig.c: Include basic-block.h.
25980 * hsa-dump.c: Include basic-block.h.
25981 * hsa-gen.c (T sum_slice): Update.
25982 (convert_switch_statements):Update.
25983 * hsa-regalloc.c: Include basic-block.h.
25984 * ipa-chkp.c (chkp_produce_thunks): Update.
25985 * ipa-cp.c (struct caller_statistics): Update.
25986 (init_caller_stats): Update.
25987 (gather_caller_stats): Update.
25988 (ipcp_cloning_candidate_p): Update.
25989 (good_cloning_opportunity_p): Update.
25990 (get_info_about_necessary_edges): Update.
25991 (dump_profile_updates): Update.
25992 (update_profiling_info): Update.
25993 (update_specialized_profile): Update.
25994 (perhaps_add_new_callers): Update.
25995 (decide_about_value): Update.
25996 (ipa_cp_c_finalize): Update.
25997 * ipa-devirt.c (struct odr_type_warn_count): Update.
25998 (struct decl_warn_count): Update.
25999 (struct final_warning_record): Update.
26000 (possible_polymorphic_call_targets): Update.
26001 (ipa_devirt): Update.
26002 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26003 * ipa-icf.c (sem_function::merge): Update.
26004 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26005 * ipa-inline.c (compute_uninlined_call_time): Update.
26006 (compute_inlined_call_time): Update.
26007 (want_inline_small_function_p): Update.
26008 (want_inline_self_recursive_call_p): Update.
26009 (edge_badness): Update.
26010 (lookup_recursive_calls): Update.
26011 (recursive_inlining): Update.
26012 (inline_small_functions): Update.
26013 (dump_overall_stats): Update.
26014 (dump_inline_stats): Update.
26015 * ipa-profile.c (ipa_profile_generate_summary): Update.
26016 (ipa_propagate_frequency): Update.
26017 (ipa_profile): Update.
26018 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26019 * ipa-utils.c (ipa_merge_profiles): Update.
26020 * loop-doloop.c (doloop_modify): Update.
26021 * loop-unroll.c (report_unroll): Update.
26022 (unroll_loop_runtime_iterations): Update.
26023 * lto-cgraph.c (lto_output_edge): Update.
26024 (lto_output_node): Update.
26025 (input_node): Update.
26026 (input_edge): Update.
26027 (merge_profile_summaries): Update.
26028 * lto-streamer-in.c (input_cfg): Update.
26029 * lto-streamer-out.c (output_cfg): Update.
26030 * mcf.c (create_fixup_graph): Update.
26031 (adjust_cfg_counts): Update.
26032 (sum_edge_counts): Update.
26033 * modulo-sched.c (sms_schedule): Update.
26034 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26035 * predict.c (maybe_hot_count_p): Update.
26036 (probably_never_executed): Update.
26037 (dump_prediction): Update.
26038 (combine_predictions_for_bb): Update.
26039 (propagate_freq): Update.
26040 (handle_missing_profiles): Update.
26041 (counts_to_freqs): Update.
26042 (rebuild_frequencies): Update.
26043 (force_edge_cold): Update.
26044 * predict.h: Include profile-count.h
26045 (maybe_hot_count_p, counts_to_freqs): UPdate.
26046 * print-rtl-function.c: Do not include cfg.h
26047 * print-rtl.c: Include basic-block.h
26048 * profile-count.c: New file.
26049 * profile-count.h: New file.
26050 * profile.c (is_edge_inconsistent): Update.
26051 (correct_negative_edge_counts): Update.
26052 (is_inconsistent): Update.
26053 (set_bb_counts): Update.
26054 (read_profile_edge_counts): Update.
26055 (compute_frequency_overlap): Update.
26056 (compute_branch_probabilities): Update; Initialize and deinitialize
26058 (branch_prob): Update.
26059 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26060 (edge_gcov_count): New.
26061 (bb_gcov_count): New.
26062 * shrink-wrap.c (try_shrink_wrapping): Update.
26063 * tracer.c (better_p): Update.
26064 * trans-mem.c (expand_transaction): Update.
26065 (ipa_tm_insert_irr_call): Update.
26066 (ipa_tm_insert_gettmclone_call): Update.
26067 * tree-call-cdce.c: Update.
26068 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26069 (gimple_duplicate_sese_tail): Update.
26070 (gimple_account_profile_record): Update.
26071 (execute_fixup_cfg): Update.
26072 * tree-inline.c (copy_bb): Update.
26073 (copy_edges_for_bb): Update.
26074 (initialize_cfun): Update.
26075 (freqs_to_counts): Update.
26076 (copy_cfg_body): Update.
26077 (expand_call_inline): Update.
26078 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26079 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26080 (try_unroll_loop_completely): Update.
26081 (try_peel_loop): Update.
26082 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26083 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26084 * tree-ssa-loop-split.c (connect_loops): Update.
26085 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26086 * tree-ssa-reassoc.c (branch_fixup): Update.
26087 * tree-ssa-tail-merge.c (replace_block_by): Update.
26088 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26089 (compute_path_counts): Update.
26090 (update_profile): Update.
26091 (recompute_probabilities): Update.
26092 (update_joiner_offpath_counts): Update.
26093 (estimated_freqs_path): Update.
26094 (freqs_to_counts_path): Update.
26095 (clear_counts_path): Update.
26096 (ssa_fix_duplicate_block_edges): Update.
26097 (duplicate_thread_path): Update.
26098 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26099 (struct switch_conv_info): Update.
26100 * tree-tailcall.c (decrease_profile): Update.
26101 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26102 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26103 * value-prof.c (check_counter): Update.
26104 (gimple_divmod_fixed_value): Update.
26105 (gimple_mod_pow2): Update.
26106 (gimple_mod_subtract): Update.
26107 (gimple_ic_transform): Update.
26108 (gimple_stringop_fixed_value): Update.
26109 * value-prof.h (gimple_ic): Update.
26111 2017-06-02 Carl Love <cel@us.ibm.com>
26113 * config/rs6000/rs6000-c: Add support for built-in functions
26114 vector double vec_doublee (vector signed int);
26115 vector double vec_doublee (vector unsigned int);
26116 vector double vec_doublee (vector float);
26117 vector double vec_doubleh (vector signed int);
26118 vector double vec_doubleh (vector unsigned int);
26119 vector double vec_doubleh (vector float);
26120 vector double vec_doublel (vector signed int);
26121 vector double vec_doublel (vector unsigned int);
26122 vector double vec_doublel (vector float);
26123 vector double vec_doubleo (vector signed int);
26124 vector double vec_doubleo (vector unsigned int);
26125 vector double vec_doubleo (vector float);.
26126 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26127 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26129 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26130 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26131 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26133 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26134 vec_doublel, vec_doubleh.
26135 * doc/extend.texi: Update the built-in documentation file for the
26136 new built-in functions.
26138 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26141 * ipa-inline-analysis.c (free_growth_caches): Set
26142 edge_removal_hook_holder to NULL after removing it.
26144 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26146 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26147 comparision with zero.
26149 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26150 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26151 for early expansion of vec_min and vec_max builtins.
26152 (builtin_function_type): Add min/max unsigned variants to those
26153 identified as having unsigned arguments.
26155 2017-06-02 Olivier Hainque <hainque@adacore.com>
26157 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26159 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26161 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26162 Use VALL_F16 iterator rather than VALL.
26164 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26166 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26167 Emit CBNZ inside loop when doing a strong exchange and comparing
26168 against zero. Generate the CC flags after the loop.
26170 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26172 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26173 (dl_section_ref): New.
26174 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26175 On AIX, append an expression to subtract the size of the
26176 section length to dl_section_ref.
26178 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26180 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26181 for early expansion of vector absolute builtins.
26183 2017-06-02 Richard Biener <rguenther@suse.de>
26185 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26186 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26188 2017-06-02 Richard Biener <rguenther@suse.de>
26190 PR tree-optimization/80948
26191 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26192 stmt order as well.
26194 2017-06-02 Richard Biener <rguenther@suse.de>
26196 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26198 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26199 for inductions as relevant.
26201 2017-06-02 Richard Biener <rguenther@suse.de>
26203 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26204 (vectorizable_induction): ... this. Remove dead code.
26206 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26208 * builtins. (expand_builtin_alloca): Remove second parameter and
26209 infer its value from the first parameter instead.
26210 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26212 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26214 PR rtl-optimization/80903
26215 * loop-doloop.c (add_test): Unshare sequence.
26217 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26219 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26221 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26223 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26225 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26226 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26227 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26228 (xlogue_layout::get_stub_rtx): Make static.
26229 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26230 (xlogue_layout::compute_stub_managed_regs): Rename to...
26231 (xlogue_layout::count_stub_managed_regs): ...this.
26232 (xlogue_layout::is_stub_managed_reg): New function.
26233 (xlogue_layout::m_stub_names): Rename to...
26234 (xlogue_layout::s_stub_names): ...this, make static.
26235 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26236 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26237 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26238 xlogue_layout::s_stub_names): Instantiate statics.
26239 (stub_managed_regs): Remove.
26240 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26241 (disable_call_ms2sysv_xlogues): Rename to...
26242 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26243 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26245 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26246 change after reload_completed.
26247 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26249 (ix86_expand_prologue): Likewise.
26250 (ix86_expand_epilogue): Likewise.
26251 (ix86_expand_split_stack_prologue): Likewise.
26252 (ix86_compute_frame_layout): Remove frame parameter ...
26253 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26254 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26256 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26257 (ix86_frame): Move from here ...
26258 * config/i386/i386.h (ix86_frame): ... to here.
26259 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26260 complete ix86_frame data structure instead. Remove some_ld_name.
26262 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26264 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26265 symbols that hold a DECL_VALUE_EXPR.
26267 2017-06-01 Martin Jambor <mjambor@suse.cz>
26269 PR tree-optimization/80898
26270 * tree-sra.c (process_subtree_disqualification): Removed.
26271 (disqualify_candidate): Do not acll
26272 process_subtree_disqualification.
26273 (subtree_mark_written_and_enqueue): New function.
26274 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26275 RHS has been disqualified and re-queue LHS if necessary. Apart
26276 from that, ignore disqualified RHS.
26278 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26280 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26281 address fetch for z10 or later.
26283 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26285 * config/arc/arc.md (tst_movb): Add guard when splitting.
26287 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26289 * config/arc/arc.c (arc_can_eliminate): Test against
26290 arc_frame_pointer_needed.
26292 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26294 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26295 to prevent store reordering.
26296 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26297 (type): Add block type.
26298 (stack_tie): Define special instruction to be used in
26301 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26303 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26304 constraint. It is not valid for the pattern.
26305 (noncommutative_binary_comparison): Likewise.
26307 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26309 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26312 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26314 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26315 be used by the reg-alloc.
26317 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26319 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26320 reg-alloc when having mul64 or mul32x16 instructions.
26321 (mulsidi3): Likewise.
26322 (umulsidi3): Likewise.
26323 (mulsi32x16): New pattern.
26324 (mulsi64): Likewise.
26325 (mulsidi64): Likewise.
26326 (umulsidi64): Likewise.
26327 (MUL32x16_REG): Define.
26328 (mul64_600): Use MUL32x16_REG.
26329 (mac64_600): Likewise.
26330 (umul64_600): Likewise.
26331 (umac64_600): Likewise.
26333 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26335 * config/arc/arc.md (mulsi3_700): Make it commutative.
26337 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26339 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26341 (*sign_extendsidi2_insn): Likewise for movstosw.
26343 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26345 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26346 the type of the input discriminant value. Convert the
26347 discriminant value of signedness vary.
26349 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26351 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26352 Add to -Wall section.
26354 2017-06-01 Richard Biener <rguenther@suse.de>
26356 PR middle-end/66313
26357 * fold-const.c (fold_plusminus_mult_expr): If the factored
26358 factor may be zero use a wrapping type for the inner operation.
26359 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26360 and handle moved defs.
26361 (process_assignment): Properly guard the unary op case. Return a
26362 tri-state indicating that moving the stmt before the call may allow
26363 to continue. Pass through to_move.
26364 (find_tail_calls): Handle moving unrelated defs before
26367 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26370 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26371 splitter result in the canonical way.
26373 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26375 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26376 also for 32bit target. Update insn attributes.
26377 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26379 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26381 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26382 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26383 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26384 (_mm_maskz_min_ss): New intrinsics.
26386 2017-05-31 Martin Liska <mliska@suse.cz>
26388 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26389 Change comment style to one we normally use.
26390 (vectorizable_reduction): Likewise.
26391 (vectorizable_induction): Likewise.
26392 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26393 (vectorizable_call): Likewise.
26394 (vectorizable_simd_clone_call): Likewise.
26395 (vectorizable_conversion): Likewise.
26396 (vectorizable_assignment): Likewise.
26397 (vectorizable_shift): Likewise.
26398 (vectorizable_operation): Likewise.
26399 (vectorizable_store): Likewise.
26400 (vectorizable_load): Likewise.
26401 * tree-vectorizer.h: Likewise.
26403 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26405 * passes.c (emergency_dump_function): New.
26406 * tree-pass.h (emergency_dump_function): Declare.
26407 * plugin.c (plugins_internal_error_function): Remove.
26408 * plugin.h (plugins_internal_error_function): Remove declaration.
26409 * toplev.c (internal_error_function): New static function. Use it...
26410 (general_init): ...here.
26412 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26414 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26415 (arc_rtx_costs): Add costs for new patterns.
26416 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26417 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26419 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26421 * tree-ssa-strlen.c (get_next_strinfo): New function.
26422 (get_stridx_plus_constant): Use it.
26423 (zero_length_string): Likewise.
26424 (adjust_related_strinfos): Likewise.
26425 (adjust_last_stmt): Likewise.
26427 2017-05-31 Richard Biener <rguenther@suse.de>
26430 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26431 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26433 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26435 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26436 loop_vinfo argument and use of dependence distance vectors.
26437 Check instead whether the two references differ only in their
26438 initial value and assume that they have the same alignment if the
26439 difference is a multiple of the vector alignment.
26440 (vect_analyze_data_refs_alignment): Update call accordingly.
26442 2017-05-31 Martin Liska <mliska@suse.cz>
26445 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26447 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26449 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26450 (create_intersect_range_checks): Move from ...
26451 * tree-data-ref.c (create_intersect_range_checks_index)
26452 (create_intersect_range_checks): ... to here.
26453 (create_runtime_alias_checks): New function factored from ...
26454 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26455 here. Call above function.
26456 * tree-data-ref.h (create_runtime_alias_checks): New function.
26458 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26460 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26461 segment length for dr_b and compute it in wide_int.
26463 2017-05-31 Richard Biener <rguenther@suse.de>
26465 PR tree-optimization/80906
26466 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26467 and pass through iv_map.
26468 (copy_bb_and_scalar_dependences): Adjust.
26469 (translate_pending_phi_nodes): Likewise.
26470 (copy_loop_close_phi_args): Handle code-generating IVs instead
26473 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26475 * diagnostic-color.c (color_dict): Add "type-diff".
26476 (parse_gcc_colors): Update comment.
26477 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26478 -fdiagnostics-show-template-tree and -fno-elide-type.
26479 (GCC_COLORS): Add type-diff to example.
26481 (-fdiagnostics-show-template-tree): New.
26482 (-fno-elide-type): New.
26483 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26484 the pp_format_decoder callback. Call any m_format_postprocessor's
26486 (pretty_printer::pretty_printer): Initialize
26487 m_format_postprocessor.
26488 (pretty_printer::~pretty_printer): Delete any
26489 m_format_postprocessor.
26490 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26491 (class format_postprocessor): New class.
26492 (struct pretty_printer::format_decoder): Document the new parameters.
26493 (struct pretty_printer::m_format_postprocessor): New field.
26494 * tree-diagnostic.c (default_tree_printer): Update for new
26495 bool and const char ** params.
26496 * tree-diagnostic.h (default_tree_printer): Likewise.
26498 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26500 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26501 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26502 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26503 rs6000_gen_cell_microcode code.
26504 (rs6000_final_prescan_insn): Delete.
26505 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26506 "warn-cell-microcode" entries.
26507 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26508 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26509 throughout. Change cc_reg_not_micro_cr0_operand to
26510 cc_reg_not_cr0_operand throughout.
26511 (*extendhi<mode>2_noload): Delete.
26512 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26513 (mwarn-cell-microcode): Delete.
26514 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26515 -mgen-cell-microcode and -mwarn-cell-microcode.
26517 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26520 * config/i386/constraints.md (Yd): New constraint.
26522 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26523 and (?Yd, r) alternatives. Update insn attributes.
26524 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26525 and (?*Yd, r) alternatives. Update insn attributes.
26526 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26528 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26530 * gimplify.c (gimplify_modify_expr): Don't create a
26531 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26534 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26536 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26538 2017-05-30 Richard Biener <rguenther@suse.de>
26540 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26541 and reduc_def fields.
26542 (STMT_VINFO_REDUC_TYPE): New define.
26543 (STMT_VINFO_REDUC_DEF): Likewise.
26544 (vect_force_simple_reduction): Adjust prototype.
26545 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26546 (vect_is_simple_reduction): Remove check_reduction argument.
26547 (vect_force_simple_reduction): Adjust and set
26548 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26549 (vectorizable_reduction): Do not re-do reduction analysis
26550 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26551 * tree-parloops.c (gather_scalar_reductions): Adjust.
26553 2017-05-30 Richard Biener <rguenther@suse.de>
26555 PR middle-end/80901
26556 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26559 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26561 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26562 Introduce unknown_misalignment parameter and remove vf.
26563 (vect_peeling_hash_get_lowest_cost):
26564 Pass unknown_misalignment parameter.
26565 (vect_enhance_data_refs_alignment):
26566 Fix unsupportable data ref treatment.
26568 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26570 * tree-vect-data-refs.c (vect_get_data_access_cost):
26571 Workaround for SLP handling.
26572 (vect_enhance_data_refs_alignment):
26573 Compute costs for doing no peeling at all, compare to the best
26574 peeling costs so far and avoid peeling if cheaper.
26576 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26578 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26579 Return peeling info and set costs to zero for unlimited cost
26581 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26582 with unknown misalignment. Compute and costs for unknown
26583 misalignment, compare them to the costs for known misalignment
26584 and choose the cheapest for peeling.
26586 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26588 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26589 (vect_get_peeling_costs_all_drs): Create function.
26590 (vect_peeling_hash_get_lowest_cost):
26591 Use vect_get_peeling_costs_all_drs.
26592 (vect_peeling_supportable): Create function.
26593 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26595 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26597 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26598 DR_HAS_NEGATIVE_STEP.
26599 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26600 (vect_enhance_data_refs_alignment): Use.
26601 (vect_duplicate_ssa_name_ptr_info): Use.
26602 * tree-vectorizer.h (dr_misalignment): Use.
26603 (known_alignment_for_access_p): Use.
26605 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26608 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26610 (has_section_name): New function.
26612 2017-05-30 Martin Liska <mliska@suse.cz>
26615 * auto-profile.c (get_function_decl_from_block): Fix
26618 2017-05-30 Richard Biener <rguenther@suse.de>
26620 PR middle-end/80876
26621 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26623 2017-05-30 Martin Liska <mliska@suse.cz>
26625 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26626 * dumpfile.h (struct dump_file_info): Remove ctors.
26628 2017-05-30 Martin Liska <mliska@suse.cz>
26630 * predict.def: Fix GNU coding style.
26632 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26634 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26635 Mark 'to' argument with ATTRIBUTE_UNUSED.
26637 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26639 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26640 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26641 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26644 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26646 * doc/install.texi (Options specification): Restore entry of
26647 --enable-sjlj-exceptions.
26649 2017-05-27 Michael Eager <eager@eagercon.com>
26652 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26654 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26656 * config/microblaze/microblaze.h
26657 (FIXED_REGISTERS): Update in macro.
26658 (CALL_USED_REGISTERS): Update in macro.
26660 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26662 * doc/install.texi: Add links to macOS binary distributions.
26664 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26668 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26670 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26672 2017-05-26 Martin Liska <mliska@suse.cz>
26674 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26676 2017-05-26 Martin Liska <mliska@suse.cz>
26678 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26679 always leading ';; '.
26680 (dump_bb_info): Likewise.
26681 (brief_dump_cfg): Likewise.
26682 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26683 * dumpfile.c: Remove usage of TDF_VERBOSE.
26684 * dumpfile.h (enum dump_kind): Likewise.
26685 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26686 * print-tree.c (debug_verbose): Remove.
26687 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26688 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26689 * tree-diagnostic.c (default_tree_printer): Replace
26690 TDF_DIAGNOSTIC with TDF_SLIM.
26692 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26694 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26695 in parameter loop, rather than loop_vinfo.
26696 (create_intersect_range_checks): Ditto.
26697 (vect_create_cond_for_alias_checks): Update call to above functions.
26699 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26701 PR tree-optimization/80815
26702 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26703 for merging runtime alias checks. Handle negative DR_STEPs.
26705 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26707 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26709 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26710 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26711 out code pruning runtime alias checks.
26712 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26713 factored out from above.
26714 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26716 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26718 (prune_runtime_alias_test_list): New decalaration.
26720 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26722 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26723 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26724 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26725 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26726 (operator==, comp_dr_with_seg_len_pair): Ditto.
26727 (vect_prune_runtime_alias_test_list): Ditto.
26729 2017-05-26 Martin Liska <mliska@suse.cz>
26732 * params.def: Bound partial-inlining-entry-probability param.
26734 2017-05-26 Marek Polacek <polacek@redhat.com>
26737 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26740 2017-05-26 Richard Biener <rguenther@suse.de>
26742 PR tree-optimization/80842
26743 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26746 2017-05-26 Richard Biener <rguenther@suse.de>
26748 PR tree-optimization/80844
26749 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26751 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26753 * doc/md.texi (Machine Constraints): Update x86 family
26754 machine constraints section to match 'config/i386/constraints.md'.
26756 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26758 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26760 2017-05-25 Nathan Sidwell <nathan@acm.org>
26762 * doc/invoke.texi (--enable-languages): Update documentation.
26764 2017-05-25 Martin Liska <mliska@suse.cz>
26766 * dumpfile.c: Add TDF_FOLDING.
26767 * dumpfile.h (enum dump_kind): Likewise.
26768 * genmatch.c (dt_simplify::gen_1): Use it.
26770 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26772 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26774 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26776 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26777 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26779 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26781 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26782 * match.pd (X == C): Rewrite it here.
26783 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26784 with_certain_nonzero_bits2): New predicates.
26785 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26787 2017-05-24 Nathan Sidwell <nathan@acm.org>
26789 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26792 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26795 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26797 * config/powerpcspe: New port. Files are copied from the rs6000
26798 port, with "rs6000" in filenames replaced by "powerpcspe".
26800 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26802 PR rtl-optimization/80754
26803 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26805 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26807 * config/sparc/sparc.md (length): Return the correct value for -mflat
26808 sibcalls to match output_sibcall.
26810 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26814 * config/rs6000/rs6000.c (struct machine_function): Add new field
26816 (rs6000_get_separate_components): Init that field, use it.
26817 (rs6000_components_for_bb): Use the field.
26819 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26821 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26823 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26825 PR middle-end/80823
26826 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26828 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26831 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26832 address against address_operand predicate.
26833 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26835 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26837 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26838 parameters passed indirectly.
26840 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26842 * config/i386/i386.md (*movdi_internal): Remove SSE4
26843 alternative 18 (?r, *v). Update insn attributes.
26844 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26845 Update insn attributes.
26846 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26847 Update insn attributes.
26848 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26849 alternative 1 (r, v). Remove isa attribute.
26850 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26851 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26852 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26854 2017-05-23 Tom de Vries <tom@codesourcery.com>
26856 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26859 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26861 * cgraphunit.c (symbol_table::process_new_functions): Update.
26862 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26863 (inline_generate_summary): Rename to ...
26864 (ipa_fn_summary_generate): ... this one.
26865 (inline_read_summary): Rename to ...
26866 (ipa_fn_summary_read): ... this one.
26867 (inline_write_summary): Rename to ...
26868 (ipa_fn_summary_write): ... this one.
26869 (inline_free_summary): Rename to ...
26870 (ipa_free_fn_summary): ... this one.
26871 (pass_data_local_fn_summary, pass_local_fn_summary,
26872 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26873 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26874 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26875 make_pass_ipa_fn_summary): New.
26876 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26877 inline_write_summary, inline_free_summary): Remove.
26878 (ipa_free_fn_summary) : New.
26879 * ipa-inline.c (ipa_inline): Update.
26880 (pass_ipa_inline): Do not generate summaries.
26881 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26883 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26884 and add pass_ipa_fn_summary.
26885 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26887 (make_pass_inline_parameters): Remove.
26889 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26891 * omp-low.c (struct omp_context): Remove "default_kind" member.
26894 * omp-offload.c (execute_oacc_device_lower): Remove the
26895 parallelism dimensions function attributes for unparallelized
26896 OpenACC kernels constructs.
26898 2017-05-23 Martin Liska <mliska@suse.cz>
26900 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26902 (cgraph_edge::make_speculative): Likewise.
26903 (cgraph_edge::resolve_speculation): Likewise.
26904 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26905 (cgraph_node::dump): Likewise.
26906 * cgraph.h: Likewise.
26907 * cgraphunit.c (analyze_functions): Likewise.
26908 (symbol_table::compile): Likewise.
26909 * ipa-cp.c (print_all_lattices): Likewise.
26910 (determine_versionability): Likewise.
26911 (initialize_node_lattices): Likewise.
26912 (ipcp_verify_propagated_values): Likewise.
26913 (estimate_local_effects): Likewise.
26914 (update_profiling_info): Likewise.
26915 (create_specialized_node): Likewise.
26916 (perhaps_add_new_callers): Likewise.
26917 (decide_about_value): Likewise.
26918 (decide_whether_version_node): Likewise.
26919 (identify_dead_nodes): Likewise.
26920 (ipcp_store_bits_results): Likewise.
26921 * ipa-devirt.c (dump_targets): Likewise.
26922 (ipa_devirt): Likewise.
26923 * ipa-icf.c (sem_item::dump): Likewise.
26924 (sem_function::equals): Likewise.
26925 (sem_variable::equals): Likewise.
26926 (sem_item_optimizer::read_section): Likewise.
26927 (sem_item_optimizer::execute): Likewise.
26928 (congruence_class::dump): Likewise.
26929 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26930 (dump_inline_summary): Likewise.
26931 (estimate_node_size_and_time): Likewise.
26932 (inline_analyze_function): Likewise.
26933 * ipa-inline-transform.c (inline_call): Likewise.
26934 * ipa-inline.c (report_inline_failed_reason): Likewise.
26935 (want_early_inline_function_p): Likewise.
26936 (edge_badness): Likewise.
26937 (update_edge_key): Likewise.
26938 (inline_small_functions): Likewise.
26939 * ipa-profile.c (ipa_profile): Likewise.
26940 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26941 (ipa_make_edge_direct_to_target): Likewise.
26942 (remove_described_reference): Likewise.
26943 (ipa_impossible_devirt_target): Likewise.
26944 (propagate_controlled_uses): Likewise.
26945 (ipa_print_node_params): Likewise.
26946 (ipcp_transform_function): Likewise.
26947 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26948 (propagate_pure_const): Likewise.
26949 * ipa-reference.c (generate_summary): Likewise.
26950 (read_write_all_from_decl): Likewise.
26951 (propagate): Likewise.
26952 (ipa_reference_read_optimization_summary): Likewise.
26953 * ipa-utils.c (ipa_merge_profiles): Likewise.
26954 * ipa.c (walk_polymorphic_call_targets): Likewise.
26955 (symbol_table::remove_unreachable_nodes): Likewise.
26956 (ipa_single_use): Likewise.
26957 * passes.c (execute_todo): Likewise.
26958 * predict.c (drop_profile): Likewise.
26959 * symtab.c (symtab_node::get_dump_name): New function.
26960 (symtab_node::dump_name): Likewise.
26961 (symtab_node::dump_asm_name): Likewise.
26962 (symtab_node::dump_references): Likewise.
26963 (symtab_node::dump_referring): Likewise.
26964 (symtab_node::dump_base): Likewise.
26965 (symtab_node::debug_symtab): Likewise.
26966 * tree-sra.c (convert_callers_for_node): Likewise.
26967 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26968 * value-prof.c (init_node_map): Likewise.
26970 2017-05-23 Martin Liska <mliska@suse.cz>
26972 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26973 and symtab_node::debug_symtab to symbol_table::debug.
26974 * cgraphunit.c (analyze_functions): Use the renamed function.
26975 (symbol_table::compile): Likewise.
26976 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26977 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26978 * passes.c (execute_todo): Likewise.
26979 * symtab.c (symbol_table::dump): New function.
26980 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26982 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26984 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26985 that nonconst implies exec.
26987 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26989 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26990 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26991 (inline_edge_summary_vec): Turn into ...
26992 (ipa_call_summaries): ... this one.
26993 (redirect_to_unreachable, edge_set_predicate,
26994 evaluate_properties_for_edge, inline_summary_alloc,
26995 reset_ipa_call_summary, reset_inline_summary,
26996 inline_summary_t::duplicate): Update.
26997 (inline_edge_duplication_hook): Turn to ...
26998 (ipa_call_summary_t::duplicate): ... this one.
26999 (inline_edge_removal_hook): Turn to ...
27000 (ipa_call_summary_t::remove): ... this one.
27001 (dump_inline_edge_summary): Turn to ...
27002 (dump_ipa_call_summary): ... this one.
27003 (estimate_function_body_sizes): Update.
27004 (inline_update_callee_summaries): Update.
27005 (remap_edge_change_prob): Update.
27006 (remap_edge_summaries): Update.
27007 (inline_merge_summary): Update.
27008 (do_estimate_edge_time): Update.
27009 (inline_generate_summary): Update.
27010 (inline_read_section): Update.
27011 (inline_read_summary): Update.
27012 (inline_free_summary): Update.
27013 * ipa-inline.c (can_inline_edge_p): Update.
27014 (compute_inlined_call_time): Update.
27015 (want_inline_small_function_p): Update.
27016 (edge_badness): Update.
27017 (early_inliner): Update.
27018 * ipa-inline.h (inline_edge_summary): Turn to ...
27019 (ipa_call_summary): ... this one.
27020 (ipa_call_summary_t): New class.
27021 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27022 (ipa_call_summaries): New.
27023 (inline_edge_summary): Remove.
27024 (estimate_edge_growth): Update.
27025 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27026 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27027 * ipa-split.c (execute_split_functions): Update.
27028 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27030 2017-05-23 Tom de Vries <tom@codesourcery.com>
27032 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27033 attributes): Document rdrand effective target.
27035 2017-05-23 Tom de Vries <tom@codesourcery.com>
27037 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27038 attributes): Sort alphabetically.
27040 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27042 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27044 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27047 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27048 V2DF/V2DI splat into two separate patterns, one that handles
27049 registers, and the other that only handles memory. Drop support
27050 for splatting from a GPR on ISA 2.07 and then splitting the
27051 splat into direct move and splat.
27052 (vsx_splat_<mode>_reg): Likewise.
27053 (vsx_splat_<mode>_mem): Likewise.
27055 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27057 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27059 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27061 PR middle-end/80809
27062 * omp-low.c (finish_taskreg_remap): New function.
27063 (finish_taskreg_scan): If unit size of ctx->record_type
27064 is non-constant, unshare the size expression and replace
27065 decls in it with possible outer var refs.
27067 PR middle-end/80809
27068 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27069 GOVD_SHARED rather than GOVD_PRIVATE with it.
27070 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27071 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27073 PR middle-end/80853
27074 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27075 as last argument to build_outer_var_ref for pointer bases of array
27076 section reductions.
27078 2017-05-19 Martin Sebor <msebor@redhat.com>
27080 * print-tree.c (print_node): Print DECL_READ_P flag.
27082 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27084 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27085 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27086 * cgraph.c: Likewise.
27087 * cgraphunit.c: Likewise.
27088 * gengtype.c: Likewise.
27089 * ipa-cp.c: Likewise.
27090 * ipa-devirt.c: Likewise.
27091 * ipa-icf.c: Likewise.
27092 * ipa-predicate.c: Likewise.
27093 * ipa-profile.c: Likewise.
27094 * ipa-prop.c: Likewise.
27095 * ipa-split.c: Likewise.
27097 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27098 edge_predicate_pool, dump_inline_hints,
27099 inline_summary::account_size_time, redirect_to_unreachable,
27100 edge_set_predicate, set_hint_predicate,
27101 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27102 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27103 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27104 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27105 ipa_call_summary_t::remove, initialize_growth_caches,
27106 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27107 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27108 mark_modified, unmodified_parm_1, unmodified_parm,
27109 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27110 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27111 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27112 will_be_nonconstant_predicate, record_modified_bb_info,
27113 get_minimal_bb, record_modified, param_change_prob,
27114 phi_result_unknown_predicate, predicate_for_phi_result,
27115 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27116 estimate_function_body_sizes, compute_inline_parameters,
27117 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27118 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27119 inline_update_callee_summaries, remap_edge_change_prob,
27120 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27121 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27122 inline_analyze_function, inline_summary_t::insert,
27123 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27124 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27125 inline_free_summary): Move to ipa-fnsummary.h
27126 (predicate_t): Remove.
27127 * ipa-fnsummary.c: New file.
27128 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27129 (enum inline_hints_vals, inline_hints, agg_position_info,
27130 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27131 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27132 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27133 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27134 inline_read_summary, inline_write_summary, inline_free_summary,
27135 inline_analyze_function, initialize_inline_failed,
27136 inline_merge_summary, inline_update_overall_summary,
27137 compute_inline_parameters): Move to ipa-fnsummary.h
27138 * ipa-fnsummary.h: New file.
27139 * ipa-inline-transform.h: Include ipa-inline.h.
27140 * ipa-inline.c: LIkewise.
27142 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27144 * ipa-inline.c (edge_badness): Use inlined_time instead of
27145 inline_summaries->get.
27147 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27149 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27151 2017-05-22 Nathan Sidwell <nathan@acm.org>
27153 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27154 (fdump-lang): Document 'raw' option.
27155 * dumpfile.h (TDI_tu): Delete.
27156 * dumpfile.c (dump_files): Remove translation-unit.
27157 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27159 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27161 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27162 command option from $(AWK) call.
27163 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27164 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27165 [FORMAT]: Remove handling of variable.
27166 * config/avr/t-multilib: Regenerate.
27168 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27170 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27172 (dump_inline_summary): Do not print self_time.
27173 (estimate_function_body_sizes): Do not set self_time.
27174 (compute_inline_parameters): Likewise.
27175 (inline_read_section, inline_write_summary): Do not stream self_time.
27176 * ipa-inline.h (inline_summary): Drop self_time.
27178 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27180 * ipa-inline-analysis.c (account_size_time): Rename to ...
27181 (inline_summary::account_size_time): ... this one.
27182 (reset_ipa_call_summary): Turn to ...
27183 (ipa_call_summary::reset): ... this one.
27184 (reset_inline_summary): Turn to ...
27185 (inline_summary::reset): ... this one.
27186 (inline_summary_t::remove): Update.
27187 (inline_summary_t::duplicate): Update.
27188 (ipa_call_summary_t::remove): Update.
27189 (dump_inline_summary): Update.
27190 (estimate_function_body_sizes): Update.
27191 (compute_inline_parameters): Update.
27192 (estimate_node_size_and_time): Update.
27193 (inline_merge_summary): Update.
27194 (inline_update_overall_summary): Update.
27195 (inline_read_section): Update.
27196 (inline_write_summary): Update.
27197 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27198 add account_size_time and reset member functions.
27199 (ipa_call_summary): Add reset function.
27200 * ipa-predicate.h (predicate::operator &): Constify.
27202 2017-05-22 Richard Biener <rguenther@suse.de>
27204 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27206 2017-05-19 Jason Merrill <jason@redhat.com>
27208 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27210 2017-05-19 Marek Polacek <polacek@redhat.com>
27213 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27214 TYPE_OVERFLOW_WRAPS checks.
27216 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27218 * tree-core.h (enum omp_clause_default_kind): Add
27219 "OMP_CLAUSE_DEFAULT_PRESENT".
27220 * tree-pretty-print.c (dump_omp_clause): Handle it.
27221 * gimplify.c (enum gimplify_omp_var_data): Add
27222 "GOVD_MAP_FORCE_PRESENT".
27223 (gimplify_adjust_omp_clauses_1): Map it to
27224 "GOMP_MAP_FORCE_PRESENT".
27225 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27227 * gimplify.c (oacc_default_clause): Clarify.
27229 2017-05-19 Nathan Sidwell <nathan@acm.org>
27231 LANG_HOOK_REGISTER_DUMPS
27232 * toplev.c (general_init): Call register dump lang hook.
27233 * doc/invoke.texi: Document -fdump-lang option family.
27234 * dumpfile.c (dump_files): Remove class dump here.
27235 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27236 * dumpfile.h (tree_dump_index): Remove TDI_class.
27237 * langhooks-def.h (lhd_register_dumps): Declare.
27238 (LANG_HOOKS_REGISTER_DUMPS): Define.
27239 (LANG_HOOKS_INITIALIZER): Add it.
27240 * langhooks.c (lhd_register_dumps): Define.
27241 * langhooks.h (struct lang_hooks): Add register_dumps.
27243 2017-05-19 Nathan Sidwell <nathan@acm.org>
27245 * context.h (context::set_passes): New.
27246 * context.c (context::context): Do not create pass manager.
27247 * toplev.c (general_init): Create pass manager here.
27249 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27251 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27252 use this splitter if two add or or instructions would also work for
27253 the constant we want to generate.
27255 2017-05-19 Richard Biener <rguenther@suse.de>
27258 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27259 predicate evaluation.
27261 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27263 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27265 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27268 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27270 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27271 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27272 (inline_edge_summary_vec): Turn into ...
27273 (ipa_call_summaries): ... this one.
27274 (redirect_to_unreachable, edge_set_predicate,
27275 evaluate_properties_for_edge, inline_summary_alloc,
27276 reset_ipa_call_summary, reset_inline_summary,
27277 inline_summary_t::duplicate): Update.
27278 (inline_edge_duplication_hook): Turn to ...
27279 (ipa_call_summary_t::duplicate): ... this one.
27280 (inline_edge_removal_hook): Turn to ...
27281 (ipa_call_summary_t::remove): ... this one.
27282 (dump_inline_edge_summary): Turn to ...
27283 (dump_ipa_call_summary): ... this one.
27284 (estimate_function_body_sizes): Update.
27285 (inline_update_callee_summaries): Update.
27286 (remap_edge_change_prob): Update.
27287 (remap_edge_summaries): Update.
27288 (inline_merge_summary): Update.
27289 (do_estimate_edge_time): Update.
27290 (inline_generate_summary): Update.
27291 (inline_read_section): Update.
27292 (inline_read_summary): Update.
27293 (inline_free_summary): Update.
27294 * ipa-inline.c (can_inline_edge_p): Update.
27295 (compute_inlined_call_time): Update.
27296 (want_inline_small_function_p): Update.
27297 (edge_badness): Update.
27298 (early_inliner): Update.
27299 * ipa-inline.h (inline_edge_summary): Turn to ...
27300 (ipa_call_summary): ... this one.
27301 (ipa_call_summary_t): New class.
27302 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27303 (ipa_call_summaries): New.
27304 (inline_edge_summary): Remove.
27305 (estimate_edge_growth): Update.
27306 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27307 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27308 * ipa-split.c (execute_split_functions): Update.
27309 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27311 2017-05-19 Richard Biener <rguenther@suse.de>
27313 PR middle-end/80764
27314 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27316 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27318 * config/rs6000/rs6000.c (struct machine_function): Add field
27319 fpr_is_wrapped_separately.
27320 (rs6000_get_separate_components): Use 64 components. Handle the
27321 new FPR components.
27322 (rs6000_components_for_bb): Handle the FPR components.
27323 (rs6000_emit_prologue_components): Handle the FPR components.
27324 (rs6000_emit_epilogue_components): Handle the FPR components.
27325 (rs6000_set_handled_components): Handle the FPR components.
27326 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27327 that are already separately shrink-wrapped.
27328 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27329 that are already separately shrink-wrapped.
27331 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27334 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27337 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27338 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27339 cases where the register allocator uses a move and an offsettable
27340 memory operation to/from a FPR register on ISA 2.06/2.07.
27341 (define_peephole2 for Altivec d-form store): Likewise.
27343 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27346 * config/i386/mmx.md (*mov<mode>_internal): Enable
27347 alternatives 11, 12, 13 and 14 also for 32bit targets.
27348 Remove alternatives 15, 16, 17 and 18.
27349 * config/i386/sse.md (vec_concatv2di): Change
27350 alternative (!x, *y) to (x, ?!*Yn).
27352 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27354 * dumpfile.h (enum dump_kind): Remove stray comma.
27356 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27358 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27359 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27360 predicate::num_conditions
27361 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27362 (CHANGED): turn into predicate::changed.
27363 (agg_position_info): Move to ipa-predicate.h
27364 (add_condition, predicate::add_clause, predicate::operator &=,
27365 predicate::or_with, predicate::evaluate, predicate::probability,
27366 dump_condition, dump_clause, predicate::dump,
27367 predicate::remap_after_duplication, predicate::remap_after_inlining,
27368 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27369 (evaluate_conditions_for_known_args): Update.
27370 (set_cond_stmt_execution_predicate): Update.
27371 * ipa-inline.h: Include ipa-predicate.h
27372 (condition, inline_param_summary, conditions, agg_position_info,
27373 predicate): Move to ipa-predicate.h
27374 * ipa-predicate.c: New file.
27375 * ipa-predicate.h: New file.
27377 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27379 * final.c (leaf_function_p): Check we are not in a sequence.
27381 2017-05-18 Martin Liska <mliska@suse.cz>
27383 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27384 * dumpfile.c (dump_register): Use new enum dump_kind.
27385 (get_dump_file_name): Likewise.
27386 (dump_enable_all): Likewise.
27387 (dump_switch_p_1): Likewise.
27388 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27389 * dumpfile.h (enum dump_kind): New enum type.
27390 (struct dump_file_info): Create constructor and
27391 format fields and comments.
27392 * passes.c (pass_manager::register_one_dump_file):
27394 * statistics.c (statistics_early_init): Likewise.
27395 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27396 TDF_TREE with TDF_SLIM.
27397 (gather_memory_references_ref): Likewise.
27399 2017-05-18 Martin Liska <mliska@suse.cz>
27401 * vec.h (struct vnull): Use it.
27403 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27405 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27406 (true_predicate, false_predicate, true_predicate_p,
27407 false_predicate_p): Remove.
27408 (single_cond_predicate, not_inlined_predicate): Turn to member function
27410 (add_condition): Update.
27411 (add_clause): Turn to...
27412 (predicate::add_clause): ... this one; update; allow passing NULL
27414 (and_predicates): Turn to ...
27415 (predicate::operator &=): ... this one.
27416 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27417 (or_predicates): Turn to ...
27418 (predicate::or_with): ... this one.
27419 (evaluate_predicate): Turn to ...
27420 (predicate::evaluate): ... this one.
27421 (predicate_probability): Turn to ...
27422 (predicate::probability): ... this one.
27423 (dump_condition): Update.
27424 (dump_predicate): Turn to ...
27425 (predicate::dump): ... this one.
27426 (account_size_time): Update.
27427 (edge_set_predicate): Update.
27428 (set_hint_predicate): UPdate.
27429 (evaluate_conditions_for_known_args): Update.
27430 (evaluate_properties_for_edge): Update.
27431 (remap_predicate_after_duplication): Turn to...
27432 (predicate::remap_after_duplication): ... this one.
27433 (remap_hint_predicate_after_duplication): Update.
27434 (inline_summary_t::duplicate): UPdate.
27435 (dump_inline_edge_summary): Update.
27436 (dump_inline_summary): Update.
27437 (set_cond_stmt_execution_predicate): Update.
27438 (set_switch_stmt_execution_predicate): Update.
27439 (compute_bb_predicates): Update.
27440 (will_be_nonconstant_expr_predicate): Update.
27441 (will_be_nonconstant_predicate): Update.
27442 (phi_result_unknown_predicate): Update.
27443 (predicate_for_phi_result): Update.
27444 (array_index_predicate): Update.
27445 (estimate_function_body_sizes): Update.
27446 (estimate_node_size_and_time): Update.
27447 (estimate_ipcp_clone_size_and_time): Update.
27448 (remap_predicate): Rename to ...
27449 (predicate::remap_after_inlining): ... this one.
27450 (remap_hint_predicate): Update.
27451 (inline_merge_summary): Update.
27452 (inline_update_overall_summary): Update.
27453 (estimate_size_after_inlining): Update.
27454 (read_predicate): Rename to ...
27455 (predicate::stream_in): ... this one.
27456 (read_inline_edge_summary): Update.
27457 (write_predicate): Rename to ...
27458 (predicate::stream_out): ... this one.
27459 (write_inline_edge_summary): Update.
27460 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27461 (clause_t): Turn to uint32_t
27462 (predicate): Turn to class; implement constructor and operators
27464 (size_time_entry): Update.
27465 (inline_summary): Update.
27466 (inline_edge_summary): Update.
27468 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27470 * fold-const.c (fold_binary_loc): Move transformation...
27471 * match.pd (C - X CMP X): ... here.
27473 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27475 * config/sparc/sparc.c (sparc_option_override): Set function
27476 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27477 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27479 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27481 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27483 2017-05-18 Marek Polacek <polacek@redhat.com>
27486 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27487 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27488 gimple_assign_load_p.
27490 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27492 PR middle-end/80692
27493 * real.c (do_compare): Give decimal_do_compare preference over
27494 comparing just the signs.
27496 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27498 * doc/md.texi (Canonicalization of Instructions): Describe the
27499 canonical form of instructions that inherently set a condition
27502 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27504 PR middle-end/80775
27505 * tree-cfg.c: Move deletion of unreachable case statements to after
27506 the merging of consecutive case labels.
27508 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27510 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27511 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27512 restoring of callee-saved registers.
27514 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27516 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27517 * config/visium/visium.c (single_set_and_flags): Likewise.
27518 * config/visium/visium.md (Substitutions): Likewise.
27520 2017-05-17 Martin Liska <mliska@suse.cz>
27522 * cfg.c: Introduce dump_flags_t type and
27523 use it instead of int type.
27525 * cfghooks.c: Likewise.
27526 * cfghooks.h (struct cfg_hooks): Likewise.
27527 * cfgrtl.c: Likewise.
27528 * cfgrtl.h: Likewise.
27529 * cgraph.c (cgraph_node::get_body): Likewise.
27530 * coretypes.h: Likewise.
27531 * domwalk.c: Likewise.
27532 * domwalk.h: Likewise.
27533 * dumpfile.c (struct dump_option_value_info): Likewise.
27534 (dump_enable_all): Likewise.
27535 (dump_switch_p_1): Likewise.
27536 (opt_info_switch_p): Likewise.
27537 * dumpfile.h (enum tree_dump_index): Likewise.
27538 (struct dump_file_info): Likewise.
27539 * genemit.c: Likewise.
27540 * generic-match-head.c: Likewise.
27541 * gengtype.c (open_base_files): Likewise.
27542 * gimple-pretty-print.c: Likewise.
27543 * gimple-pretty-print.h: Likewise.
27544 * graph.c (print_graph_cfg): Likewise.
27545 * graphite-scop-detection.c (dot_all_sese): Likewise.
27546 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27547 * loop-unroll.c (report_unroll): Likewise.
27548 * passes.c (pass_manager::register_one_dump_file): Likewise.
27549 * print-tree.c: Likewise.
27550 * statistics.c: Likewise.
27551 * tree-cfg.c: Likewise.
27552 * tree-cfg.h: Likewise.
27553 * tree-dfa.c: Likewise.
27554 * tree-dfa.h: Likewise.
27555 * tree-dump.c (dump_function): Likewise.
27556 * tree-dump.h (struct dump_info): Likewise.
27557 * tree-pretty-print.c: Likewise.
27558 * tree-pretty-print.h: Likewise.
27559 * tree-ssa-live.c: Likewise.
27560 * tree-ssa-live.h: Likewise.
27561 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27562 * tree-vect-loop.c: Likewise.
27563 * tree-vect-slp.c: Likewise.
27565 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27566 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27568 PR tree-optimization/80457
27569 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27570 of all arguments to a statement as scalar_to_vec operations.
27571 (vectorizable_call): Adjust call to vect_model_simple_cost for
27573 (vectorizable_conversion): Likewise.
27574 (vectorizable_assignment): Likewise.
27575 (vectorizable_shift): Likewise.
27576 (vectorizable_operation): Likewise.
27577 (vectorizable_comparison): Likewise.
27578 (vect_is_simple_cond): Record the def types for operands.
27579 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27580 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27581 for statement argument count.
27583 2017-05-16 Carl Love <cel@us.ibm.com>
27585 * config/rs6000/rs6000-c: Add support for built-in functions
27586 vector unsigned long long vec_bperm (vector unsigned long long,
27587 vector unsigned char)
27588 vector signed long long vec_mule (vector signed int,
27590 vector unsigned long long vec_mule (vector unsigned int,
27591 vector unsigned int)
27592 vector signed long long vec_mulo (vector signed int,
27594 vector unsigned long long vec_mulo (vector unsigned int,
27595 vector unsigned int)
27596 vector signed char vec_sldw (vector signed char,
27597 vector signed char,
27599 vector unsigned char vec_sldw (vector unsigned char,
27600 vector unsigned char,
27602 vector signed short vec_sldw (vector signed short,
27603 vector signed short,
27605 vector unsigned short vec_sldw (vector unsigned short,
27606 vector unsigned short,
27608 vector signed int vec_sldw (vector signed int,
27611 vector unsigned int vec_sldw (vector unsigned int,
27612 vector unsigned int,
27614 vector signed long long vec_sldw (vector signed long long,
27615 vector signed long long,
27617 vector unsigned long long vec_sldw (vector unsigned long long,
27618 vector unsigned long long,
27620 * config/rs6000/rs6000-c: Add support for built-in functions
27621 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27622 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27623 * doc/extend.texi: Update the built-in documentation for the
27624 new built-in functions.
27626 2017-05-16 Marek Polacek <polacek@redhat.com>
27630 * tree.c (save_expr): Don't fold the expression.
27632 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27634 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27635 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27636 and (?*y,m). Update insn attributes.
27638 2017-05-16 Martin Liska <mliska@suse.cz>
27640 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27641 flags argument of print_gimple_stmt, print_gimple_expr,
27642 print_generic_stmt and print_generic_expr.
27643 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27644 * coretypes.h: Likewise.
27645 * except.c (dump_eh_tree): Likewise.
27646 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27647 * gimple-pretty-print.h: Likewise.
27648 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27649 (backprop::push_to_worklist): Likewise.
27650 (backprop::pop_from_worklist): Likewise.
27651 (backprop::process_use): Likewise.
27652 (backprop::intersect_uses): Likewise.
27653 (note_replacement): Likewise.
27654 * gimple-ssa-store-merging.c
27655 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27656 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27657 (pass_store_merging::execute): Likewise.
27658 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27659 (ssa_base_cand_dump_callback): Likewise.
27660 (dump_incr_vec): Likewise.
27661 (replace_refs): Likewise.
27662 (replace_mult_candidate): Likewise.
27663 (create_add_on_incoming_edge): Likewise.
27664 (create_phi_basis): Likewise.
27665 (insert_initializers): Likewise.
27666 (all_phi_incrs_profitable): Likewise.
27667 (introduce_cast_before_cand): Likewise.
27668 (replace_one_candidate): Likewise.
27669 * gimplify.c (gimplify_expr): Likewise.
27670 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27671 (set_rename): Likewise.
27672 (rename_uses): Likewise.
27673 (copy_loop_phi_nodes): Likewise.
27674 (add_close_phis_to_merge_points): Likewise.
27675 (copy_loop_close_phi_args): Likewise.
27676 (copy_cond_phi_args): Likewise.
27677 (graphite_copy_stmts_from_block): Likewise.
27678 (translate_pending_phi_nodes): Likewise.
27679 * graphite-poly.c (print_pdr): Likewise.
27680 (dump_gbb_cases): Likewise.
27681 (dump_gbb_conditions): Likewise.
27682 (print_scop_params): Likewise.
27683 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27684 (build_cross_bb_scalars_use): Likewise.
27685 (gather_bbs::before_dom_children): Likewise.
27686 * hsa-dump.c (dump_hsa_immed): Likewise.
27687 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27688 (get_replacement_map): Likewise.
27689 * ipa-inline-analysis.c (dump_condition): Likewise.
27690 (estimate_function_body_sizes): Likewise.
27691 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27692 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27693 * ipa-prop.c (ipa_dump_param): Likewise.
27694 (ipa_print_node_jump_functions_for_edge): Likewise.
27695 (ipa_modify_call_arguments): Likewise.
27696 (ipa_modify_expr): Likewise.
27697 (ipa_dump_param_adjustments): Likewise.
27698 (ipa_dump_agg_replacement_values): Likewise.
27699 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27700 * ipa-pure-const.c (check_stmt): Likewise.
27701 (pass_nothrow::execute): Likewise.
27702 * ipa-split.c (execute_split_functions): Likewise.
27703 * omp-offload.c (dump_oacc_loop_part): Likewise.
27704 (dump_oacc_loop): Likewise.
27705 * trans-mem.c (tm_log_emit): Likewise.
27706 (tm_memopt_accumulate_memops): Likewise.
27707 (dump_tm_memopt_set): Likewise.
27708 (dump_tm_memopt_transform): Likewise.
27709 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27710 (print_loop): Likewise.
27711 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27712 (chkp_gather_checks_info): Likewise.
27713 (chkp_get_check_result): Likewise.
27714 (chkp_remove_check_if_pass): Likewise.
27715 (chkp_use_outer_bounds_if_possible): Likewise.
27716 (chkp_reduce_bounds_lifetime): Likewise.
27717 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27718 (chkp_mark_completed_bounds): Likewise.
27719 (chkp_register_incomplete_bounds): Likewise.
27720 (chkp_mark_invalid_bounds): Likewise.
27721 (chkp_maybe_copy_and_register_bounds): Likewise.
27722 (chkp_build_returned_bound): Likewise.
27723 (chkp_get_bound_for_parm): Likewise.
27724 (chkp_build_bndldx): Likewise.
27725 (chkp_get_bounds_by_definition): Likewise.
27726 (chkp_generate_extern_var_bounds): Likewise.
27727 (chkp_get_bounds_for_decl_addr): Likewise.
27728 * tree-chrec.c (chrec_apply): Likewise.
27729 * tree-data-ref.c (dump_data_reference): Likewise.
27730 (dump_subscript): Likewise.
27731 (dump_data_dependence_relation): Likewise.
27732 (analyze_overlapping_iterations): Likewise.
27733 * tree-inline.c (expand_call_inline): Likewise.
27734 (tree_function_versioning): Likewise.
27735 * tree-into-ssa.c (dump_defs_stack): Likewise.
27736 (dump_currdefs): Likewise.
27737 (dump_names_replaced_by): Likewise.
27738 (dump_update_ssa): Likewise.
27739 (update_ssa): Likewise.
27740 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27741 * tree-parloops.c (build_new_reduction): Likewise.
27742 (try_create_reduction_list): Likewise.
27743 (ref_conflicts_with_region): Likewise.
27744 (oacc_entry_exit_ok_1): Likewise.
27745 (oacc_entry_exit_single_gang): Likewise.
27746 * tree-pretty-print.h: Likewise.
27747 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27748 (get_scalar_evolution): Likewise.
27749 (add_to_evolution): Likewise.
27750 (get_loop_exit_condition): Likewise.
27751 (analyze_evolution_in_loop): Likewise.
27752 (analyze_initial_condition): Likewise.
27753 (analyze_scalar_evolution): Likewise.
27754 (instantiate_scev): Likewise.
27755 (number_of_latch_executions): Likewise.
27756 (gather_chrec_stats): Likewise.
27757 (final_value_replacement_loop): Likewise.
27758 (scev_const_prop): Likewise.
27759 * tree-sra.c (dump_access): Likewise.
27760 (disqualify_candidate): Likewise.
27761 (create_access): Likewise.
27762 (reject): Likewise.
27763 (maybe_add_sra_candidate): Likewise.
27764 (create_access_replacement): Likewise.
27765 (analyze_access_subtree): Likewise.
27766 (analyze_all_variable_accesses): Likewise.
27767 (sra_modify_assign): Likewise.
27768 (initialize_constant_pool_replacements): Likewise.
27769 (find_param_candidates): Likewise.
27770 (decide_one_param_reduction): Likewise.
27771 (replace_removed_params_ssa_names): Likewise.
27772 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27773 * tree-ssa-copy.c (dump_copy_of): Likewise.
27774 (copy_prop_visit_cond_stmt): Likewise.
27775 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27776 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27777 (record_equivalences_from_stmt): Likewise.
27778 * tree-ssa-dse.c (compute_trims): Likewise.
27779 (delete_dead_call): Likewise.
27780 (delete_dead_assignment): Likewise.
27781 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27782 (forward_propagate_into_cond): Likewise.
27783 (pass_forwprop::execute): Likewise.
27784 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27785 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27787 (move_computations_worker): Likewise.
27788 (execute_sm): Likewise.
27789 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27790 (remove_exits_and_undefined_stmts): Likewise.
27791 (remove_redundant_iv_tests): Likewise.
27792 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27793 (adjust_iv_update_pos): Likewise.
27794 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27795 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27796 (value_replacement): Likewise.
27797 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27798 * tree-ssa-pre.c (print_pre_expr): Likewise.
27799 (get_representative_for): Likewise.
27800 (create_expression_by_pieces): Likewise.
27801 (insert_into_preds_of_block): Likewise.
27802 (eliminate_insert): Likewise.
27803 (eliminate_dom_walker::before_dom_children): Likewise.
27804 (eliminate): Likewise.
27805 (remove_dead_inserted_code): Likewise.
27806 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27807 * tree-ssa-reassoc.c (get_rank): Likewise.
27808 (eliminate_duplicate_pair): Likewise.
27809 (eliminate_plus_minus_pair): Likewise.
27810 (eliminate_not_pairs): Likewise.
27811 (undistribute_ops_list): Likewise.
27812 (eliminate_redundant_comparison): Likewise.
27813 (update_range_test): Likewise.
27814 (optimize_range_tests_var_bound): Likewise.
27815 (optimize_vec_cond_expr): Likewise.
27816 (rewrite_expr_tree): Likewise.
27817 (rewrite_expr_tree_parallel): Likewise.
27818 (linearize_expr): Likewise.
27819 (break_up_subtract): Likewise.
27820 (linearize_expr_tree): Likewise.
27821 (attempt_builtin_powi): Likewise.
27822 (attempt_builtin_copysign): Likewise.
27823 (transform_stmt_to_copy): Likewise.
27824 (transform_stmt_to_multiply): Likewise.
27825 (dump_ops_vector): Likewise.
27826 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27827 (print_scc): Likewise.
27828 (set_ssa_val_to): Likewise.
27829 (visit_reference_op_store): Likewise.
27830 (visit_use): Likewise.
27831 (sccvn_dom_walker::before_dom_children): Likewise.
27832 (run_scc_vn): Likewise.
27833 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27835 (expr_hash_elt::print): Likewise.
27836 (const_and_copies::pop_to_marker): Likewise.
27837 (const_and_copies::record_const_or_copy_raw): Likewise.
27838 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27839 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27840 (dump_predicates): Likewise.
27841 (find_uninit_use): Likewise.
27842 (warn_uninitialized_phi): Likewise.
27843 (pass_late_warn_uninitialized::execute): Likewise.
27844 * tree-ssa.c (verify_vssa): Likewise.
27845 (verify_ssa): Likewise.
27846 (maybe_optimize_var): Likewise.
27847 * tree-vrp.c (dump_value_range): Likewise.
27848 (dump_all_value_ranges): Likewise.
27849 (dump_asserts_for): Likewise.
27850 (register_edge_assert_for_2): Likewise.
27851 (vrp_visit_cond_stmt): Likewise.
27852 (vrp_visit_switch_stmt): Likewise.
27853 (vrp_visit_stmt): Likewise.
27854 (vrp_visit_phi_node): Likewise.
27855 (simplify_cond_using_ranges_1): Likewise.
27856 (fold_predicate_in): Likewise.
27857 (evrp_dom_walker::before_dom_children): Likewise.
27858 (evrp_dom_walker::push_value_range): Likewise.
27859 (evrp_dom_walker::pop_value_range): Likewise.
27860 (execute_early_vrp): Likewise.
27862 2017-05-16 Richard Biener <rguenther@suse.de>
27864 * dwarf2out.c (loc_list_from_tree_1): Do not create
27865 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27867 2017-05-16 Richard Biener <rguenther@suse.de>
27869 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27871 (note_variable_value_in_expr): If we resolved the decl ref
27872 do not push to the stack.
27874 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27876 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27877 operations in fast-math mode.
27878 (vaddq_f16): Likewise.
27879 (vmul_f16): Likewise.
27880 (vmulq_f16): Likewise.
27881 (vsub_f16): Likewise.
27882 (vsubq_f16): Likewise.
27883 * config/arm/neon.md (add<mode>3): New.
27885 (fma:<VH:mode>3): New. Also remove outdated comment.
27888 2017-05-16 Martin Liska <mliska@suse.cz>
27892 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27893 (odr_types_equivalent_p): Likewise.
27895 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27897 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27899 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27902 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27903 non-interunit SSE move alternatives with '?'.
27904 (zero-extendsidi peephole2): New peephole to skip intermediate
27905 general register in SSE zero-extend sequence.
27907 2017-05-15 Jeff Law <law@redhat.com>
27909 * reorg.c (relax_delay_slots): Create a new variable to hold
27910 the temporary target rather than clobbering TARGET_LABEL.
27912 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27913 missing argument to extract_bit_field call.
27914 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27916 2017-05-15 Martin Liska <mliska@suse.cz>
27919 * gcc.c (process_command): Do not allow empty argument of -o option.
27921 2017-05-15 Renlin Li <renlin.li@arm.com>
27923 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27924 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27925 * config/aarch64/constraints.md (Usf): Add long call check.
27926 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27927 (call_value): Likewise.
27928 (sibcall): Likewise.
27929 (sibcall_value): Likewise.
27931 (call_value_insn): New.
27932 (sibcall_insn): Update rtx pattern.
27933 (sibcall_value_insn): Likewise.
27934 (call_internal): Remove.
27935 (call_value_internal): Likewise.
27936 (sibcall_internal): Likewise.
27937 (sibcall_value_internal): Likewise.
27938 (call_reg): Likewise.
27939 (call_symbol): Likewise.
27940 (call_value_reg): Likewise.
27941 (call_value_symbol): Likewise.
27943 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27946 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27948 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27950 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27951 compatible with CCGOCmode and with CCZmode.
27953 2017-05-14 Martin Sebor <msebor@redhat.com>
27955 PR middle-end/77671
27956 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27957 (gimple_fold_builtin_snprintf): Same.
27958 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27959 (gimple_fold_builtin_snprintf): Same.
27960 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27961 of character types.
27962 (is_call_safe): New function.
27963 (try_substitute_return_value): Call it.
27964 (try_simplify_call): New function.
27965 (pass_sprintf_length::handle_gimple_call): Call it.
27967 2017-05-14 Martin Sebor <msebor@redhat.com>
27969 PR middle-end/80669
27970 * builtins.c (expand_builtin_stpncpy): Simplify.
27972 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27974 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27975 * config/i386/i386.h
27976 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27977 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27978 (struct machine_function): Add new members call_ms2sysv,
27979 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27980 (struct machine_frame_state): New fields sp_realigned and
27981 sp_realigned_offset.
27982 * config/i386/i386.c
27983 (enum xlogue_stub): New enum.
27984 (enum xlogue_stub_sets): New enum.
27985 (class xlogue_layout): New class.
27986 (struct ix86_frame): New fields stack_realign_allocate_offset,
27987 stack_realign_offset and outlined_save_offset. Modify comments to
27988 detail stack layout when using out-of-line stubs.
27989 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27990 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27991 -mcall-ms2sysv-xlogues.
27992 (stub_managed_regs): New static variable.
27993 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27994 registers managed by out-of-line stub.
27995 (disable_call_ms2sysv_xlogues): New function.
27996 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27997 m->call_ms2sysv when appropriate and compute frame layout for
27999 (sp_valid_at, fp_valid_at): New inline functions.
28000 (choose_basereg): New function.
28001 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28003 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28004 Use align parameter of choose_baseaddr to generated aligned SSE movs
28006 (pro_epilogue_adjust_stack): Modify to track
28007 machine_frame_state::sp_realigned.
28008 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28009 (ix86_nsaved_sseregs): Likewise.
28010 (ix86_emit_save_regs): Likewise.
28011 (ix86_emit_save_regs_using_mov): Likewise.
28012 (ix86_emit_save_sse_regs_using_mov): Likewise.
28013 (get_scratch_register_on_entry): Likewise.
28014 (gen_frame_set): New function.
28015 (gen_frame_load): Likewise.
28016 (gen_frame_store): Likewise.
28017 (emit_outlined_ms2sysv_save): Likewise.
28018 (emit_outlined_ms2sysv_restore): Likewise.
28019 (ix86_expand_prologue): Modify stack re-alignment code and call
28020 emit_outlined_ms2sysv_save when appropriate.
28021 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28022 parameter rtx_insn *insn, which allows the function to be used to only
28023 generate the notes.
28024 (ix86_expand_epilogue): Modify validity checks of frame and stack
28025 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28026 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28027 * config/i386/predicates.md
28028 (save_multiple): New predicate.
28029 (restore_multiple): Likewise.
28030 * config/i386/sse.md
28031 (save_multiple<mode>): New pattern.
28032 (save_multiple_realign<mode>): Likewise.
28033 (restore_multiple<mode>): Likewise.
28034 (restore_multiple_and_return<mode>): Likewise.
28035 (restore_multiple_leave_return<mode>): Likewise.
28036 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28038 2017-05-14 Julia Koval <julia.koval@intel.com>
28040 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28041 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28042 (__builtin_ia32_xsetbv): New builtins.
28043 * config/i386/i386.c (ix86_expand_special_args_builtin):
28045 (ix86_expand_builtin): Special expand for new intrinsics.
28046 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28047 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28048 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28050 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28052 * cfganal.c (inverted_post_order_compute): Change argument type
28054 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28055 * df-core.c (rest_of_handle_df_initialize): Adjust.
28056 (rest_of_handle_df_finish): Likewise.
28057 (df_analyze_1): Likewise.
28058 (df_analyze): Likewise.
28059 (loop_inverted_post_order_compute): Change argument to be a vec *.
28060 (df_analyze_loop): Adjust.
28061 (df_get_n_blocks): Likewise.
28062 (df_get_postorder): Likewise.
28063 * df.h (struct df_d): Change field to be a vec.
28064 * lcm.c (compute_laterin): Adjust.
28065 (compute_available): Likewise.
28066 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28067 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28068 * tree-ssa-pre.c (compute_antic): Likewise.
28070 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28072 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28073 (depth_first_search::depth_first_search): Change structure init
28074 function to this constructor.
28075 (depth_first_search::add_bb): Rename function to this member.
28076 (depth_first_search::execute): Likewise.
28077 (flow_dfs_compute_reverse_finish): Adjust.
28079 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28081 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28082 (longest_simple_path): Likewise.
28083 * shrink-wrap.c (spread_components): Likewise.
28084 (disqualify_problematic_components): Likewise.
28085 (emit_common_heads_for_components): Likewise.
28086 (emit_common_tails_for_components): Likewise.
28087 (insert_prologue_epilogue_for_components): Likewise.
28089 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28091 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28094 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28096 * df-core.c (df_set_blocks): Start using auto_bitmap.
28097 (df_compact_blocks): Likewise.
28098 * df-problems.c (df_rd_confluence_n): Likewise.
28099 * df-scan.c (df_insn_rescan_all): Likewise.
28100 (df_process_deferred_rescans): Likewise.
28101 (df_update_entry_block_defs): Likewise.
28102 (df_update_exit_block_uses): Likewise.
28103 (df_entry_block_bitmap_verify): Likewise.
28104 (df_exit_block_bitmap_verify): Likewise.
28105 (df_scan_verify): Likewise.
28106 * lra-constraints.c (lra_constraints): Likewise.
28107 (undo_optional_reloads): Likewise.
28108 (lra_undo_inheritance): Likewise.
28109 * lra-remat.c (calculate_gen_cands): Likewise.
28110 (do_remat): Likewise.
28111 * lra-spills.c (assign_spill_hard_regs): Likewise.
28112 (spill_pseudos): Likewise.
28113 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28114 (bitmap_set_subtract_values): Likewise.
28116 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28118 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28119 management with auto_bitmap.
28120 (fix_inter_tick): Likewise.
28121 (fix_recovery_deps): Likewise.
28122 * ira.c (add_store_equivs): Likewise.
28123 (find_moveable_pseudos): Likewise.
28124 (split_live_ranges_for_shrink_wrap): Likewise.
28125 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28126 (rtx_reuse_manager::seen_def_p): Likewise.
28127 (rtx_reuse_manager::set_seen_def): Likewise.
28128 * print-rtl.h (class rtx_reuse_manager): Likewise.
28130 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28132 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28134 (migrate_btr_def): Likewise.
28135 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28136 * df-core.c (loop_post_order_compute): Likewise.
28137 (loop_inverted_post_order_compute): Likewise.
28138 * hsa-common.h: Likewise.
28139 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28140 * init-regs.c (initialize_uninitialized_regs): Likewise.
28141 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28142 (inline_small_functions): Likewise.
28143 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28144 * ira.c (combine_and_move_insns): Likewise.
28145 (build_insn_chain): Likewise.
28146 * loop-invariant.c (find_invariants): Likewise.
28147 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28148 * predict.c (tree_predict_by_opcode): Likewise.
28149 (predict_paths_leading_to): Likewise.
28150 (predict_paths_leading_to_edge): Likewise.
28151 (estimate_loops_at_level): Likewise.
28152 (estimate_loops): Likewise.
28153 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28154 (spread_components): Likewise.
28155 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28156 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28157 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28158 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28159 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28160 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28161 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28162 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28163 (mark_threaded_blocks): Likewise.
28164 (thread_through_all_blocks): Likewise.
28165 * tree-ssa.c (verify_ssa): Likewise.
28166 (execute_update_addresses_taken): Likewise.
28167 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28169 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28171 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28173 (post_order_compute): Likewise.
28174 (inverted_post_order_compute): Likewise.
28175 (pre_and_rev_post_order_compute_fn): Likewise.
28177 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28179 * genrecog.c (int_set::int_set): Explicitly construct our
28180 auto_vec base class.
28181 * vec.h (auto_vec::auto_vec): New constructor.
28183 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28185 * bitmap.h (class auto_bitmap): New constructor taking
28186 bitmap_obstack * argument.
28188 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28190 * bitmap.h (class auto_bitmap): Change type of m_bits to
28191 bitmap_head, and adjust ctor / dtor and member operators.
28193 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28195 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28196 when returned register mode doesn't match original mode.
28198 2017-05-12 Jeff Law <law@redhat.com>
28199 Jakub Jelinek <jakub@redhat.com>
28201 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28202 we look for cc setter after the compare-elim changes.
28203 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28204 within the vector to match what compare-elim now expects.
28205 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28206 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28208 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28209 after the compare-elim changes.
28210 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28211 the vector to match what compare-elim now expects.
28212 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28213 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28214 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28215 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28216 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28218 * config/visium/visium.c (single_set_and_flags): Fix where
28219 we look for cc setter after the compare-elim changes.
28220 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28221 with the vector to match what compare-elim now expects.
28222 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28223 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28224 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28225 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28226 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28227 (neg<mode>2_insn_set_overflow): Likewise.
28229 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28231 PR middle-end/79794
28232 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28233 maybe_expand_insn call, set ops[0].target. If still set after call,
28234 set alt_rtl. Add extra arg to recursive calls.
28235 (extract_bit_field): Add alt_rtl argument. Pass to
28237 * expmed.h (extract_bit_field): Fix prototype.
28238 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28239 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28240 to extract_bit_field_calls.
28241 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28242 Pass alt_rtl to extract_bit_field calls.
28243 * calls.c (store_unaligned_arguments_into_psuedos)
28244 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28245 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28247 * optabs.h (struct expand_operand): Add target bitfield.
28249 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28251 * compare-elim.c (try_eliminate_compare): Canonicalize
28252 operation with embedded compare to
28253 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28254 (set (reg) (operation)].
28256 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28258 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28261 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28262 cost of adding a carry flag for ADC instruction.
28263 [case MINUS]: Ignore the cost of subtracting a carry flag
28264 for SBB instruction.
28266 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28268 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28270 * config/rs6000/bmiintrin.h: New file.
28271 * config/rs6000/bmi2intrin.h: New file.
28272 * config/rs6000/x86intrin.h: New file.
28274 2017-05-12 Jeff Law <law@redhat.com>
28276 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28279 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28281 PR middle-end/80707
28282 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28284 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28286 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28287 early expansion of vector divide builtins.
28288 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28289 builtins identified as having unsigned arguments.
28291 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28293 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28294 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28295 expansion of vector logical operations (and, andc, or, xor,
28298 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28300 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28301 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28303 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28305 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28306 early GIMPLE expansion of vector multiplies.
28308 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28310 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28311 TARGET_HAVE_MOVT conditional.
28312 (movt splitter): Likewise.
28314 2017-05-12 Richard Biener <rguenther@suse.de>
28316 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28317 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28318 Fold all stmts not inplace.
28320 2017-05-12 Richard Biener <rguenther@suse.de>
28322 PR tree-optimization/80713
28323 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28324 inserted_exprs bit for not removed stmts.
28326 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28328 PR middle-end/69921
28329 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28330 parallelized" attribute for parallelized OpenACC kernels.
28331 * omp-offload.c (execute_oacc_device_lower): Use it.
28333 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28334 Set "oacc kernels" attribute.
28335 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28336 parameter. Adjust all users.
28337 (oacc_fn_attrib_kernels_p): Remove function.
28338 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28339 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28340 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28341 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28342 assert "oacc kernels" attribute is set.
28344 2017-05-11 Carl Love <cel@us.ibm.com>
28346 * config/rs6000/rs6000-c: Add support for built-in functions
28347 vector unsigned char vec_popcnt (vector signed char)
28348 vector unsigned char vec_popcnt (vector unsigned char)
28349 vector unsigned short vec_popcnt (vector signed short)
28350 vector unsigned short vec_popcnt (vector unsigned short)
28351 vector unsigned int vec_popcnt (vector signed int)
28352 vector unsigned int vec_popcnt (vector unsigned int)
28353 vector unsigned long long vec_popcnt (vector signed long long)
28354 vector unsigned long long vec_popcnt (vector unsigned long long)
28355 vector signed long long vec_slo (vector signed long long,
28356 vector signed char)
28357 vector signed long long vec_slo (vector signed long long,
28358 vector unsigned char)
28359 vector unsigned long long vec_slo (vector unsigned long long,
28360 vector signed char)
28361 vector unsigned long long vec_slo (vector unsigned long long,
28362 vector unsigned char)
28363 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28364 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28365 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28366 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28367 * doc/extend.texi: Update the built-in documentation file for the
28368 new built-in functions.
28370 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28372 * attribs.h (sorted_attr_string): Move machine independent
28373 functions for target clone support from the i386 port to common
28374 code. Rename ix86_function_versions to common_function_versions.
28375 Rename make_name to make_unique_name.
28376 (common_function_versions): Likewise.
28377 (make_unique_name): Likewise.
28378 (make_dispatcher_decl): Likewise.
28379 (is_function_default_version): Likewise.
28380 * attribs.c (attr_strcmp): Likewise.
28381 (sorted_attr_string): Likewise.
28382 (common_function_versions): Likewise.
28383 (make_unique_name): Likewise.
28384 (make_dispatcher_decl): Likewise.
28385 (is_function_default_version): Likewise.
28386 * config/i386/i386.c (attr_strcmp): Likewise.
28387 (sorted_attr_string): Likewise.
28388 (ix86_function_versions): Likewise.
28389 (make_name): Likewise.
28390 (make_dispatcher_decl): Likewise.
28391 (is_function_default_version): Likewise.
28392 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28394 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28397 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28398 Account for direct move costs for vec_construct of integer
28401 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28404 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28405 (UNSPEC_STX_ATOMIC): Ditto.
28406 (loaddi_via_sse): New insn.
28407 (storedi_via_sse): Ditto.
28408 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28409 Update corresponding peephole2 patterns.
28410 (atomic_storedi_fpu): Ditto.
28412 2017-05-11 Julia Koval <julia.koval@intel.com>
28414 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28415 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28417 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28418 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28419 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28421 2017-05-11 Nathan Sidwell <nathan@acm.org>
28423 * graphite-poly.c: Include dumpfile.h.
28425 * dumpfle.h (dump_function): Declare here ...
28426 * tree-dump.h (dump_function): ... not here.
28427 * dumpfile.c: #include tree-cfg.h.
28428 (dump_function): Move here from ...
28429 * tree-dump.c (dump_function): ... here.
28430 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28431 * graphite-poly.c: Don't include tree-dump.h.
28432 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28433 * print-tree.c: Likewise.
28434 * stor-layout.c: Likewise.
28435 * tree-nested.c: Likewise.
28437 * dumpfile.c (dump_start): Use TDF_FLAGS.
28438 (dump_enable_all): Fix TDF_KIND check thinko.
28440 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28442 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28443 array entries to represent two legal parameterizations of the
28444 overloaded __builtin_cmpb function, as represented by the
28445 P6_OV_BUILTIN_CMPB constant.
28446 (altivec_resolve_overloaded_builtin): Add special case handling
28447 for the __builtin_cmpb function, as represented by the
28448 P6_OV_BUILTIN_CMPB constant.
28449 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28450 (BU_P6_64BIT_2): New macro.
28451 (BU_P6_OVERLOAD_2): New macro
28452 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28453 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28454 (CMPB): Add overload support to represent both 32-bit and 64-bit
28455 compare-bytes function.
28456 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28457 support for TARGET_CMPB.
28458 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28459 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28460 documentation of the __builtin_cmpb overloaded built-in function.
28462 2017-05-11 Richard Biener <rguenther@suse.de>
28464 PR tree-optimization/80705
28465 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28466 bases are not vectorizable.
28468 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28470 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28471 when counting register pressure.
28473 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28475 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28476 (iv_ca_more_deps): Renamed to ...
28477 (iv_ca_compare_deps): ... this.
28478 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28480 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28482 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28484 (determine_group_iv_costs): ... here.
28485 (find_inv_vars_cb): Record inv var if it's not recorded before.
28487 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28489 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28490 (get_shiftadd_cost): Ditto.
28492 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28494 * tree-ssa-address.c: Include header file.
28495 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28497 (add_to_parts): Refactor.
28498 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28499 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28502 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28504 PR tree-optimization/53090
28505 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28507 (extract_cond_operands): Detect condition with IV on both sides
28508 and return COMP_IV_EXPR_2.
28509 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28510 (rewrite_use_compare): Simplify by removing call to function
28511 extract_cond_operands.
28513 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28515 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28516 (extract_cond_operands): Detect condition comparing against non-
28517 invariant bound and return appropriate enum value.
28518 (find_interesting_uses_cond): Update use of extract_cond_operands.
28519 Handle its return value accordingly.
28520 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28522 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28524 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28525 nonlinear iv_use computation in loop invariant sensitive way.
28527 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28529 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28530 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28532 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28534 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28535 (dump_cand): Support iv_cand.inv_exprs.
28536 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28538 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28541 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28543 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28545 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28546 as local function. Include necessary header files.
28547 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28549 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28551 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28553 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28555 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28556 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28557 RSHIFT_EXPR and BIT_NOT_EXPR.
28559 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28561 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28562 (adjust_setup_cost): New parameter supporting round up adjustment.
28563 (struct address_cost_data): Delete.
28564 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28565 (split_address_cost, ptr_difference_cost): Delete.
28566 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28567 (struct ainc_cost_data): New struct.
28568 (get_address_cost_ainc): New function.
28569 (get_address_cost, get_computation_cost): Reimplement.
28570 (determine_group_iv_cost_address): Record inv_expr for all uses of
28572 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28573 (iv_ca_has_deps): Reimplemented to ...
28574 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28576 (iv_ca_extend): Call iv_ca_more_deps.
28578 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28580 * tree-ssa-address.c (struct mem_address): Move to header file.
28581 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28582 * tree-ssa-address.h (struct mem_address): Move from C file.
28583 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28585 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28587 * tree-affine.h (aff_combination_type): New interface.
28588 (aff_combination_zero_p): Remove static.
28589 (aff_combination_const_p): New interface.
28590 (aff_combination_singleton_var_p): New interfaces.
28592 2017-05-11 Richard Biener <rguenther@suse.de>
28594 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28595 Skip unreachable blocks and destinations.
28596 (eliminate): Move stmt removal and fixup ...
28597 (fini_eliminate): ... here. Skip inserted exprs.
28598 (pass_pre::execute): Move fini_pre after fini_eliminate.
28599 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28600 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28601 PRE to get rid of dead code that has invalid SSA form and
28602 split critical edges again.
28604 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28606 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28608 2017-05-11 Richard Biener <rguenther@suse.de>
28610 * passes.c (execute_function_todo): Verify loops if they are
28611 said to be up-to-date.
28612 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28613 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28615 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28618 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28619 handle calling assemble_external ourself.
28622 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28623 modes with zero size. Enhance comment.
28625 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28627 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28628 built-ins for vec_xl and vec_xst with short and char pointer
28631 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28633 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28634 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28635 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28636 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28637 (_mm_maskz_min_round_ss): New intrinsics.
28638 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28639 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28640 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28641 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28642 (__builtin_ia32_minss_mask_round): New builtins.
28643 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28644 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28645 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28647 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28648 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28650 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28653 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28655 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28656 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28657 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28658 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28659 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28660 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28661 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28662 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28663 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28664 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28665 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28666 (__builtin_ia32_mulss_mask_round): New builtins.
28667 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28668 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28669 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28671 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28672 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28674 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28677 2017-05-10 Julia Koval <julia.koval@intel.com>
28679 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28680 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28681 (_mm256_setr_m128i): New intrinsics.
28683 2017-05-10 Julia Koval <julia.koval@intel.com>
28685 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28686 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28687 (_mm_maskz_rcp14_ss): New intrinsics.
28688 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28689 (__builtin_ia32_rcp14ss_mask): New builtins.
28690 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28692 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28694 PR tree-optimization/51513
28695 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28696 (assert_unreachable_fallthru_edge_p): Use it.
28697 (group_case_labels_stmt): Likewise.
28698 * tree-cfg.h: Prototype it.
28699 * stmt.c: Include cfghooks.h and tree-cfg.h.
28700 (emit_case_dispatch_table) <gap_label>: New local variable.
28701 Use it to fill dispatch table gaps.
28702 Test for default_label before updating probabilities.
28703 (expand_case) <default_label>: Remove unneeded initialization.
28704 Test for unreachable default case statement and remove its edge.
28705 Set default_label accordingly.
28706 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28708 2017-05-10 Carl Love <cel@us.ibm.com>
28710 * config/rs6000/rs6000-c: Add support for built-in functions
28711 vector signed char vec_neg (vector signed char)
28712 vector signed short int vec_neg (vector short int)
28713 vector signed int vec_neg (vector signed int)
28714 vector signed long long vec_neg (vector signed long long)
28715 vector float vec_neg (vector float)
28716 vector double vec_neg (vector double)
28717 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28719 * config/rs6000/altivec.h: Add define for vec_neg
28720 * doc/extend.texi: Update the built-in documentation for the
28721 new built-in functions.
28723 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28725 PR tree-optimization/77644
28726 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28728 2017-05-10 Nathan Sidwell <nathan@acm.org>
28730 * dumpfile.h (TDI_lang_all): New.
28731 (TDF_KIND): New. Renumber others
28732 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28734 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28736 (get_dump_file_name): Adjust suffix generation.
28737 (dump_enable_all): Use TDF_KIND.
28738 * doc/invoke.texi (-fdump-lang-all): Document.
28740 * dumpfile.h: Tabify.
28742 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28745 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28746 Move member access before delete.
28748 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28750 * tree-inline.c (expand_call_inline): Split block at stmt
28753 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28756 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28757 are now unused after splitting mov{sf,sd}_hardfloat.
28758 (f32_lr2): Likewise.
28759 (f32_lm): Likewise.
28760 (f32_lm2): Likewise.
28761 (f32_li): Likewise.
28762 (f32_li2): Likewise.
28763 (f32_lv): Likewise.
28764 (f32_sr): Likewise.
28765 (f32_sr2): Likewise.
28766 (f32_sm): Likewise.
28767 (f32_sm2): Likewise.
28768 (f32_si): Likewise.
28769 (f32_si2): Likewise.
28770 (f32_sv): Likewise.
28771 (f32_dm): Likewise.
28772 (f32_vsx): Likewise.
28773 (f32_av): Likewise.
28774 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28775 For movsf, order stores so the VSX stores occur before the GPR
28776 store which encourages the register allocator to use a traditional
28777 FPR instead of a GPR. For movsd, order the stores so that the GPR
28778 store comes before the VSX stores to allow the power6 to work.
28779 This is due to the power6 not having a 32-bit integer store
28780 instruction from a FPR.
28781 (movsf_hardfloat): Likewise.
28782 (movsd_hardfloat): Likewise.
28784 2017-05-09 Martin Sebor <msebor@redhat.com>
28786 PR translation/80280
28787 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28790 PR translation/80280
28791 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28792 data member added in r247778.
28793 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28795 2017-05-09 Nathan Sidwell <nathan@acm.org>
28797 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28799 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28802 2017-05-09 Marek Polacek <polacek@redhat.com>
28804 * doc/invoke.texi: Fix typo.
28806 2017-05-09 Richard Biener <rguenther@suse.de>
28808 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28809 (vrp_val_is_min): Likewise.
28810 (set_value_range_to_value): Likewise.
28811 (set_value_range_to_nonnegative): Likewise.
28812 (gimple_assign_nonzero_p): Likewise.
28813 (gimple_stmt_nonzero_p): Likewise.
28814 (vrp_int_const_binop): Likewise. Remove unreachable case.
28815 (adjust_range_with_scev): Adjust comments.
28816 (compare_range_with_value): Likewise.
28817 (extract_range_from_phi_node): Likewise.
28818 (test_for_singularity): Likewise.
28820 2017-05-09 Richard Biener <rguenther@suse.de>
28822 * tree-vrp.c (get_single_symbol): Add assert that we don't
28823 get overflowed constants as invariant part.
28824 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28825 checks. Use wi::cmp instead of recursing for integer constants.
28826 (compare_values): Just ignore whether we assumed undefined
28827 overflow instead of failing the compare.
28828 (extract_range_for_var_from_comparison_expr): Add comment before the
28829 TREE_NO_WARNING sets.
28830 (test_for_singularity): Likewise.
28831 (extract_range_from_comparison): Do not disable optimization
28832 when we assumed undefined overflow.
28833 (extract_range_basic): Remove init of unused var.
28835 2017-05-09 Richard Biener <rguenther@suse.de>
28837 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28838 (extract_range_from_multiplicative_op_1): Adjust.
28839 (extract_range_from_binary_expr_1): Use int_const_binop.
28841 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28844 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28845 rs6000_store_data_bypass_p in seven define_bypass directives and
28846 in several comments.
28847 * config/rs6000/rs6000-protos.h: Add prototype for
28848 rs6000_store_data_bypass_p function.
28849 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28850 function implements slightly different (rs6000-specific) semantics
28851 than store_data_bypass_p, returning false rather than aborting
28852 with assertion error when arguments do not satisfy the
28853 requirements of store data bypass.
28854 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28855 rs6000_store_data_bypass_p.
28857 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28859 * config/xtensa/xtensa-protos.h
28860 (xtensa_initial_elimination_offset): New declaration.
28861 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28862 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28863 macro definition, add case for FRAME_POINTER_REGNUM when
28864 FRAME_GROWS_DOWNWARD.
28865 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28866 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28867 xtensa_initial_elimination_offset.
28869 2017-05-08 Nathan Sidwell <nathan@acm.org>
28871 * doc/invoke.texi: Alphabetize -fdump options.
28873 2017-05-08 Martin Sebor <msebor@redhat.com>
28875 PR translation/80280
28876 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28878 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28880 * target.def (compute_frame_layout): New optional target hook.
28881 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28882 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28883 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28885 * reload1.c (verify_initial_elim_offsets): Likewise.
28886 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28887 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28888 (arm_get_frame_offsets): Split up into this ...
28889 (arm_compute_frame_layout): ... and this function.
28891 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28893 * config/aarch64/constraints.md (Usa): New constraint.
28894 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28896 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28898 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28899 with_multilib_list after it has been checked.
28901 2017-05-08 Richard Biener <rguenther@suse.de>
28903 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28904 (bitmap_set_subtract_values): Likewise.
28906 2017-05-08 Richard Biener <rguenther@suse.de>
28908 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28909 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28911 (gimple_stmt_nonzero_warnv_p): Rename to ...
28912 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28914 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28915 (extract_range_basic): Adjust, do not disable propagation on
28916 strict overflow sensitive simplification.
28917 (vrp_visit_cond_stmt): Likewise.
28919 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28921 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28922 body size unconditionally.
28924 2017-05-07 Jeff Law <law@redhat.com>
28927 2017-05-06 Jeff Law <law@redhat.com>
28928 PR tree-optimization/78496
28929 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28932 PR tree-optimization/78496
28933 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28934 (simplify_stmt_using_ranges): Call it.
28935 (vrp_dom_walker::before_dom_children): Extract equivalences
28936 from an ASSERT_EXPR with an equality comparison against a
28939 2017-05-06 Jeff Law <law@redhat.com>
28941 PR tree-optimization/78496
28942 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28945 PR tree-optimization/78496
28946 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28947 (simplify_stmt_using_ranges): Call it.
28948 (vrp_dom_walker::before_dom_children): Extract equivalences
28949 from an ASSERT_EXPR with an equality comparison against a
28952 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28954 * lra-constraints.c (lra_copy_reg_equiv): New function.
28955 (split_reg): Use it to copy equivalence information from the
28956 original register to the spill register.
28958 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28960 PR rtl-optimization/75964
28961 * simplify-rtx.c (simplify_const_relational_operation): Remove
28962 invalid handling of comparisons of integer ABS.
28964 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28966 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28967 initialize to zero.
28968 (init_regs): Remove declaration.
28969 (function_arg_advance_32): Initialize error_p as boolean variable.
28971 2017-05-05 Nathan Sidwell <nathan@acm.org>
28973 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28974 lines. Use for (;;).
28976 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28978 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28979 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28980 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28981 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28982 VF=2 that require versioning.
28984 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28986 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28989 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28991 * diagnostic.h (diagnostic_override_option_index): Convert from
28992 macro to inline function.
28994 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28996 * diagnostic.c (last_module_changed_p): New function.
28997 (set_last_module): New function.
28998 (diagnostic_report_current_module): Convert macro usage to
28999 the above functions.
29000 * diagnostic.h (diagnostic_context::last_module): Strengthen
29001 from const line_map * to const line_map_ordinary *.
29002 (diagnostic_last_module_changed): Delete macro.
29003 (diagnostic_set_last_module): Delete macro.
29005 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29007 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29008 with diagnostic_report_diagnostic.
29009 (diagnostic_n_impl_richloc): Likewise.
29010 * diagnostic.h (report_diagnostic): Delete macro.
29011 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29012 with diagnostic_report_diagnostic.
29013 * substring-locations.c (format_warning_va): Likewise.
29015 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29017 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29018 save/restor of format_spec. Move option-printing code to...
29019 (print_option_information): ...this new function, and
29020 reimplement by simply printing to the pretty_printer,
29021 rather than appending to the format string.
29023 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29025 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29026 handling logic into...
29027 (update_effective_level_from_pragmas): ...this new function.
29029 2017-05-04 Andrew Waterman <andrew@sifive.com>
29031 * config/riscv/riscv.opt (mstrict-align): New option.
29032 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29033 (SLOW_UNALIGNED_ACCESS): Define.
29034 (riscv_slow_unaligned_access): Declare.
29035 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29037 (riscv_slow_unaligned_access): New variable.
29038 (rocket_tune_info): Set slow_unaligned_access to true.
29039 (optimize_size_tune_info): Set slow_unaligned_access to false.
29040 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29041 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29042 (riscv_option_override): Set riscv_slow_unaligned_access.
29043 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29045 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29047 * config/riscv/riscv.md: Unify indentation.
29049 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29054 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29056 (extendsi<mode>2): Add support for doing sign extension via
29057 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29058 don't have ISA 3.0 instructions.
29059 (extendsi<mode>2 splitter): Likewise.
29060 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29061 generate the normal insns since SImode can now go in vector
29062 registers. Disallow the special UNSPECs needed for previous
29063 machines to hide SImode being used. Add new insns
29064 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29065 (fix_trunc<mode>si2_stfiwx): Likewise.
29066 (fix_trunc<mode>si2_internal): Likewise.
29067 (fixuns_trunc<mode>si2): Likewise.
29068 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29069 (fctiw<u>z_<mode>_smallint): Likewise.
29070 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29071 of floating point to 32-bit integer from doing a direct move to
29072 the GPR registers to do a store.
29073 (fctiwz_<mode>): Break long line.
29075 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29077 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29078 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29079 (addr_list, addr_offset_valid_p): New.
29080 (split_address_groups): Check offset validity with above function.
29081 (gt-tree-ssa-loop-ivopts.h): Include header file.
29083 2017-05-05 Nathan Sidwell <nathan@acm.org>
29085 * config.gcc (arm*-*-*): Add missing 'fi'.
29087 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29089 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29090 included in -fopt-info does not matter.
29091 * doc/optinfo.texi (-fopt-info): Fix description of default
29092 behavour. Explicitly say order of options included in -fopt-info
29095 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29097 * config.gcc: Allow combinations of aprofile and rmprofile values for
29098 --with-multilib-list.
29099 * config/arm/t-multilib: New file.
29100 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29101 variables. Remove setting of ISA and floating-point ABI in
29102 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29103 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29104 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29105 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29107 * config/arm/t-rmprofile: Likewise except for the matches changes.
29108 * doc/install.texi (--with-multilib-list): Document the combination of
29109 aprofile and rmprofile values and warn about pitfalls in doing that.
29111 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29113 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29114 (movdi_aarch64): Likewise.
29116 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29118 PR tree-optimization/80632
29119 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29121 (build_arrays): Initialize it for virtual phis.
29122 (fix_phi_nodes): Use it for virtual phis.
29124 PR tree-optimization/80558
29125 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29126 [x, y] op z into [x op, y op z] for op & or | if conditions
29129 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29130 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29133 * config/arm/arm.md (use_literal_pool): Remove.
29134 (64-bit immediate split): No longer takes cost into consideration
29135 if arm_disable_literal_pool is enabled.
29136 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29137 used when arm_disable_literal_pool is enabled.
29138 (arm_max_const_double_inline_cost): Remove use of
29139 arm_disable_literal_pool.
29140 (push_minipool_fix): Add assert.
29141 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29142 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29143 (no_literal_pool_sf_immediate): New.
29145 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29147 PR tree-optimization/80613
29148 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29149 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29151 2017-05-05 Richard Biener <rguenther@suse.de>
29153 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29155 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29157 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29158 of this flag from insn conditions due to removal from r247495.
29160 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29162 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29164 (arm_early_store_addr_dep_ptr): Likewise.
29165 * config/arm/aarch-common-protos.h
29166 (arm_early_load_addr_dep_ptr): Add prototype.
29167 (arm_early_store_addr_dep_ptr): Likewise.
29168 * config/arm/cortex-a53.md: Add new bypasses.
29170 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29172 * tree.c (next_type_uid): Change type to unsigned.
29173 (type_hash_canon): Decrement back next_type_uid if
29174 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29175 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29178 2017-05-04 Martin Sebor <msebor@redhat.com>
29180 * builtins.c: Fix a trivial typo in a comment.
29182 PR middle-end/79234
29183 * builtins.c (check_sizes): Adjust to handle reading past the end.
29184 Avoid printing excessive upper bound of ranges. Use %E to print
29185 tree nodes instead of converting them to %wu.
29186 (expand_builtin_memchr): New function.
29187 (compute_dest_size): Rename...
29188 (compute_objsize): ...to this.
29189 (expand_builtin_memcpy): Adjust.
29190 (expand_builtin_mempcpy): Adjust.
29191 (expand_builtin_strcat): Adjust.
29192 (expand_builtin_strcpy): Adjust.
29193 (check_strncat_sizes): Adjust.
29194 (expand_builtin_strncat): Adjust.
29195 (expand_builtin_strncpy): Adjust and simplify.
29196 (expand_builtin_memset): Adjust.
29197 (expand_builtin_bzero): Adjust.
29198 (expand_builtin_memcmp): Adjust.
29199 (expand_builtin): Handle memcmp.
29200 (maybe_emit_chk_warning): Check strncat just once.
29202 2017-05-04 Martin Sebor <msebor@redhat.com>
29204 PR preprocessor/79214
29205 PR middle-end/79222
29206 PR middle-end/79223
29207 * builtins.c (check_sizes): Add inlining context and issue
29208 warnings even when -Wno-system-headers is set.
29209 (check_strncat_sizes): Same.
29210 (expand_builtin_strncat): Same.
29211 (expand_builtin_memmove): New function.
29212 (expand_builtin_stpncpy): Same.
29213 (expand_builtin): Handle memmove and stpncpy.
29215 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29217 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29218 which is not used any more.
29220 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29222 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29224 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29226 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29227 (cortexa53_tunings): Likewise.
29228 (cortexa57_tunings): Likewise.
29229 (cortexa72_tunings): Likewise.
29230 (cortexa73_tunings): Likewise.
29232 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29234 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29235 Set loop alignment to 8.
29237 2017-05-04 Martin Sebor <msebor@redhat.com>
29239 PR translation/80280
29240 * builtins.c (expand_builtin_object_size): Add missing quoting to
29241 %D and like directives.
29242 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29243 (hsa_type_for_tree_type): Same.
29244 (verify_function_arguments): Same.
29245 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29246 * varasm.c (get_section): Same.
29249 2017-05-04 Martin Sebor <msebor@redhat.com>
29251 PR translation/80280
29252 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29254 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29256 * config/aarch64/aarch64.c (generic_addrcost_table):
29257 Change HI/TI mode setting.
29259 2017-05-04 Martin Jambor <mjambor@suse.cz>
29261 PR tree-optimization/80622
29262 * tree-sra.c (comes_initialized_p): New function.
29263 (build_accesses_from_assign): Only set write lazily when
29264 comes_initialized_p is false.
29265 (analyze_access_subtree): Use comes_initialized_p.
29266 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29267 instead of testing for PARM_DECL.
29269 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29271 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29272 constraint on operand 0 to allow more general addressing modes.
29273 Adjust output template.
29274 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29276 * config/aarch64/aarch64-protos.h
29277 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29278 * config/aarch64/constraints.md (Dp): New address constraint.
29279 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29282 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29284 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29285 update use of estimate_ipcp_clone_size_and_time.
29286 (estimate_local_effects): Update use of
29287 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29288 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29289 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29290 Return nonspecialized time.
29292 2017-05-04 Richard Biener <rguenther@suse.de>
29294 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29295 for the last VUSE which def dominates the PHI. Directly call
29297 (get_continuation_for_phi_1): Remove.
29299 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29301 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29302 to explain the use of truncating division. Cap the number of
29303 iterations to the maximum given by nb_iterations_upper_bound,
29306 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29308 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29309 * configure: Regenerate.
29310 * config.in: Regenerate.
29311 * config/i386/driver-mingw32.c: new file.
29312 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29313 * config.host: Link driver-mingw32.o on MinGW host.
29314 * doc/install.texi: Document new --enable-mingw-wildcard configure
29317 2017-05-04 Marek Polacek <polacek@redhat.com>
29319 PR tree-optimization/80612
29320 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29322 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29323 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29325 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29326 (movt splitter): Likewise.
29327 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29328 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29329 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29330 block for Thumb-1 with MOVT.
29331 (thumb2_legitimate_address_p): Move code block ...
29332 (can_avoid_literal_pool_for_label_p): ... into this new function.
29333 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29335 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29336 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29337 "M-profile targets with the MOVT instruction".
29339 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29341 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29342 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29343 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29345 2017-05-04 Martin Liska <mliska@suse.cz>
29347 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29348 variable cond_code.
29350 2017-05-04 Richard Biener <rguenther@suse.de>
29352 * tree.c (array_at_struct_end_p): Handle arrays at struct
29353 end with flexarrays more conservatively. Refactor and treat
29354 arrays of arrays or aggregates more strict. Fix
29355 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29356 * tree.c (array_at_struct_end_p): Adjust prototype.
29357 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29358 * gimple-fold.c (get_range_strlen): Likewise.
29359 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29361 2017-05-04 Richard Biener <rguenther@suse.de>
29363 PR tree-optimization/31130
29364 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29366 (supports_overflow_infinity): Likewise.
29367 (is_negative_overflow_infinity): Likewise.
29368 (is_positive_overflow_infinity): Likewise.
29369 (is_overflow_infinity): Likewise.
29370 (stmt_overflow_infinity): Likewise.
29371 (overflow_infinity_range_p): Likewise.
29372 (usable_range_p): Remove as always returning true.
29373 (make_overflow_infinity): Remove.
29374 (negative_overflow_infinity): Likewise.
29375 (positive_overflow_infinity): Likewise.
29376 (avoid_overflow_infinity): Likewise.
29377 (set_value_range): Adjust accordingly.
29378 (set_value_range_to_nonnegative): Likewise, remove now unused
29379 overflow_infinity arg.
29380 (vrp_operand_equal_p): Adjust.
29381 (update_value_range): Likewise.
29382 (range_int_cst_singleton_p): Likewise.
29383 (operand_less_p): Likewise.
29384 (compare_values_warnv): Likewise.
29385 (extract_range_for_var_from_comparison_expr): Likewise.
29386 (vrp_int_const_binop): Likewise.
29387 (zero_nonzero_bits_from_vr): Likewise.
29388 (extract_range_from_multiplicative_op_1): Likewise.
29389 (extract_range_from_binary_expr_1): Likewise.
29390 (extract_range_from_unary_expr): Likewise.
29391 (extract_range_from_comparison): Likewise.
29392 (extract_range_basic): Likewise.
29393 (adjust_range_with_scev): Likewise.
29394 (compare_ranges): Likewise.
29395 (compare_range_with_value): Likewise.
29396 (dump_value_range): Likewise.
29397 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29399 (simplify_cond_using_ranges): Adjust.
29401 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29403 * brig-builtins.def: Added a builtin for class_f64.
29404 * builtin-types.def: Added a builtin type needed by class_f64.
29406 2017-05-03 Jason Merrill <jason@redhat.com>
29408 * timevar.def: Add TV_CONSTEXPR.
29410 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29412 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29414 2017-05-03 Martin Jambor <mjambor@suse.cz>
29416 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29417 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29418 * ipa-cp.c (ipcp_propagate_stage): Do not call
29419 ipa_update_after_lto_read.
29420 * ipa-inline.c (ipa_inline): Likewise.
29422 2017-05-03 Martin Jambor <mjambor@suse.cz>
29424 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29425 tag. Added a default constructor and a destructor.
29426 (ipa_edge_args_sum_t): New class;
29427 (ipa_edge_args_sum): Declare.
29428 (ipa_edge_args_vector): Remove declaration.
29429 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29430 (ipa_free_edge_args_substructures): Remove declaration.
29431 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29432 (ipa_edge_args_info_available_for_edge_p): Likewise.
29433 * ipa-prop.c (ipa_edge_args_vector): Removed.
29434 (edge_removal_hook_holder): Likewise.
29435 (edge_duplication_hook_holder): Likewise.
29436 (ipa_edge_args_sum): New variable.
29437 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29438 ipa_edge_args_vector.
29439 (ipa_free_edge_args_substructures): Likewise.
29440 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29441 ipa_edge_args_vector.
29442 (ipa_edge_removal_hook): Turned into method
29443 ipa_edge_args_sum_t::remove.
29444 (ipa_edge_duplication_hook): Turned into method
29445 ipa_edge_args_sum_t::duplicate.
29446 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29447 registering edge hooks.
29448 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29449 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29450 ipa_edge_args_sum instead of ipa_edge_args_vector.
29451 * ipa-profile.c (ipa_profile): Likewise.
29453 2017-05-03 Martin Jambor <mjambor@suse.cz>
29455 * symbol-summary.h (function_summary): New method exists.
29456 (function_summary::symtab_removal): Deallocate through release.
29457 (call_summary): New class.
29458 (gt_ggc_mx): New overload.
29459 (gt_pch_nx): Likewise.
29460 (gt_pch_nx): Likewise.
29462 2017-05-03 Jeff Law <law@redhat.com>
29464 PR tree-optimization/78496
29465 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29466 from simplify_cond_using_ranges. Split off code to walk
29467 backwards through casts into ...
29468 (simplify_cond_using_ranges_2): New function.
29469 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29470 (execute_vrp): After identifying jump threads, call
29471 simplify_cond_using_ranges_2.
29473 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29476 * ipa-inline.h (inline_summary): Add ctor.
29477 (create_ggc): Do not use ggc_cleared_alloc.
29479 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29480 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29482 * gcc.c (handle_braces): Support escaping in switch matching
29484 * doc/invoke.texi (Spec Files): Document it.
29485 Remove superfluous @code markup in items.
29487 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29489 * diagnostic-show-locus.c (struct column_range): New struct.
29490 (get_affected_columns): New function.
29491 (get_printed_columns): New function.
29492 (struct correction): New struct.
29493 (correction::ensure_capacity): New function.
29494 (correction::ensure_terminated): New function.
29495 (struct line_corrections): New struct.
29496 (line_corrections::~line_corrections): New dtor.
29497 (line_corrections::add_hint): New function.
29498 (layout::print_trailing_fixits): Reimplement in terms of the new
29500 (selftest::test_overlapped_fixit_printing): New function.
29501 (selftest::diagnostic_show_locus_c_tests): Call it.
29503 2017-05-03 Nathan Sidwell <nathan@acm.org>
29505 Canonicalize canonical type hashing
29506 * tree.h (type_hash_canon_hash): Declare.
29507 * tree.c (type_hash_list, attribute_hash_list): Move into
29508 type_hash_canon_hash.
29509 (build_type_attribute_qual_variant): Break out hash code calc into
29510 type_hash_canon_hash.
29511 (type_hash_canon_hash): New. Generic type hash computation.
29512 (build_range_type_1, build_array_type_1, build_function_type,
29513 build_method_type_directly, build_offset_type, build_complex_type,
29514 make_vector_type): Call it.
29516 2017-05-03 Richard Biener <rguenther@suse.de>
29518 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29519 When all DRs have unknown misaligned do not always peel
29520 when there is a store but apply the same costing model as if
29521 there were only loads.
29523 2017-05-03 Richard Biener <rguenther@suse.de>
29526 PR tree-optimization/80492
29527 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29528 compare_base_decls returning dont-know properly.
29530 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29532 * config/arm/iterators.md (CCSI): New mode iterator.
29533 (arch): New mode attribute.
29534 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29535 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29536 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29537 code iterator for success result mode.
29538 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29539 the corresponding new insn generators.
29541 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29544 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29545 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29547 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29549 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29550 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29551 (DDR_A): Wrap DDR argument in brackets.
29552 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29553 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29554 (DDR_REVERSED_P): Likewise.
29556 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29558 PR tree-optimization/79472
29559 * tree-switch-conversion.c (struct switch_conv_info): Add
29560 contiguous_range and default_case_nonstandard fields.
29561 (collect_switch_conv_info): Compute contiguous_range and
29562 default_case_nonstandard fields, don't clear final_bb if
29563 contiguous_range and only the default case doesn't have the required
29565 (check_all_empty_except_final): Set default_case_nonstandard instead
29566 of failing if contiguous_range and the default case doesn't have empty
29568 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29569 and only the default case doesn't have the required constants. Skip
29571 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29572 if default_case_nonstandard.
29573 (build_constructors): Build constant 1 just once. Assert that default
29574 values aren't inserted in between cases if contiguous_range. Skip
29576 (build_arrays): Skip virtual phis.
29577 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29578 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29579 Handle virtual phis.
29580 (gen_inbound_check): Handle default_case_nonstandard case.
29581 (process_switch): Adjust check_final_bb caller. Call
29582 gather_default_values with the first non-default case instead of
29583 default case if default_case_nonstandard.
29585 2017-05-02 Nathan Sidwell <nathan@acm.org>
29587 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29588 check. Fix formatting.
29590 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29592 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29593 errors when comparing specialized and unspecialized times.
29595 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29597 * diagnostic-show-locus.c
29598 (layout::should_print_annotation_line_p): Make private.
29599 (layout::print_annotation_line): Make private.
29600 (layout::annotation_line_showed_range_p): Make private.
29601 (layout::show_ruler): Make private.
29602 (layout::print_source_line): Make private. Pass in line and
29603 line_width, rather than calling location_get_source_line. Drop
29605 (layout::print_leading_fixits): New method.
29606 (layout::print_any_fixits): Rename to...
29607 (layout::print_trailing_fixits): ...this, and make private.
29608 Don't print newline fixits.
29609 (diagnostic_show_locus): Move logic for printing one row into...
29610 (layout::print_line): ...this new function. Move the
29611 location_get_source_line call and error-handling from
29612 print_source_line to here. Call print_leading_fixits, and rename
29613 print_any_fixits to print_trailing_fixits.
29614 (selftest::test_fixit_insert_containing_newline): Update now that
29615 newlines are partially supported.
29616 (selftest::test_fixit_insert_containing_newline_2): New test.
29617 (selftest::test_fixit_replace_containing_newline): Update comments.
29618 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29619 * edit-context.c (class added_line): New class.
29620 (class edited_line): Describe newline handling in comment.
29621 (edited_line::actually_edited_p): New method.
29622 (edited_line::print_content): Delete redundant decl.
29623 (edited_line::m_predecessors): New field.
29624 (edited_file::print_content): Call edited_line::print_content.
29625 (edited_file::print_diff): Update to support newlines.
29626 (edited_file::print_diff_hunk): Likewise.
29627 (edited_file::print_run_of_changed_lines): New function.
29628 (edited_file::print_diff_line): Convert to...
29629 (print_diff_line): ...this.
29630 (edited_file::get_effective_line_count): New function.
29631 (edited_line::edited_line): Initialize new field m_predecessors.
29632 (edited_line::~edited_line): Clean up m_predecessors.
29633 (edited_line::apply_fixit): Handle newlines.
29634 (edited_line::get_effective_line_count): New function.
29635 (edited_line::print_content): New function.
29636 (edited_line::print_diff_lines): New function.
29637 (selftest::test_applying_fixits_insert_containing_newline): New
29639 (selftest::test_applying_fixits_replace_containing_newline): New
29641 (selftest::insert_line): New function.
29642 (selftest::test_applying_fixits_multiple_lines): Add example of
29644 (selftest::edit_context_c_tests): Call the new tests.
29646 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29648 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29649 parameter cand. Update dump information.
29650 (get_computation_cost): Update uses.
29652 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29654 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29655 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29656 (get_computation_at, rewrite_use_address): Update use of
29657 get_computation_aff.
29659 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29661 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29662 (get_computation): Delete.
29663 (get_computation_cost): Implement like get_computation_cost_at.
29664 Use get_computation_at.
29665 (get_computation_cost_at): Delete.
29666 (rewrite_use_nonlinear_expr): Use get_computation_at.
29667 (rewrite_use_compare, remove_unused_ivs): Ditto.
29669 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29671 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29673 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29675 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29676 (ivopts_global_cost_for_size): Rename parameter and update uses.
29677 (iv_ca_recount_cost): Update uses.
29678 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29679 candidates seperately in n_invs and n_cands.
29680 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29682 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29684 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29685 (find_inv_vars_cb): New.
29686 (find_depends): Renamed to ...
29687 (find_inv_vars): ... this.
29688 (add_candidate_1, force_var_cost): Call find_inv_vars.
29689 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29691 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29693 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29694 inv_vars. Add inv_exprs.
29695 (struct iv_cand): Rename depends_on to inv_vars.
29696 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29697 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29698 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29699 (dump_cand): Dump inv_vars.
29700 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29701 (record_invariant, find_depends, add_candidate_1): Ditto.
29702 (set_group_iv_cost, force_var_cost): Ditto.
29703 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29704 (get_computation_cost_at, get_computation_cost): Ditto.
29705 (determine_group_iv_cost_generic): Ditto.
29706 (determine_group_iv_cost_address): Ditto.
29707 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29708 (determine_group_iv_costs): Ditto.
29709 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29710 (iv_ca_set_remove_invariants): Renamed to ...
29711 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29712 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29713 (iv_ca_set_add_invariants): Renamed to ...
29714 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29715 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29716 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29717 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29718 (create_new_ivs): Remove useless dump.
29720 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29722 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29724 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29725 (iv_ca_set_no_cp, create_new_iv): Ditto.
29727 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29729 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29731 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29733 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29734 function tree_check2.
29736 2017-05-02 Martin Liska <mliska@suse.cz>
29738 * doc/gcov.texi: Add missing preposition.
29739 * gcov.c (function_info::function_info): Properly fill up
29740 all member variables.
29742 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29744 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29746 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29748 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29750 2017-05-02 Martin Liska <mliska@suse.cz>
29753 * lto-streamer-in.c (lto_read_tree_1): Remove
29754 LTO_STREAMER_DEBUG.
29755 * lto-streamer.c (struct tree_hash_entry): Likewise.
29756 (struct tree_entry_hasher): Likewise.
29757 (tree_entry_hasher::hash): Likewise.
29758 (tree_entry_hasher::equal): Likewise.
29759 (lto_streamer_init): Likewise.
29760 (lto_orig_address_map): Likewise.
29761 (lto_orig_address_get): Likewise.
29762 (lto_orig_address_remove): Likewise.
29763 * lto-streamer.h: Likewise.
29764 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29765 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29767 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29769 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29770 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29771 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29772 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29773 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29774 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29775 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29776 (mm_maskz_sub_ss): New intrinsics.
29777 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29778 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29779 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29780 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29781 (__builtin_ia32_subss_mask_round): New builtins.
29782 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29783 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29784 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29786 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29787 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29789 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29792 2017-05-02 Martin Jambor <mjambor@suse.cz>
29794 PR tree-optimization/78687
29795 * tree-sra.c (access): New field parent.
29796 (process_subtree_disqualification): New function.
29797 (disqualify_candidate): Call it.
29798 (build_accesses_from_assign): Reset write flag if creating an
29800 (build_access_subtree): Fill in parent field and also prpagate
29801 down grp_write flag.
29802 (create_artificial_child_access): New parameter set_grp_write, set
29803 grp_write to its value.
29804 (propagate_subaccesses_across_link): Also propagate grp_write flag
29806 (propagate_all_subaccesses): Push the closest parent back to work
29807 queue if add_access_to_work_queue returned true.
29809 2017-05-02 Richard Biener <rguenther@suse.de>
29811 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29812 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29813 -fstrict-overflow documentation.
29814 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29815 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29816 flag_strict_overflow.
29817 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29818 * lto-opts.c (lto_write_options): Do not stream it.
29819 * lto-wrapper.c (merge_and_complain): Do not handle it.
29820 * opts.c (default_options_table): Do not set -fstrict-overflow.
29821 (finish_options): Likewise do not clear it when sanitizing.
29822 * simplify-rtx.c (simplify_const_relational_operation): Do not
29823 test flag_strict_overflow.
29825 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29827 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29828 using enabled attribute.
29829 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29830 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29831 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29832 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29833 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29834 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29835 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29836 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29837 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29838 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29840 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29842 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29844 2017-05-02 Richard Biener <rguenther@suse.de>
29846 PR tree-optimization/80591
29848 2017-04-10 Richard Biener <rguenther@suse.de>
29850 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29853 2017-05-02 Richard Biener <rguenther@suse.de>
29855 PR tree-optimization/80549
29856 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29857 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29858 headers if they do not have a preheader.
29860 2017-05-02 Martin Liska <mliska@suse.cz>
29863 * common.opt: Fix typo.
29864 * doc/invoke.texi: Likewise.
29866 2017-05-01 Jan Beulich <jbeulich@suse.com>
29868 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29869 swapping, add (x,x,m,x,n) alternative.
29871 2017-05-01 Nathan Sidwell <nathan@acm.org>
29873 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29874 unnecessary unadjusted_alignment check.
29876 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29879 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29881 * gimplify.c (gimplify_cilk_detach): New function.
29882 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29883 * tree-core.h: Document EXPR_CILK_SPAWN.
29884 * tree.h (EXPR_CILK_SPAWN): Define.
29886 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29888 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29889 to use new fixit_hint representation, using the "replace" logic.
29890 (get_line_span_for_fixit_hint): Likewise.
29891 (layout::print_any_fixits): Likewise.
29892 (selftest::test_one_liner_many_fixits): Rename to...
29893 (selftest::test_one_liner_many_fixits_1): ...this, and update
29894 comment and expected output to reflect that the multiple fix-it
29895 hints are now consolidated into one insertion.
29896 (selftest::test_one_liner_many_fixits_2): New test.
29897 (selftest::test_diagnostic_show_locus_one_liner): Update for
29899 (selftest::test_fixit_consolidation): Update for fix-it API
29901 * diagnostic.c (print_parseable_fixits): Likewise.
29902 * edit-context.c (edited_line::m_line_events): Convert from
29903 auto_vec <line_event *> to auto_vec <line_event>.
29904 (class line_event): Convert from abstract base class to a concrete
29905 class, taking over the role of replace_event.
29906 (class insert_event): Delete.
29907 (class replace_event): Rename to class line_event. Convert to
29909 (edit_context::add_fixits): Reimplement.
29910 (edit_context::apply_insert): Delete.
29911 (edit_context::apply_replace): Rename to...
29912 (edit_context::apply_fixit): ...this. Convert to half-open range.
29913 (edited_file::apply_insert): Delete.
29914 (edited_file::apply_replace): Rename to...
29915 (edited_file::apply_fixit): ...this.
29916 (edited_line::~edited_line): Drop deletion of events.
29917 (edited_line::apply_insert): Delete.
29918 (edited_line::apply_replace): Rename to...
29919 (edited_line::apply_fixit): ...this. Convert to half-open range.
29920 Update for change to type of m_line_events.
29921 * edit-context.h (edit_context::apply_insert): Delete.
29922 (edit_context::apply_replace): Rename to...
29923 (edit_context::apply_fixit): ...this.
29925 2017-05-01 Martin Sebor <msebor@redhat.com>
29927 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29930 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29933 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29934 __get_cpuid_max returns 0.
29935 (__get_cpuid_count): Ditto.
29937 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29939 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29940 replacement expression is another instance of one of its arguments.
29942 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29945 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29946 check for stack push/pop autoinc.
29947 * config/i386/i386.c (ix86_agi_dependent): Return false
29948 if the only reason why modified_in_p returned true is that
29949 addr is SP based and set_insn is a push or pop.
29951 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29953 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29956 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29959 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29960 (account_size_time): Use two predicates - exec_pred and
29962 (evaluate_conditions_for_known_args): Compute both clause and
29964 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29965 (inline_summary_t::duplicate): Update.
29966 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29968 (compute_inline_parameters): Likewise.
29969 (estimate_edge_size_and_time): Update caluclation of time.
29970 (estimate_node_size_and_time): Compute both time and nonspecialized
29972 (estimate_ipcp_clone_size_and_time): Update.
29973 (inline_merge_summary): Update.
29974 (do_estimate_edge_time): Update.
29975 (do_estimate_edge_size): Update.
29976 (do_estimate_edge_hints): Update.
29977 (inline_read_section, inline_write_summary): Stream both new predicates.
29978 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29980 (compute_inlined_call_time): Cleanup.
29981 (big_speedup_p): Update.
29982 (edge_badness): Update.
29983 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29984 (size_time_entry): Replace predicate by exec_predicate and
29985 nonconst_predicate.
29986 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29987 (estimate_edge_time): Return also nonspec_time.
29988 (reset_edge_growth_cache): Update.
29990 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29992 PR rtl-optimization/80491
29993 * ifcvt.c (noce_process_if_block): When looking for x setter
29994 with missing else_bb, don't check only the insn right before
29995 cond_earliest, but look for the last insn that x is modified in
29996 within the same bb.
29998 PR rtl-optimization/80491
29999 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30001 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30003 PR tree-optimization/80487
30004 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30006 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30008 PR tree-optimization/79697
30009 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30010 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30011 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30013 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30014 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30016 2017-04-28 Martin Sebor <msebor@redhat.com>
30018 PR tree-optimization/80523
30019 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30020 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30022 (maybe_warn, format_directive, parse_directive): Use new functions.
30023 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30025 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30027 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30029 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30031 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30032 target_header_dir): Set correctly.
30033 * configure: Regenerated.
30034 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30035 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30036 instead of SYSTEM_HEADER_DIR.
30038 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30040 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30041 (estimate_local_effects): Likewise.
30042 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30043 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30044 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30045 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30046 do_estimate_edge_time, estimate_edge_time): Likewise.
30047 * ipa-inline-analysis.c (estimate_node_size_and_time,
30048 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30049 (estimate_time_after_inlining): Remove.
30051 2017-04-28 Martin Liska <mliska@suse.cz>
30053 * doc/gcov.texi: Enhance documentation of gcov.
30055 2017-04-28 Martin Liska <mliska@suse.cz>
30057 * doc/gcov.texi: Sort options in alphabetic order.
30058 * doc/gcov-dump.texi: Likewise.
30059 * doc/gcov-tool.texi: Likewise.
30060 * gcov.c (print_usage): Likewise.
30061 * gcov-dump.c (print_usage): Likewise.
30062 * gcov-tool.c (print_merge_usage_message): Likewise.
30063 (print_rewrite_usage_message): Likewise.
30064 (print_overlap_usage_message): Likewise.
30066 2017-04-28 Martin Liska <mliska@suse.cz>
30068 PR gcov-profile/53915
30069 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30071 2017-04-28 Martin Liska <mliska@suse.cz>
30073 PR gcov-profile/79891
30074 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30075 is marked by compiler as living on a line.
30076 (get_cycles_count): Remove usage of the union.
30077 (output_intermediate_file): Likewise.
30078 (find_source): Fix GNU coding style.
30079 (accumulate_line_counts): Remove old non-all block mode.
30080 (output_lines): Remove usage of the union.
30081 * profile.c (output_location): Include all BBs, even if
30082 belonging to a same line (and file) as a previous BB.
30084 2017-04-28 Martin Liska <mliska@suse.cz>
30086 * gcov.c (process_args): Handle new argument 'w'.
30087 (read_graph_file): Assign ID to BBs.
30088 (output_branch_count): Display BB # if verbose flag is set.
30089 (output_lines): Likewise for arcs.
30090 (print_usage): Add '--verbose' option help.
30091 * doc/gcov.texi: Document --verbose (-w) option.
30093 2017-04-28 Martin Liska <mliska@suse.cz>
30095 * gcov.c (struct block_location_info): New struct.
30096 (process_file): Fill up the new structure.
30097 (read_graph_file): Replace usage of encoding by the newly added
30099 (add_line_counts): Likewise.
30100 (accumulate_line_counts): Remove usage of the union.
30101 (function_info::function_info): New function.
30102 (function_info::~function_info): Likewise.
30103 (process_file): Call delete instead of release_function.
30104 (release_function): Release the function.
30105 (release_structures): Call delete instead of release_function.
30106 (solve_flow_graph): Replace usage of num_blocks.
30107 (find_exception_blocks): Likewise.
30108 (output_lines): Fix GNU coding style.
30110 2017-04-28 Martin Liska <mliska@suse.cz>
30113 * coverage.c (coverage_remove_note_file): New function.
30114 * coverage.h: Declare the function.
30115 * toplev.c (finalize): Clean if an error has been seen.
30117 2017-04-28 Martin Liska <mliska@suse.cz>
30119 PR gcov-profile/80031
30120 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30121 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30122 * gcov.c (read_graph_file): Read just number of blocks.
30123 * profile.c (branch_prob): Do not stream 0 flags per a basic
30126 2017-04-28 Martin Liska <mliska@suse.cz>
30128 * gcov-dump.c (tag_*): Add new argument to declarations.
30129 (dump_gcov_file): Likewise.
30130 (tag_blocks): Add and use new argument depth.
30131 (tag_arcs): Likewise.
30132 (tag_lines): Likewise.
30133 (tag_counters): Likewise.
30134 (tag_summary): Likewise.
30135 (dump_working_sets): Use depth to do a proper indentation.
30137 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30140 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30141 * symtab.c (symtab_node::debug_symtab): Move definition here.
30143 2017-04-28 Richard Biener <rguenther@suse.de>
30145 * lto-streamer.h (LTO_major_version): Bump to 7.
30147 2017-04-28 Richard Biener <rguenther@suse.de>
30149 * tree-vrp.c (assert_info): New struct.
30150 (add_assert_info): New helper.
30151 (register_edge_assert_for_2): Refactor to add asserts to a vector
30153 (register_edge_assert_for_1): Likewise.
30154 (register_edge_assert_for): Likewise.
30155 (finish_register_edge_assert_for): New helper actually registering
30156 asserts where live on edge.
30157 (find_conditional_asserts): Adjust.
30158 (find_switch_asserts): Likewise.
30159 (evrp_dom_walker::try_find_new_range): Generalize.
30160 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30162 2017-04-27 Marek Polacek <polacek@redhat.com>
30165 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30166 arg10 and arg11 to itype.
30168 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30170 * doc/extend.texi (Object Size Checking): Improve grammar.
30172 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30175 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30176 that the logic for permitting reciprocal estimates matches that
30179 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30182 * tree.c (type_cache_hasher::equal): Only compare
30183 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30184 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30185 non-aggregate element types.
30186 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30187 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30189 2017-04-27 Richard Biener <rguenther@suse.de>
30191 PR middle-end/80533
30192 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30193 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30194 keeping a reference to a trailing array.
30196 2017-04-27 Richard Biener <rguenther@suse.de>
30198 PR middle-end/80539
30199 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30200 being in loop-closed SSA form conservatively.
30201 (chrec_fold_multiply_poly_poly): Likewise.
30203 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30205 PR middle-end/79665
30206 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30207 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30209 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30212 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30213 (aarch64_function_arg_alignment): Return unsigned int again, but still
30214 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30215 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30216 Don't emit -Wpsabi note.
30217 (aarch64_function_arg_boundary): Likewise.
30218 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30221 2017-04-26 Nathan Sidwell <nathan@acm.org>
30223 * tree.h (crc32_unsigned_n): Declare.
30224 (crc32_unsigned, crc32_unsigned): Make inline.
30225 * tree.c (crc32_unsigned_bits): Replace with ...
30226 (crc32_unsigned_n): ... this.
30227 (crc32_unsigned, crc32_byte): Remove.
30228 (crc32_string): Remove unnecessary braces.
30230 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30232 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30233 * ipa-inline-analysis.c (MAX_TIME): Remove.
30234 (account_size_time): Use sreal for time.
30235 (dump_inline_summary): Update.
30236 (estimate_function_body_sizes): Update.
30237 (estimate_edge_size_and_time): Update.
30238 (estimate_calls_size_and_time): Update.
30239 (estimate_node_size_and_time): Update.
30240 (inline_merge_summary): Update.
30241 (inline_update_overall_summary): Update.
30242 (estimate_time_after_inlining): Update.
30243 (inline_read_section): Update.
30244 (inline_write_summary): Update.
30245 * ipa-inline.c (compute_uninlined_call_time): Update.
30246 (compute_inlined_call_time): Update.
30247 (recursive_inlining): Update.
30248 (inline_small_functions): Update.
30249 (dump_overall_stats): Update.
30250 * ipa-inline.h: Include sreal.h.
30251 (size_time_entry): Turn time to sreal.
30252 (inline_summary): Turn self_time nad time to sreal.
30254 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30256 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30258 (sreal::stream_out, sreal::stream_in): New.
30259 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30261 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30263 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30266 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30269 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30271 (dimode_scalar_chain::compute_convert_gain): Ditto.
30272 (dimode_scalar_chain::make_vector_copies): Ditto.
30273 (dimode_scalar_chain::convert_reg): Ditto.
30274 (dimode_scalar_chain::convert_insn): Ditto.
30275 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30276 (VI248_AVX512BW_1): New mode iterator.
30277 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30278 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30281 2017-04-25 Martin Sebor <msebor@redhat.com>
30283 PR tree-optimization/80497
30284 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30285 constants are representable in HOST_WIDE_INT.
30286 (parse_directive): Ditto.
30288 2017-04-25 Martin Sebor <msebor@redhat.com>
30291 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30292 (new_zero_array): Adjust signature.
30293 (dom_info::dom_init): Used unsigned rather that size_t.
30294 (dom_info::dom_info): Same.
30296 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30297 Jakub Jelinek <jakub@redhat.com>
30300 * config/arm/arm.c: Include gimple.h.
30301 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30302 returns negative, increment ncrn only if it returned positive.
30303 (arm_needs_doubleword_align): Return int instead of bool,
30304 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30305 members, but if there is any such non-FIELD_DECL
30306 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30307 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30308 returns negative, increment nregs only if it returned positive.
30309 (arm_setup_incoming_varargs): Likewise.
30310 (arm_function_arg_boundary): Emit -Wpsabi note if
30311 arm_needs_doubleword_align returns negative, return
30312 DOUBLEWORD_ALIGNMENT only if it returned positive.
30314 2017-04-25 Marek Polacek <polacek@redhat.com>
30317 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30318 first argument to type.
30320 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30323 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30324 type checks to test for compatibility instead of equality.
30326 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30327 Jakub Jelinek <jakub@redhat.com>
30330 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30332 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30333 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30334 the alignment computation, but return their maximum in warn_alignment.
30335 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30336 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30338 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30339 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30342 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30344 * config/arc/simdext.md (dmpyh): Fix typo.
30346 2017-04-25 Richard Biener <rguenther@suse.de>
30348 PR tree-optimization/80492
30349 * alias.c (compare_base_decls): Handle registers with asm
30350 specification conservatively.
30351 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30352 compare_base_decls returning dont-know properly.
30354 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30356 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30357 (legitimate_offset_address_p): New function.
30358 (arc_legitimate_address_p): Use above function.
30360 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30362 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30364 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30366 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30367 ACCH registers whenever they are available.
30369 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30371 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30372 double regs fix when not used.
30374 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30376 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30378 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30379 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30381 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30383 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30384 when emitting short ADD instructions.
30386 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30388 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30390 (cmpsi_cc_c_insn): Likewise.
30391 (cbranchsi4_scratch): Compute proper instruction length using
30392 compact_hreg_operand.
30393 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30395 2017-04-25 Richard Biener <rguenther@suse.de>
30397 PR middle-end/80509
30398 * passes.c (pass_manager::pass_manager): Initialize
30399 m_name_to_pass_map.
30401 2017-04-25 Richard Biener <rguenther@suse.de>
30403 PR tree-optimization/79201
30404 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30406 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30409 * config/s390/vector.md: Split MEM->GPR vector moves for
30410 non-s_operand addresses.
30412 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30415 * config/s390/predicates.md (reload_const_wide_int_operand): New
30417 * config/s390/s390.md ("movti"): Remove d/P alternative.
30418 ("movti_bigconst"): New pattern definition.
30420 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30423 * s390-protos.h (s390_expand_cs_hqi): Removed.
30424 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30425 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30426 modes as well as CCZ1mode and CCZmode.
30427 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30428 signature of s390_emit_compare_and_swap.
30429 (s390_expand_cs_hqi): Likewise, make static.
30430 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30431 compare-and-swap, in some cases.
30432 (s390_expand_cs): Wrapper function.
30433 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30435 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30436 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30437 patterns for small and large integers. Forbid symref memory operands.
30438 Move expander to s390.c. Require cc register.
30439 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30440 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30441 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30442 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30443 symref memory operands. Remove CC mode and call s390_match_ccmode
30445 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30447 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30449 * config/s390/s390.md (define_peephole2): New peephole to help
30450 combining the load-and-test pattern with volatile memory.
30452 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30454 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30455 with CCZmode for TARGET_Z196.
30457 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30459 PR rtl-optimization/80501
30460 * combine.c (make_compound_operation_int): Set subreg_code to SET
30461 even for AND with mask of the sign bit of mode.
30463 PR rtl-optimization/80500
30464 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30465 sum's initial value.
30467 2017-04-25 Julian Brown <julian@codesourcery.com>
30468 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30470 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30472 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30474 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30476 2017-04-25 Julian Brown <julian@codesourcery.com>
30477 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30479 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30480 (thunderx2t99_sha): New Reservation.
30482 2017-04-25 Julian Brown <julian@codesourcery.com>
30483 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30485 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30486 type for 1-element load.
30488 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30490 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30492 2017-04-24 Martin Jambor <mjambor@suse.cz>
30494 PR tree-optimization/80293
30495 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30496 char arrays not totally scalarizable if it is false.
30497 (analyze_all_variable_accesses): Pass correct value in the new
30498 parameter. Add a statistics counter.
30500 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30502 PR middle-end/79931
30503 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30505 2017-04-24 Richard Biener <rguenther@suse.de>
30507 PR tree-optimization/80494
30508 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30509 out for complex types.
30511 2017-04-24 Richard Biener <rguenther@suse.de>
30513 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30514 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30515 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30516 (DFS): Adjust and never fail.
30517 (sccvn_dom_walker::fail): Remove.
30518 (sccvn_dom_walker::before_dom_children): Adjust.
30519 (run_scc_vn): Likewise and never fail.
30520 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30521 (pass_fre::execute): Likewise.
30523 2017-04-24 Richard Biener <rguenther@suse.de>
30525 PR tree-optimization/79725
30526 * tree-ssa-sink.c (statement_sink_location): Return whether
30527 failure reason was zero uses. Move that check later.
30528 (sink_code_in_bb): Deal with zero uses by removing the stmt
30531 2017-04-24 Richard Biener <rguenther@suse.de>
30534 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30535 pointer-based references.
30537 2017-04-24 Richard Biener <rguenther@suse.de>
30540 * pass_manager.h (pass_manager::operator new): Remove.
30541 (pass_manager::operator delete): Likewise.
30542 * passes.c (pass_manager::operator new): Remove.
30543 (pass_manager::operator delete): Likewise.
30544 (pass_manager::pass_manager): Zero individual pass members.
30546 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30549 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30550 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30551 Check "XEXP (src, 1)" operand here.
30552 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30553 Check "XEXP (src, 1)" operand here.
30554 (dimode_scalar_chain::make_vector_copies): Detect count register
30555 of a shift instruction. Zero extend count register from QImode
30556 to DImode to satisfy vector shift pattern count operand predicate.
30557 Substitute vector shift count operand with a DImode copy.
30558 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30561 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30563 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30564 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30565 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30566 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30567 (UNSPEC_NOREX_MEM): Remove definition.
30569 2017-04-21 Richard Biener <rguenther@suse.de>
30571 PR tree-optimization/79547
30572 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30573 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30574 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30575 without any constraints.
30577 2017-04-21 Richard Biener <rguenther@suse.de>
30579 PR tree-optimization/78847
30580 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30582 2017-04-21 Richard Biener <rguenther@suse.de>
30584 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30585 (build_distinct_type_copy): Likewise.
30586 (build_variant_type_copy): Likewise.
30587 * tree.c (build_qualified_type): Pass down mem-stat info.
30588 (build_distinct_type_copy): Likewise.
30589 (build_variant_type_copy): Likewise.
30591 2017-04-21 Richard Biener <rguenther@suse.de>
30593 PR tree-optimization/80237
30594 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30596 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30597 for a simplified result.
30599 2016-04-21 Richard Biener <rguenther@suse.de>
30601 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30602 sth as strict as a simple_iv but a chrec without symbols and an
30603 operand defined in the loop we are peeling (and not some subloop).
30604 (propagate_constants_for_unrolling): Propagate all constants.
30606 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30609 * config/i386/i386.c (print_reg): Remove assert for disalowed
30610 regno values, call output_operand_lossage instead.
30612 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30615 * config/i386/constraints.md (Yc): New register constraint.
30616 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30617 Use Yc constraint for alternative 2 of operand 0. Remove
30618 preferred_for_speed attribute.
30620 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30622 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30623 lastprivate clauses in SIMT case.
30625 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30627 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30629 2017-04-20 Richard Biener <rguenther@suse.de>
30631 PR tree-optimization/57796
30632 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30633 as N scalar stores.
30634 (vect_model_load_cost): Cost gathers as N scalar loads.
30636 2017-04-20 Richard Biener <rguenther@suse.de>
30638 * ggc-page.c (ggc_allocated_p): Rename to ...
30639 (safe_lookup_page_table_entry): ... this and return the lookup
30641 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30643 2017-04-20 Richard Biener <rguenther@suse.de>
30645 PR tree-optimization/80453
30646 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30647 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30648 from the conditions.
30649 (vn_phi_eq): Pass them down.
30650 (vn_phi_lookup): Record them.
30651 (vn_phi_insert): Likewise.
30653 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30655 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30656 uninitialized variable warning to avoid buffer overrun.
30658 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30661 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30662 is suppressed for '{ 0 }' in C.
30664 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30666 * BASE-VER: Set to 8.0.0.
30668 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30670 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30671 priority .init_array and .fini_array section with SECTION_NOTYPE
30674 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30676 PR middle-end/80423
30677 * tree.h (build_array_type): Add typeless_storage default argument.
30678 * tree.c (type_cache_hasher::equal): Also compare
30679 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30680 (build_array_type): Add typeless_storage argument, set
30681 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30683 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30684 (build_array_type): Likewise. Add typeless_storage argument.
30686 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30687 Jakub Jelinek <jakub@redhat.com>
30689 PR tree-optimization/80426
30690 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30691 operation on symbolic operands, also compute the overflow for the
30692 invariant part when the operation degenerates into a negation.
30694 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30697 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30698 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30701 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30703 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30706 * config/avr/avr.c (tree.h): Include it.
30707 (cgraph.h): Include it.
30708 (avr_encode_section_info): Don't warn for uninitialized progmem
30709 variable if it's just an alias.
30711 2017-04-19 Richard Biener <rguenther@suse.de>
30714 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30715 when needed by AutoPGO.
30717 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30720 * doc/lto.texi: Remove an extra 'that'.
30722 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30724 PR rtl-optimization/80429
30725 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30726 are only used in debug insns.
30728 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30729 Vladimir Makarov <vmakarov@redhat.com>
30731 * config/sparc/predicates.md (input_operand): Add comment. Return
30732 true for any memory operand when LRA is in progress.
30733 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30735 2017-04-18 Jeff Law <law@redhat.com>
30738 * mips.md ({return,simple_return}_internal): Do not overwrite
30741 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30743 PR tree-optimization/80443
30744 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30745 instead of adding 1, subtract -1 and similarly instead of subtracting
30748 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30750 PR rtl-optimization/80357
30751 * haifa-sched.c (tmp_bitmap): New variable.
30752 (model_recompute): Handle duplicate use records.
30753 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30754 (free_global_sched_pressure_data): Free it.
30756 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30759 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30760 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30761 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30762 instead of SYSTEM_HEADER_DIR.
30764 2017-04-18 Jeff Law <law@redhat.com>
30766 PR middle-end/80422
30767 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30768 predecessors after walking up the insn chain.
30770 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30773 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30774 sizetype type into debug info.
30776 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30779 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30780 unneeded test for TARGET_UPPER_REGS_SF.
30781 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30783 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30786 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30787 instead of gsi_after_labels.
30789 2017-04-18 Jeff Law <law@redhat.com>
30791 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30795 2017-04-13 Jeff Law <law@redhat.com>
30796 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30797 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30799 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30802 * config/avr/avr.c (intl.h): Include it.
30803 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30805 2017-04-18 Martin Liska <mliska@suse.cz>
30807 PR gcov-profile/78783
30808 * gcov-tool.c (gcov_output_files): Validate that destination
30809 file is either removed by the tool or by a user.
30811 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30812 Guy Benyei <guybe@mellanox.com>
30814 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30815 block, and do not negate it, the stored id is already negative.
30817 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30819 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30821 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30824 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30825 masks of options that should be turned off if the VSX vector
30826 options are turned off.
30827 (OTHER_P8_VECTOR_MASKS): Likewise.
30828 (OTHER_VSX_VECTOR_MASKS): Likewise.
30829 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30830 rs6000_disable_incompatible_switches to validate no type switches
30832 (rs6000_incompatible_switch): New function to disallow turning on
30833 other vector options if -mno-vsx, -mno-power8-vector, or
30834 -mno-power9-vector are specified.
30836 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30838 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30840 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30842 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30843 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30844 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30845 (ARG_POINTER_CFA_OFFSET): Likewise.
30847 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30849 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30850 conditions to take advantage of various optimizations.
30852 2017-04-13 Jeff Law <law@redhat.com>
30854 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30855 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30856 (zero_extendsidi2_dext): Likewise.
30858 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30861 * fold-const.c (fold_ternary_loc): Revert
30862 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30865 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30867 PR rtl-optimization/80343
30868 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30869 new scratch pseudo.
30871 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30874 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30875 to ubsan_encode_value.
30877 2017-04-13 Jeff Law <law@redhat.com>
30879 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30880 appearing in DEBUG_INSNs.
30882 2017-04-13 Martin Liska <mliska@suse.cz>
30884 PR gcov-profile/80413
30885 * gcov-io.c (gcov_write_string): Copy to buffer just when
30886 allocated size is greater than zero.
30888 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30891 * dwarf2out.c (decls_for_scope): Ignore declarations of
30892 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30894 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30897 * ipa-visibility.c (non_local_p): Fix typos.
30898 (localize_node): When localizing symbol in same comdat group,
30899 dissolve the group only when we know external symbols are going
30901 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30903 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30905 PR tree-optimization/79390
30906 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30907 order does not result in usable sequence, retry with reversed operand
30913 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30914 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30915 op0 instead of fold_convert_loc (loc, type, arg0).
30917 2017-04-12 Jeff Law <law@redhat.com>
30919 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30920 has a delay slot in the generated code.
30922 * config/cris/cris.md (cris_preferred_reload_class): Return
30923 GENNONACR_REGS rather than GENERAL_REGS.
30925 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30928 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30929 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30930 signedness of the result type.
30932 2017-04-12 Richard Biener <rguenther@suse.de>
30933 Jeff Law <law@redhat.com>
30935 PR tree-optimization/80359
30936 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30937 trim stores to TARGET_MEM_REFs.
30939 2017-04-12 Richard Biener <rguenther@suse.de>
30941 PR tree-optimization/79390
30942 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30943 threading case even more.
30945 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30948 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30949 for quad_address_p for TImode, instead of just not indexed_address.
30951 2017-04-12 Richard Biener <rguenther@suse.de>
30952 Bernd Edlinger <bernd.edlinger@hotmail.de>
30954 PR middle-end/79671
30955 * alias.c (component_uses_parent_alias_set_from): Handle
30956 TYPE_TYPELESS_STORAGE.
30957 (get_alias_set): Likewise.
30958 * tree-core.h (tree_type_common): Add typeless_storage flag.
30959 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30960 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30961 for types containing members with TYPE_TYPELESS_STORAGE.
30962 (place_field): Likewise.
30963 (layout_type): Likewise for ARRAY_TYPE.
30964 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30965 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30966 TYPE_TYPELESS_STORAGE.
30967 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30969 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30972 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30973 first argument to type.
30975 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30979 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30980 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30981 (rs6000_expand_binop_builtin): Likewise.
30982 (rs6000_expand_ternop_builtin): Likewise; also add missing
30983 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30984 vshasigma built-ins.
30985 * doc/extend.texi: Document that vec_xxpermdi's third argument
30986 must be a constant.
30988 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30990 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30991 Use shift_const cost parameter when calculating gain of STV shifts.
30993 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30995 PR rtl-optimization/70478
30996 * lra-constraints.c (process_alt_operands): Check memory for
30997 disfavoring memory insn operand.
30999 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31001 PR middle-end/80100
31002 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31003 left shift in unsigned HOST_WIDE_INT type.
31005 PR rtl-optimization/80385
31006 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31007 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31010 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31011 if they have any depend clauses.
31013 2017-04-11 Martin Liska <mliska@suse.cz>
31016 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31017 * ipa-split.c (split_function): Create a local comdat symbol
31018 if caller is in a comdat group.
31020 2017-04-11 Martin Liska <mliska@suse.cz>
31023 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31026 2017-04-11 Martin Sebor <msebor@redhat.com>
31028 PR middle-end/80364
31029 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31030 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31032 (directive::set_width, directive::set_precision, format_character):
31034 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31037 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31040 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31041 conflict, set target->arch_name instead of target->cpu_name.
31043 2017-04-11 Richard Biener <rguenther@suse.de>
31045 PR tree-optimization/80374
31046 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31047 build_zero_cst, remove fold_convertible_p check again.
31049 2017-04-11 Martin Liska <mliska@suse.cz>
31052 * ubsan.c (instrument_object_size): Do not instrument register
31055 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31058 * config/i386/i386-builtin-types.def
31059 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31060 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31061 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31062 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31063 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31064 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31065 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31066 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31067 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31068 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31069 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31070 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31071 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31072 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31073 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31074 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31075 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31076 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31077 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31078 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31079 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31080 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31081 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31082 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31083 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31084 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31085 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31086 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31087 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31088 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31089 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31090 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31091 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31092 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31093 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31094 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31095 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31096 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31097 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31098 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31099 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31100 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31101 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31102 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31103 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31104 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31106 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31107 flag to second_arg_count, handle 4 argument function type _COUNT
31108 aliases, handle second_arg_count on second argument rather than last.
31110 2017-04-10 Jeff Law <law@redhat.com>
31112 PR tree-optimization/80374
31113 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31114 record anything if we can not convert integer_zero_node to the
31117 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31120 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31121 Enhance special handling given to the TARGET_P9_MINMAX option in
31122 relation to certain other options.
31124 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31126 PR tree-optimization/80153
31127 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31128 remove POINTER_PLUS_EXPR's base part directly, rather than through
31131 2017-04-10 Richard Biener <rguenther@suse.de>
31132 Bin Cheng <bin.cheng@arm.com>
31134 PR tree-optimization/80153
31135 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31136 the first element of pointer type aff_tree. Build result expr in
31138 (add_elt_to_tree): Convert to type unconditionally. Remove other
31139 fold_convert calls.
31140 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31141 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31143 2017-04-10 Richard Biener <rguenther@suse.de>
31145 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31148 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31150 PR rtl-optimization/70478
31151 * lra-constraints.c (curr_small_class_check): New.
31152 (update_and_check_small_class_inputs): New.
31153 (process_alt_operands): Update curr_small_class_check. Disfavor
31154 alternative insn memory operands. Check available regs for small
31157 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31160 * config/mips/mips.opt (-mvirt): Update description.
31161 * doc/invoke.texi (-mvirt): Likewise.
31163 2017-04-10 Richard Biener <rguenther@suse.de>
31165 PR middle-end/80362
31166 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31167 looking for NEGATE_EXPR in -A / -B to A / B folding.
31169 2017-04-10 Martin Liska <mliska@suse.cz>
31171 PR gcov-profile/80224
31172 * gcov.c (print_usage): Fix usage string.
31173 (get_gcov_intermediate_filename): Remove.
31174 (output_gcov_file): Use both for normal and intermediate format.
31175 (generate_results): Do not initialize special file for
31176 intermediate format.
31178 2017-04-10 Richard Biener <rguenther@suse.de>
31180 PR tree-optimization/80304
31181 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31184 2017-04-10 Nathan Sidwell <nathan@acm.org>
31187 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31188 (rs6000_init_builtins): Use it.
31190 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31192 * config/arm/arm.md (<mrc>): Add mode to SET source.
31193 (<mrrc>): Likewise.
31195 2017-04-10 Richard Biener <rguenther@suse.de>
31197 PR middle-end/80344
31198 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31200 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31203 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31204 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31205 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31206 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31207 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31208 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31209 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31210 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31211 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31212 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31213 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31214 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31215 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31216 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31217 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31218 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31219 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31220 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31221 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31222 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31223 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31224 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31225 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31227 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31229 PR rtl-optimization/70478
31230 * lra-constraints.c: Reverse the last patch.
31232 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31234 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31235 Add comment for WCHAR_T.
31237 2017-04-08 Martin Liska <mliska@suse.cz>
31240 2017-04-07 Martin Liska <mliska@suse.cz>
31243 * ipa-split.c (split_function): Add function part to a same comdat
31246 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31249 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31251 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31253 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31254 (vec_pairing): ...to this.
31255 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31256 (rs6000_sched_init): Adjust for name change.
31257 (struct rs6000_sched_context): Likewise.
31258 (rs6000_init_sched_context): Likewise.
31259 (rs6000_set_sched_context): Likewise.
31261 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31267 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31269 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31270 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31271 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31273 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31275 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31277 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31279 PR rtl-optimization/70703
31280 * ira-color.c (update_conflict_hard_regno_costs): Use
31281 int64_t instead of HOST_WIDE_INT.
31283 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31285 PR rtl-optimization/70478
31286 * lra-constraints.c (process_alt_operands): Disfavor alternative
31287 insn memory operands.
31289 2017-04-07 Jeff Law <law@redhat.com>
31291 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31292 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31294 2017-04-07 Martin Liska <mliska@suse.cz>
31297 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31298 Show error message instead of an ICE.
31300 2017-04-07 Martin Liska <mliska@suse.cz>
31303 * ipa-split.c (split_function): Add function part to a same comdat
31306 2017-04-07 Richard Biener <rguenther@suse.de>
31308 PR middle-end/80341
31309 * tree.c (get_unwidened): Also handle ! for_type case for
31311 * convert.c (do_narrow): Split out from ...
31312 (convert_to_integer_1): ... here. Do not pass final truncation
31313 type to get_unwidened for TRUNC_DIV_EXPR.
31315 2017-04-07 Richard Biener <rguenther@suse.de>
31317 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31319 (aff_combination_const): Adjust.
31320 (aff_combination_scale): Likewise.
31321 (aff_combination_add_elt): Likewise.
31322 (aff_combination_add_cst): Likewise.
31323 (aff_combination_convert): Likewise.
31324 (add_elt_to_tree): Likewise. Remove unused argument.
31325 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31327 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31329 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31331 * config/arm/arm.c (arm_default_short_enums): Use
31332 ARM_DEFAULT_SHORT_ENUMS.
31333 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31335 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31338 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31339 members with redundant out-of-class redeclaration.
31341 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31344 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31345 * config/i386/i386.md (*zero_extendsidi2):
31346 Add (?*x,*x) and (?*v,*v) alternatives.
31348 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31351 * config/i386/i386.c (ix86_expand_builtin)
31352 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31353 mode from insn data. Convert operands to insn operand mode.
31354 Copy operands that don't satisfy insn predicate to a register.
31356 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31358 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31361 2017-04-06 Richard Biener <rguenther@suse.de>
31363 PR tree-optimization/80334
31364 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31365 preserve alignment of accesses.
31367 2017-04-06 Richard Biener <rguenther@suse.de>
31369 PR tree-optimization/80262
31370 * tree-sra.c (build_ref_for_offset): Preserve address-space
31372 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31373 Drop useless address-space information on MEM_REF offsets.
31375 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31377 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31379 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31381 PR rtl-optimization/70703
31382 * ira-color.c (update_conflict_hard_regno_costs): Use
31383 HOST_WIDE_INT instead of long.
31385 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31388 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31389 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31391 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31392 for x86_64 target. Handle -m3dnowa option.
31394 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31396 PR rtl-optimization/70703
31397 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31398 (update_conflict_hard_regno_costs): Use long instead of unsigned
31399 arithmetic for cost calculation.
31401 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31402 Bernd Edlinger <bernd.edlinger@hotmail.de>
31405 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31408 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31411 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31412 ptr_mode with Pmode throughout.
31413 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31414 into probe_stack_range and use DImode.
31416 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31419 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31420 call_eh_return is true.
31422 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31424 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31425 Initialize last_match_fntype_index.
31427 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31430 * tree-nvr.c: Include internal-fn.h.
31431 (pass_return_slot::execute): Ignore internal calls without
31434 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31435 Richard Biener <rguenther@suse.de>
31438 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31439 captures used multiple times, except for the last use.
31440 * generic-match-head.c: Include gimplify.h.
31442 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31444 PR tree-optimization/79390
31445 * target.h (struct noce_if_info): Declare.
31446 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31447 * target.def (noce_conversion_profitable_p): New target hook.
31448 * ifcvt.h (struct noce_if_info): New type, moved from ...
31449 * ifcvt.c (struct noce_if_info): ... here.
31450 (noce_conversion_profitable_p): Renamed to ...
31451 (default_noce_conversion_profitable_p): ... this. No longer
31453 (noce_try_store_flag_constants, noce_try_addcc,
31454 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31455 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31456 instead of noce_conversion_profitable_p.
31457 * config/i386/i386.c: Include ifcvt.h.
31458 (ix86_option_override_internal): Don't override
31459 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31460 (ix86_noce_conversion_profitable_p): New function.
31461 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31462 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31463 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31464 * doc/tm.texi: Regenerated.
31466 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31468 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31471 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31474 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31475 instructions for small multiply cores.
31477 2017-04-04 Jeff Law <law@redhat.com>
31479 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31481 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31482 that are not set by the loop over the elements.
31484 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31487 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31488 int mode, convert_modes it to mode as unsigned, otherwise use
31489 lowpart_subreg to mode rather than SImode.
31490 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31491 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31492 Use DImode instead of SImode for the shift count operand.
31493 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31496 2017-04-04 Richard Biener <rguenther@suse.de>
31498 PR middle-end/80281
31499 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31500 arithmetic done for the negate or the plus. Simplify.
31501 (A - (-B) -> A + B): Likewise.
31502 * fold-const.c (split_tree): Make sure to not negate pointers.
31504 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31506 PR rtl-optimization/60818
31507 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31508 a compare of comparisons with the thing compared if this results
31509 in a different machine mode.
31511 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31513 * alias.c (base_alias_check): Fix typo in comment.
31514 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31515 * cgraphunit.c (symbol_table::compile): Likewise.
31516 * collect2.c (maybe_run_lto_and_relink): Likewise.
31517 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31518 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31519 * config/avr/avr.c (avr_map_op_t): Likewise.
31520 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31521 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31522 * config/epiphany/epiphany.md (movcc): Likewise.
31523 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31524 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31526 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31527 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31528 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31529 * config/sh/sh.c (sh_rtx_costs): Likewise.
31530 * fold-const.c (fold_truth_andor): Likewise.
31531 * genautomata.c (collapse_flag): Likewise.
31532 * gengtype.h (struct type::u::s): Likewise.
31533 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31534 * input.c (FORMAT_AMOUNT): Likewise.
31535 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31536 (known_aggs_to_agg_replacement_list): Likewise.
31537 * ipa-inline-analysis.c: Likewise.
31538 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31539 * ipa-polymorphic-call.c
31540 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31541 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31542 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31544 * modulo-sched.c (apply_reg_moves): Likewise.
31545 * omp-expand.c (build_omp_regions_1): Likewise.
31546 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31547 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31548 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31549 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31550 * value-prof.c: Likewise.
31551 * var-tracking.c (val_reset): Likewise.
31553 2017-04-03 Richard Biener <rguenther@suse.de>
31555 PR tree-optimization/80275
31556 * fold-const.c (split_address_to_core_and_offset): Handle
31559 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31561 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31562 descriptors is at least equal to that of functions.
31564 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31566 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31568 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31571 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31572 (mov<IMOD4:mode>): New expander.
31573 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31575 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31577 PR rtl-optimization/79405
31578 * fwprop.c (propagations_left): New variable.
31579 (forward_propagate_into): Decrement it.
31580 (fwprop_init): Initialize it.
31581 (fw_prop): If the variable has reached zero, stop propagating.
31582 (fwprop_addr): Ditto.
31584 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31587 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31588 a FUNCTION_DECL, pass it as decl instead of origin to
31591 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31593 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31596 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31599 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31600 TARGET_VSX_SMALL_INTEGER.
31602 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31604 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31605 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31607 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31609 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31610 extraction from odd-numbered MSA register.
31612 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31614 PR middle-end/80173
31615 * expmed.c (store_bit_field_1): Don't attempt to create
31616 a word subreg out of hard registers wider than word if they
31617 have HARD_REGNO_NREGS of 1 for their mode.
31619 PR middle-end/80163
31620 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31621 conversions to integer types wider than word and pointer.
31624 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31625 (rtx_equal_for_cselib_p): Pass 0 to it.
31626 * cselib.c (cselib_hasher::equal): Likewise.
31627 (rtx_equal_for_cselib_1): Add depth argument. If depth
31628 is 128, don't look up VALUE locs and punt. Increment
31629 depth in recursive calls when walking VALUE locs.
31631 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31633 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31634 (make_gcov_file_name): Use the canonical path name for generating
31636 (read_line): Fix handling of files with ascii null bytes.
31638 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31640 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31641 to initialise a vector register instead
31642 of using a const_int.
31644 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31646 PR translation/80189
31647 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31648 diagnostic messages.
31650 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31653 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31654 (dfp_diex_<mode>): Update mode of operand 1.
31655 * doc/extend.texi (dxex, dxexq): Document change to return type.
31656 (diex, diexq): Document change to argument type.
31658 2017-03-30 Martin Jambor <mjambor@suse.cz>
31661 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31662 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31663 it reflects the signature changes performed at the callee side.
31664 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31665 to cgraph_build_function_type_skip_args.
31666 (build_function_decl_skip_args): Adjust call to the above function.
31668 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31671 * config/i386/sse.md
31672 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31673 register as dest whenever it is a MEM not rtx_equal_p to the
31674 corresponding dup operand, and when forcing into reg move the
31675 reg into the memory afterwards.
31676 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31677 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31678 for the force_reg mode.
31679 (avx512vl_vextractf128<mode>): Use register as dest either
31680 always when a MEM, or when it is a MEM not rtx_equal_p to the
31681 corresponding dup operand, or even not when it is a CONST_VECTOR
31682 depending on the mode and lo vs. hi.
31683 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31685 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31686 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31687 Likewise. Require that operands[2] is even.
31688 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31689 Remove extraneous parens. Require that operands[2] is a multiple
31691 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31692 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31693 disallow memory then.
31695 2017-03-30 Richard Biener <rguenther@suse.de>
31697 PR tree-optimization/77498
31698 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31699 to non-constants over backedges.
31701 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31703 PR rtl-optimization/80233
31704 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31705 as last_combined_insn. Do not test for BARRIER_P separately.
31707 2017-03-29 Andreas Schwab <schwab@suse.de>
31710 * calls.c (prepare_call_address): Convert funexp to Pmode before
31711 copying to temp reg.
31713 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31715 PR tree-optimization/80158
31716 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31717 Handle possible future case of more than one alternate
31719 (replace_rhs_if_not_dup): Likewise.
31720 (replace_one_candidate): Likewise.
31722 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31724 PR rtl-optimization/80193
31725 * ira.c (ira): Do not check allocation for LRA.
31727 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31729 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31730 (nvptx_output_simt_exit): Declare.
31731 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31732 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31733 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31734 (nvptx_declare_function_name): ...here. Emit declaration of local
31735 memory space buffer for omp_simt_enter insn.
31736 (nvptx_output_unisimt_switch): New.
31737 (nvptx_output_softstack_switch): New.
31738 (nvptx_output_simt_enter): New.
31739 (nvptx_output_simt_exit): New.
31740 * config/nvptx/nvptx.h (struct machine_function): New fields
31741 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31742 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31743 (UNSPECV_SIMT_EXIT): Ditto.
31744 (omp_simt_enter_insn): New insn.
31745 (omp_simt_enter): New expansion.
31746 (omp_simt_exit): New insn.
31747 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31749 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31750 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31751 (expand_GOMP_SIMT_EXIT): New.
31752 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31753 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31754 (GOMP_SIMT_EXIT): Ditto.
31755 * target-insns.def (omp_simt_enter): New insn.
31756 (omp_simt_exit): Ditto.
31757 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31759 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31760 (lower_rec_input_clauses): Likewise.
31761 (lower_lastprivate_clauses): Handle SIMT privatization.
31763 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31764 (ompdevlow_adjust_simt_enter): New.
31765 (find_simtpriv_var_op): New.
31766 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31767 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31769 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31770 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31771 (copy_decl_for_dup_finish): Ditto.
31773 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31775 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31778 * config/i386/i386.c (ix86_option_override_internal): Always
31779 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31781 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31783 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31785 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31787 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31788 mark new edge's irreducible flag accordign to it.
31789 (vect_do_peeling): Check loop preheader edge's irreducible flag
31790 and pass it to function slpeel_add_loop_guard.
31792 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31794 PR tree-optimization/80218
31795 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31796 Update block frequencies and counts.
31798 2017-03-28 Richard Biener <rguenther@suse.de>
31800 PR tree-optimization/78644
31801 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31802 of a simplification result we may not use it at all.
31804 2017-03-28 Richard Biener <rguenther@suse.de>
31807 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31808 without arguments, generate default definition of a SSA name.
31810 2017-03-28 Richard Biener <rguenther@suse.de>
31812 PR middle-end/80222
31813 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31814 TYPE_REF_CAN_ALIAS_ALL references.
31815 * fold-const.c (fold_indirect_ref_1): Likewise.
31817 2017-03-28 Martin Liska <mliska@suse.cz>
31820 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31821 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31823 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31824 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31826 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31827 (EXTRA_SPECS): Define.
31828 (SUBTARGET_EXTRA_SPECS): Likewise.
31829 (SUBTARGET_CPP_SPEC): Likewise.
31830 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31831 SUBTARGET_EXTRA_SPECS.
31832 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31834 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31836 * config/arc/simdext.md (vst64_insn): Update pattern.
31837 (vld32wh_insn): Likewise.
31838 (vld32wl_insn): Likewise.
31839 (vld64_insn): Likewise.
31840 (vld32_insn): Likewise.
31842 2017-03-28 Marek Polacek <polacek@redhat.com>
31845 * fold-const.c (fold_comparison): Use protected_set_expr_location
31846 instead of SET_EXPR_LOCATION.
31848 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31850 * tree.c (add_expr): Avoid name lookup warning.
31852 2017-03-27 Jeff Law <law@redhat.com>
31854 PR tree-optimization/80216
31855 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31856 function name. Limit recursion depth.
31857 (record_temporary_equivalences): Corresponding changes.
31859 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31861 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31864 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31867 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31869 * cfgcleanup.c (reg_note_cfa_p): New array.
31870 (insns_have_identical_cfa_notes): New function.
31871 (old_insns_match_p): Don't cross-jump in between /f
31872 and non-/f instructions. If both i1 and i2 are frame related,
31873 verify all CFA notes, their order and content.
31875 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31878 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31879 HImode and SImode with zero extend to DImode to one insn.
31880 (bswap<mode>2_extenddi): Likewise.
31881 (bswapsi2_extenddi): Likewise.
31882 (bswaphi2_extendsi): Likewise.
31883 (bswaphi2): Combine bswap HImode and SImode into one insn.
31884 Separate memory insns from swapping register.
31885 (bswapsi2): Likewise.
31886 (bswap<mode>2): Likewise.
31887 (bswaphi2_internal): Delete, no longer used.
31888 (bswapsi2_internal): Likewise.
31889 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31890 store, and gpr<-gpr swap insns.
31891 (bswap<mode>2_store): Likewise.
31892 (bswaphi2_reg): Register only splitter, combine with the splitter.
31893 (bswaphi2 splitter): Likewise.
31894 (bswapsi2_reg): Likewise.
31895 (bswapsi2 splitter): Likewise.
31896 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31897 the insns into load, store, and register/register insns.
31898 (bswapdi2_ldbrx): Likewise.
31899 (bswapdi2_load): Likewise.
31900 (bswapdi2_store): Likewise.
31901 (bswapdi2_reg): Likewise.
31903 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31905 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31906 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31908 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31911 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31913 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31914 special handling for target option conflicts between dform
31915 options (-mpower9-dform, -mpower9-dform-vector,
31916 -mpower9-dform-scalar) and -mno-direct-move.
31918 2017-03-27 Richard Biener <rguenther@suse.de>
31920 PR tree-optimization/80181
31921 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31923 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31925 * config/arc/predicates.md (move_double_src_operand): Replace the
31926 call to move_double_src_operand with a call to address_operand.
31928 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31930 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31931 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31932 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31934 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31936 * config/arc/predicates.md (long_immediate_loadstore_operand):
31937 Consider scaled addresses cases.
31939 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31941 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31942 restored when in interrupt.
31943 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31944 doesn't have delay slot.
31946 2017-03-27 Richard Biener <rguenther@suse.de>
31949 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31950 inlined thunk clones.
31952 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31955 * asan.c (instrument_derefs): Copy over last operand from
31956 original COMPONENT_REF to the new COMPONENT_REF with
31957 DECL_BIT_FIELD_REPRESENTATIVE.
31958 * ubsan.c (instrument_object_size): Likewise.
31960 2017-03-27 Richard Biener <rguenther@suse.de>
31962 PR tree-optimization/80170
31963 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31964 sure DR/SCEV didnt fold in constants we do not see when looking
31965 at the reference base alignment.
31967 2017-03-27 Richard Biener <rguenther@suse.de>
31969 PR middle-end/80171
31970 * gimple-fold.c (fold_ctor_reference): Properly guard against
31971 NULL return value from canonicalize_constructor_val.
31973 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31976 * config/i386/i386.c (ix86_expand_builtin)
31977 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31978 flags reg setting and flags reg using instructions.
31979 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31980 clobbering instructions to zero extend op2.
31982 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31984 * doc/install.texi (Configuration) <--with-aix-soname>:
31985 Update link to AIX ld.
31987 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31989 PR rtl-optimization/80160
31990 PR rtl-optimization/80159
31991 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31992 reg_alternate_class into account.
31994 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31997 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31998 to consider in curr_insn_transform.
32000 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32002 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32003 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32004 and emit_mode_inner.
32006 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32008 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32009 argument to the overloaded builtin variants. Use the new flag to
32010 deprecate certain builtin variants.
32011 * config/s390/s390-builtin-types.def: Add new builtin types.
32012 * config/s390/s390-builtins.h: Support new flags field for
32013 overloaded builtins.
32014 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32015 (s390_macro_to_expand): Enable vector float data type.
32016 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32017 builtins by incrementing the __VEC__ version number.
32018 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32020 (s390_resolve_overloaded_builtin): Emit error messages depending
32021 on the builtin flags.
32022 * config/s390/s390.c (s390_expand_builtin): Support additional
32023 flags argument. Change error message to match the messages
32024 emitted in s390-c.c.
32025 * config/s390/s390.md: New UNSPEC_* constants.
32026 (op_type): Add new instruction types.
32027 * config/s390/vecintrin.h: Add new builtins and test data class
32029 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32030 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32031 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32032 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32033 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32034 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32036 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32037 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32038 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32039 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32041 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32042 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32043 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32044 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32046 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32047 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32048 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32049 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32050 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32051 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32052 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32054 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32055 ("vec_scatter_element<V_HW_4:mode>_DI")
32056 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32057 ("vec_fpint<mode>", "vflls")
32058 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32059 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32060 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32061 ("*vec_cmphe<mode>_cc"): ... these.
32063 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32064 mode constant instead of magic value.
32066 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32068 * config/s390/s390.c (s390_expand_vec_compare): Support other
32069 vector floating point modes than just V2DF.
32070 (s390_expand_vcond): Likewise.
32071 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32072 (s390_cannot_change_mode_class): Prevent mode changes between TF
32073 and V1TF in vector registers.
32074 * config/s390/s390.md (DF, SF): New mode attributes.
32075 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32076 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32077 SFmode support for VRs.
32078 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32080 (VFT, VF_HW): New mode iterators.
32081 (vw, sdx): New mode attributes.
32082 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32083 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32084 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32085 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32086 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32087 also the new vector floating point modes. Renaming to ...
32089 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32090 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32091 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32092 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32093 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32094 ("vec_unordered<mode>"): ... these.
32096 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32097 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32098 ("*vec_extendv2df"): New insn definitions.
32100 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32102 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32103 ("mulditi3_2", "*muldi3_sign"): New patterns.
32104 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32105 rename the pattern definition.
32107 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32109 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32111 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32113 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32115 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32116 instruction if possible.
32117 * config/s390/vector.md (vec_halfnumelts): New mode
32119 ("*vec_vllezlf<mode>"): New pattern.
32121 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32123 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32124 ("popcountv4si2", "popcountv2di2"): Rename to ...
32125 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32126 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32128 ("popcount<mode>2_vxe"): New pattern.
32130 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32132 * common/config/s390/s390-common.c (processor_flags_table): Add
32134 * config.gcc: Add arch12.
32135 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32136 Default to arch12 for unknown CPU model numbers.
32137 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32138 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32139 PROCESSOR_max sanity check.
32140 * config/s390/s390-opts.h (enum processor_type): Add
32142 * config/s390/s390.c (processor_table): Add arch12.
32143 (s390_expand_builtin): Add check for B_VXE flag.
32144 (s390_issue_rate): Add PROCESSOR_ARCH12.
32145 (s390_get_sched_attrmask): Likewise.
32146 (s390_get_unit_mask): Likewise.
32147 (s390_sched_score): Enable z13 scheduling for arch12.
32148 (s390_sched_reorder): Likewise.
32149 (s390_sched_variable_issue): Likewise.
32150 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32152 (s390_tune_attr): Use z13 scheduling also for arch12.
32153 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32154 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32155 (TARGET_VXE_P): New macros.
32156 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32157 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32158 * config/s390/s390.opt: Add arch12 as processor_type.
32160 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32162 * config/s390/s390.md
32163 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32164 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32165 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32167 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32168 Rename expanders to ...
32170 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32171 ("fixuns_truncdddi2_emu"): ... these.
32173 ("fixuns_trunc<mode>si2_emu"): New expander.
32175 ("*fixuns_truncdfdi2_z13"): Rename to ...
32176 ("*fixuns_truncdfdi2_vx"): ... this.
32178 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32180 * config/s390/2964.md: Remove the single element vector compare
32181 instructions which are no longer used.
32182 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32184 (s390_canonicalize_comparison): Remove handling of DFmode
32186 (s390_expand_vec_compare_scalar): Remove function.
32187 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32188 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32190 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32192 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32194 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32196 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32197 will anyway by matched by mov<mode>_64dfp.
32199 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32201 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32202 vlef/vstef. Add missing operand to vleif.
32204 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32206 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32207 pair for all vector types with 64 bit elements.
32208 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32209 * config/s390/vector.md (V_HW_64): ... here.
32210 (V_128_NOSINGLE): New mode iterator.
32211 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32212 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32213 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32214 ("*vec_load_pairv2di"): Change to ...
32215 ("*vec_load_pair<mode>"): ... this one.
32217 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32219 * config/s390/constraints.md: Add comments.
32220 (jKK): Reject element sizes > 8 bytes.
32221 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32223 * config/s390/s390.md: Add the s_operand checks formerly in
32224 s390_split_ok_p to various splitters where they are still
32226 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32227 for 128 bit vectors. Plus two splitters.
32229 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32231 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32234 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32237 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32238 error if the boundary argument is not constant.
32240 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32242 PR rtl-optimization/80112
32243 * loop-doloop.c (doloop_condition_get): Don't check condition
32244 if cmp isn't SET with IF_THEN_ELSE src.
32246 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32248 PR tree-optimization/80158
32249 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32250 replacing a candidate statement, also replace it for the
32251 candidate's alternate interpretation.
32252 (replace_rhs_if_not_dup): Likewise.
32253 (replace_one_candidate): Likewise.
32255 2017-03-24 Richard Biener <rguenther@suse.de>
32257 PR tree-optimization/80167
32258 * graphite-isl-ast-to-gimple.c
32259 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32261 (translate_isl_ast_to_gimple::get_rename): Likewise.
32263 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32265 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32266 handling of certain combinations of target options, including the
32267 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32268 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32270 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32273 * config/arm/arm.md (*load_multiple): Add reload_completed to
32274 matching condition.
32276 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32277 Richard Biener <rguenth@suse.de>
32279 PR tree-optimization/79908
32280 PR tree-optimization/80136
32281 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32282 been cast away, gimplify_and_add suffices.
32284 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32286 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32288 2017-03-23 Richard Biener <rguenther@suse.de>
32290 PR tree-optimization/80032
32291 * gimplify.c (gimple_push_cleanup): Forced unconditional
32292 cleanups still have to go to the conditional_cleanups
32295 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32297 PR tree-optimization/80072
32298 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32300 (next_operand_entry_id): Change type to unsigned int.
32301 (sort_by_operand_rank): Make sure to return the right return value
32302 even if unsigned fields are bigger than INT_MAX.
32303 (struct oecount): Change cnt and id type to unsigned int.
32304 (oecount_hasher::equal): Formatting fix.
32305 (oecount_cmp): Make sure to return the right return value
32306 even if unsigned fields are bigger than INT_MAX.
32307 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32310 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32311 TREE_READONLY on result if writing it more than once.
32314 * doc/invoke.texi (-fsanitize=thread): Document that with
32315 -fnon-call-exceptions atomics are not able to throw
32319 * tsan.c: Include tree-eh.h.
32320 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32321 maybe_clean_or_replace_eh_stmt where needed.
32322 (instrument_memory_accesses): Add cfg_changed argument.
32323 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32324 if it returned true.
32325 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32327 PR rtl-optimization/63191
32328 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32329 wrapper function, moved the whole old content into ...
32330 (ix86_delegitimize_address_1): ... this. New inline function.
32331 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32332 true as last argument instead of ix86_delegitimize_address.
32334 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32336 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32337 cortexa57_branch_cost.
32339 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32341 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32343 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32346 * doc/md.texi (Constraints): Document wA constraint.
32347 * config/rs6000/constraints.md (wA): New.
32348 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32349 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32350 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32351 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32353 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32356 * gimplify.c (is_oacc_declared): New function.
32357 (oacc_default_clause): Use it to set default flags for acc declared
32358 variables inside parallel regions.
32359 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32360 declared variables.
32361 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32362 declare attribute to any decl as necessary.
32364 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32367 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32368 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32369 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32370 (arm_arch_lpae): This.
32371 * config/arm/arm.c (arm_arch7ve): Rename into ...
32372 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32373 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32376 2017-03-22 Martin Liska <mliska@suse.cz>
32379 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32380 error message instead of an ICE.
32382 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32384 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32386 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32388 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32390 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32393 2017-03-21 Martin Sebor <msebor@redhat.com>
32395 * doc/extend.texi: Use "cannot" instead of "can't."
32396 * doc/hostconfig.texi: Same.
32397 * doc/install.texi: Same.
32398 * doc/invoke.texi: Same.
32399 * doc/loop.texi: Same.
32400 * doc/md.texi: Same.
32401 * doc/objc.texi: Same.
32402 * doc/rtl.texi: Same.
32403 * doc/tm.texi: Same.
32404 * doc/tm.texi.in: Same.
32405 * doc/trouble.texi: Same.
32407 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32410 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32411 (collect_checksum_attributes): Set it.
32412 (die_checksum_ordered): Use it.
32414 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32416 PR tree-optimization/79908
32417 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32418 change: For a VA_ARG whose LHS has been cast away, use
32419 force_gimple_operand to construct the side effects.
32421 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32423 PR translation/80001
32424 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32425 more amenable to translation.
32426 (oacc_loop_auto_partitions): Likewise.
32428 2017-03-21 Marek Polacek <polacek@redhat.com>
32429 Martin Sebor <msebor@redhat.com>
32431 PR tree-optimization/80109
32432 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32433 on INTEGRAL_TYPE_P.
32435 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32436 Segher Boessenkool <segher@kernel.crashing.org>
32439 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32440 check reg_used_between_p between insn and one of succ or succ2
32441 depending on if succ is artificial insn not inserted into insn
32444 2017-03-21 Martin Liska <mliska@suse.cz>
32446 PR gcov-profile/80081
32447 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32448 * doc/gcc.texi: Include gcov-dump stuff.
32449 * doc/gcov-dump.texi: New file.
32451 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32453 PR rtl-optimization/79150
32454 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32455 conditional jump, if the jump is the last insn of the loop.
32457 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32458 Richard Biener <rguenth@suse.de>
32460 PR tree-optimization/79908
32461 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32462 been cast away, use force_gimple_operand to construct the side
32465 2017-03-21 Martin Liska <mliska@suse.cz>
32467 PR libfortran/79956
32468 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32471 2017-03-21 Brad Spengler <spender@grsecurity.net>
32474 * plugin.c (htab_hash_plugin): New function.
32475 (add_new_plugin): Use it and adjust.
32476 (parse_plugin_arg_opt): Adjust.
32477 (init_one_plugin): Likewise.
32479 2017-03-21 Richard Biener <rguenther@suse.de>
32481 PR tree-optimization/80032
32482 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32483 if set force the cleanup to happen unconditionally.
32484 (gimplify_target_expr): Push inserted clobbers with force_uncond
32485 to avoid them being removed by control-dependent DCE.
32487 2017-03-21 Richard Biener <rguenther@suse.de>
32489 PR tree-optimization/80122
32490 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32491 va_arg_pack_len when the inlined call stmt requires pack
32493 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32495 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32498 * tsan.c (instrument_builtin_call): If the memory model argument
32499 is not a constant, assume it is valid.
32502 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32505 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32507 PR rtl-optimization/79910
32508 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32509 if its dest is used by an insn before I2 (other than the combined
32510 insns themselves, which are properly handled already).
32512 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32515 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32517 * combine.c (record_used_regs): New static function.
32518 (try_combine): Handle situations where there is an additional
32519 instruction between I2 and I3 which needs to have a LOG_LINK
32523 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32525 * combine.c (try_combine): Delete redundant i1 test. Call
32526 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32528 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32531 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32532 alternatives 13/14.
32534 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32536 PR tree-optimization/80054
32537 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32538 the optimization if a PHI or any of its arguments is not dominated
32539 by the candidate's basis. Use gphi* rather than gimple* as
32541 (replace_profitable_candidates): Clean up a gimple* variable that
32542 should be a gphi* variable.
32544 2017-03-20 Martin Sebor <msebor@redhat.com>
32547 * doc/extend.texi (attribute constructor): Document present limitation.
32549 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32552 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32553 __POWER9_VECTOR__ #ifdef control, change template definition to
32554 use Power9-specific built-in function.
32555 (vec_any_eq): Likewise.
32556 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32557 to control outcomes from this test.
32558 (vector_ae_<mode>p): For VEC_F modes, likewise.
32560 2017-03-20 Ian Lance Taylor <iant@google.com>
32562 * config/i386/i386.c (ix86_function_regparm): Save an extra
32563 register for -fsplit-stack with DECL_STATIC_CHAIN.
32565 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32568 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32569 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32571 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32573 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32575 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32578 2017-03-20 Marek Polacek <polacek@redhat.com>
32581 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32583 2017-03-20 Richard Biener <rguenther@suse.de>
32585 PR tree-optimization/80113
32586 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32587 allocate extra SSA name for PHI def.
32588 (add_close_phis_to_outer_loops): Likewise.
32589 (add_close_phis_to_merge_points): Likewise.
32590 (copy_loop_close_phi_args): Likewise.
32591 (copy_cond_phi_nodes): Likewise.
32593 2017-03-20 Martin Liska <mliska@suse.cz>
32595 PR middle-end/79753
32596 * tree-chkp.c (chkp_build_returned_bound): Do not build
32597 returned bounds for a LHS that's not a BOUNDED_P type.
32599 2017-03-20 Martin Liska <mliska@suse.cz>
32603 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32604 COMPLEX_CST and VECTOR_CST.
32606 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32609 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32610 target operand. A new splitter adds the clobber statement in case
32611 the target operand is dead anyway.
32613 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32615 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32616 to age-old versions of binutils and glibc.
32618 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32620 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32622 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32624 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32626 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32628 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32629 requirement for binutils 2.13.
32631 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32633 * combine.c (try_combine): Delete redundant i1 test. Call
32634 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32636 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32638 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32639 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32641 <riscv64-*-elf>: Re-arrange section
32642 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32643 <riscv32-*-linux>: Likewise.
32644 <riscv64-*-elf>: Likewise
32645 <riscv64-*-linux>: Likewise.
32647 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32650 * aarch64.opt(verbose-cost-dump): Fix typo.
32652 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32655 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32656 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32658 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32660 * reload.c (find_reloads): When reloading a nonoffsettable address,
32661 use RELOAD_OTHER for it and its address reloads.
32663 PR rtl-optimization/79910
32664 * combine.c (record_used_regs): New static function.
32665 (try_combine): Handle situations where there is an additional
32666 instruction between I2 and I3 which needs to have a LOG_LINK
32669 2017-03-17 Jeff Law <law@redhat.com>
32671 PR tree-optimization/71437
32672 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32673 conditional in the hash table first.
32674 (vrp_dom_walker::before_dom_children): Extract condition from
32675 ASSERT_EXPR. Record condition, its inverion and any implied
32676 conditions as well.
32678 2017-03-17 Marek Polacek <polacek@redhat.com>
32679 Markus Trippelsdorf <markus@trippelsdorf.de>
32681 PR tree-optimization/80079
32682 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32685 2017-03-17 Richard Biener <rguenther@suse.de>
32687 PR middle-end/80075
32688 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32689 Properly verify the LHS before the RHS possibly claims to be
32691 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32694 2017-03-17 Martin Jambor <mjambor@suse.cz>
32696 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32697 (List of -O2 options): Likewise.
32698 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32701 2017-03-17 Tom de Vries <tom@codesourcery.com>
32703 * gcov-dump.c (print_usage): Print bug_report_url.
32705 2017-03-17 Richard Biener <rguenther@suse.de>
32707 PR middle-end/80050
32708 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32709 (parser::peek): Likewise.
32711 2017-03-17 Richard Biener <rguenther@suse.de>
32713 PR tree-optimization/80048
32714 * sese.c (free_sese_info): Properly release rename_map and
32715 copied_bb_map elements.
32717 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32719 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32720 Add linked-list forward and backlinks. Insert on
32721 construction, remove on destruction.
32722 (class pass_store_merging): Add m_stores_head field.
32723 (pass_store_merging::terminate_and_process_all_chains):
32724 Iterate over m_stores_head list.
32725 (pass_store_merging::terminate_all_aliasing_chains):
32727 (pass_store_merging::execute): Check for debug stmts first.
32728 Push new chains onto the m_stores_head stack.
32730 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32733 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32734 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32735 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32737 2017-03-16 Jeff Law <law@redhat.com>
32739 PR tree-optimization/71437
32740 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32741 member function. Implementation moved into after_dom_children
32742 member function and into the threader's thread_outgoing_edges
32744 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32745 some code into new thread_outgoing_edges.
32746 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32747 definition. Simplify marker handling (do it here). Assume we always
32748 have the available expression and the const/copies tables.
32749 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32751 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32752 * tree-vrp.c (equiv_stack): No longer file scoped.
32753 (vrp_dom_walker): New class.
32754 (vrp_dom_walker::before_dom_children): New member function.
32755 (vrp_dom_walker::after_dom_children): Likewise.
32756 (identify_jump_threads): Setup domwalker. Use it rather than
32757 walking edges in a random order by hand. Simplify setup/finalization.
32758 (finalize_jump_threads): Remove.
32759 (vrp_finalize): Do not call identify_jump_threads here.
32760 (execute_vrp): Do it here instead and call thread_through_all_blocks
32763 PR tree-optimization/71437
32764 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32766 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32768 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32769 (dom_opt_dom_walker::thread_across_edge): Remove
32770 handle_dominating_asserts argument. All callers changed.
32771 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32772 changes. Remove calls to lhs_of_dominating_assert. Other
32773 uses of handle_dominating_asserts turn into unconditional code
32774 (simplify_control_stmt_condition_1): Likewise.
32775 (simplify_control_stmt_condition): Likewise.
32776 (thread_through_normal_block, thread_across_edge): Likewise.
32777 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32778 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32779 object if it is not an SSA_NAME.
32780 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32781 before calling into the VRP specific simplifiers.
32782 (identify_jump_threads): Remove handle_dominating_asserts
32785 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32788 * tree-diagnostic.c (default_tree_printer): No longer static.
32789 * tree-diagnostic.h (default_tree_printer): New prototype.
32791 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32793 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32794 Change ins into fmov.
32796 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32798 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32799 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32800 Use h_con constraint for operand 1.
32801 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32802 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32804 2017-03-15 Jeff Law <law@redhat.com>
32806 PR tree-optimization/71437
32807 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32808 (record_temporary_equivalences): Use it.
32810 PR tree-optimization/71437
32811 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32812 tree-ssa-scopedtables.
32813 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32814 (record_conditions, record_cond, vuse_eq): Likewise.
32815 (record_edge_info): Adjust to API tweak of record_conditions.
32816 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32817 (record_temporary_equivalences, optimize_stmt): Likewise.
32818 (eliminate_redundant_computations): Likewise.
32819 (record_equivalences_from_stmt): Likewise.
32820 * tree-ssa-scopedtables.c: Include options.h and params.h.
32821 (vuse_eq): New function, moved from tree-ssa-dom.c
32822 (build_and_record_new_cond): Likewise.
32823 (record_conditions): Likewise. Accept vector of conditions rather
32824 than edge_equivalence structure for first argument.
32825 for the first argument.
32826 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32827 from tree-ssa-dom.c.
32828 (avail_exprs_stack::record_cond): Likewise.
32829 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32830 from tree-ssa-dom.c.
32831 (avail_exprs_stack): Add new member functions lookup_avail_expr
32833 (record_conditions): Declare.
32835 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32838 * lra-constraints.c (process_alt_operands): Increase reject for
32839 reloading an input/output operand.
32841 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32844 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32845 insns to convert from signed/unsigned char/short to IEEE 128-bit
32847 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32849 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32852 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32853 subreg of inner mode for values already in registers.
32855 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32857 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32858 iteration reg is used after the loop.
32860 2017-03-14 Martin Sebor <msebor@redhat.com>
32862 PR tree-optimization/79800
32863 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32864 precision in negative-positive range.
32865 (format_floating): Call non-const overload with adjusted precision.
32867 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32870 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32873 2017-03-14 Martin Sebor <msebor@redhat.com>
32875 PR middle-end/80020
32876 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32877 * builtins.def (aligned_alloc): Use it.
32880 * Makefile.in (GTFILES): Add calls.c.
32881 * calls.c: Include "gt-calls.h".
32883 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32885 PR rtl-optimization/79728
32886 * regs.h (struct target_regs): New field
32887 x_contains_allocatable_regs_of_mode.
32888 (contains_allocatable_regs_of_mode): New macro.
32889 * reginfo.c (init_reg_sets_1): Initialize it, and change
32890 contains_reg_of_mode so it includes global regs as well.
32891 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32892 rather than contains_regs_of_mode.
32894 2017-03-14 Martin Liska <mliska@suse.cz>
32896 * doc/invoke.texi: Document options that can't be combined with
32897 -fcheck-pointer-bounds.
32899 2017-03-14 Martin Liska <mliska@suse.cz>
32901 PR middle-end/79831
32902 * doc/invoke.texi (-Wchkp): Document the option.
32904 2017-03-14 Martin Liska <mliska@suse.cz>
32906 * Makefile.in: Install gcov-dump.
32908 2017-03-14 Martin Liska <mliska@suse.cz>
32910 * multiple_target.c (expand_target_clones): Bail out for
32911 an invalid attribute.
32913 2017-03-14 Richard Biener <rguenther@suse.de>
32915 * alias.c (struct alias_set_entry): Pack properly.
32916 * cfgloop.h (struct loop): Likewise.
32917 * cse.c (struct set): Likewise.
32918 * ipa-utils.c (struct searchc_env): Likewise.
32919 * loop-invariant.c (struct invariant): Likewise.
32920 * lra-remat.c (struct cand): Likewise.
32921 * recog.c (struct change_t): Likewise.
32922 * rtl.h (struct address_info): Likewise.
32923 * symbol-summary.h (function_summary): Likewise.
32924 * tree-loop-distribution.c (struct partition): Likewise.
32925 * tree-object-size.c (struct object_size_info): Likewise.
32926 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32927 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32928 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32929 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32930 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32931 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32932 (struct _stmt_vec_info): Likewise.
32934 2017-03-14 Martin Liska <mliska@suse.cz>
32937 * multiple_target.c (create_dispatcher_calls): Check that
32938 a target can create a function dispatcher.
32940 2017-03-14 Martin Liska <mliska@suse.cz>
32943 * multiple_target.c (expand_target_clones): Drop local.local
32944 flag for default implementation.
32946 2017-03-14 Richard Biener <rguenther@suse.de>
32948 PR tree-optimization/80030
32949 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32951 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32953 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32954 gcc_fallthrough() instead of __attribute__((fallthrough));
32956 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32958 * doc/gcc.texi: Remove "up" link to (DIR).
32959 * doc/gccint.texi: Ditto.
32961 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32963 * doc/install.texi (Specific) <avr>: Remove reference to
32966 2017-03-13 Jeff Law <law@redhat.com>
32968 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32969 attribute rather than comments.
32971 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32972 match_scratch operand is highest.
32974 2017-03-13 Martin Liska <mliska@suse.cz>
32976 PR middle-end/78339
32977 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32978 is a CHKP clone, use original declaration.
32980 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32982 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32983 (arc_conditional_register_usage): Use a different allocation order
32984 when optimizing for size.
32985 * common/config/arc/arc-common.c (arc_option_optimization_table):
32986 Section anchors default on when optimizing for size.
32988 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32990 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32992 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32994 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32995 * config/arc/arc.md (cpu_facility): Add cd variant.
32996 (*movqi_insn): Add code density variant.
32997 (*movhi_insn): Likewise.
32998 (*movqi_insn): Likewise.
32999 (*addsi3_mixed): Likewise.
33000 (subsi3_insn): Likewise.
33002 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33004 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33006 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33008 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33009 expressions with MINUS and UNARY ops.
33011 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33014 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33016 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33017 between vec_select and vector argument.
33018 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33019 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33020 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33021 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33022 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33023 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33025 2017-03-13 Richard Biener <rguenther@suse.de>
33028 * params.def (vect-max-peeling-for-alignment): Fix typo.
33030 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33032 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33033 issue that only occurred with binutils below 2.18.
33035 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33037 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33038 refer to binutils 2.11/2.12 minimum.
33040 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33042 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33043 ftp.kernel.org and simplify binutils requirement.
33045 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33047 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33049 (Optimize Options): Ditto. Also remove redundancy.
33051 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33053 PR translation/79848
33054 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33056 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33057 to G_ to avoid double translation.
33059 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33061 PR translation/79923
33062 * auto-profile.c (get_combined_location): Convert leading
33063 character of diagnostics to lower case and remove trailing period.
33064 (read_profile): Likewise for various diagnostics.
33065 * config/arm/arm.c (arm_option_override): Remove trailing period
33066 from various diagnostics.
33067 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33068 (msp430_expand_delay_cycles): Likewise.
33070 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33073 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33074 full command-line argument, rather than just "str".
33075 (aarch64_validate_march): Likewise.
33076 (aarch64_validate_mtune): Likewise.
33078 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33080 PR rtl-optimization/78911
33081 * lra-assigns.c (must_not_spill_p): New function.
33082 (spill_for): Use it.
33084 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33086 PR tree-optimization/79981
33087 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33088 ATOMIC_COMPARE_EXCHANGE ifn result.
33089 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33090 IFN_ATOMIC_COMPARE_EXCHANGE.
33092 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33095 * opts.c (parse_sanitizer_options): Add missing question mark to
33096 "did you mean" message.
33098 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33100 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33102 (VMULEUH_UNS): Likewise.
33103 (VMULOUB_UNS): Likewise.
33104 (VMULOUH_UNS): Likewise.
33105 * config/rs6000/rs6000.c (builtin_function_type): Remove
33106 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33108 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33111 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33112 x with result of extra_parsing_for_operand_code_0.
33113 (function_reader::extra_parsing_for_operand_code_0): Convert
33114 return type from void to rtx, returning x. When reading
33115 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33116 larger size containing struct block_symbol.
33118 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33120 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33121 -mfloat128-hardware without -m64.
33123 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33126 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33127 entries to the case statement that marks unsigned arguments to
33128 overloaded functions.
33130 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33132 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33133 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33135 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33138 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33139 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33141 2017-03-10 Martin Liska <mliska@suse.cz>
33145 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33146 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33149 2017-03-10 Olivier Hainque <hainque@adacore.com>
33151 * tree-switch-conversion (array_value_type): Start by resetting
33152 candidate type to it's main variant.
33154 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33156 PR rtl-optimization/79909
33157 * combine.c (try_combine): Use simplify_replace_rtx on individual
33158 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33159 of the whole CALL_INSN_FUNCTION_USAGE.
33161 PR tree-optimization/79972
33162 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33163 get_range_info on SSA_NAMEs. Formatting fixes.
33165 2017-03-10 Richard Biener <rguenther@suse.de>
33166 Jakub Jelinek <jakub@redhat.com>
33168 PR tree-optimization/77975
33169 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33170 edge to be constant.
33171 (get_val_for): For constant x return it. Formatting fix.
33172 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33173 would use the same bases as the current one.
33175 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33177 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33178 instead of vec_select for V1TImode.
33179 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33181 (VSX_LE_128): Add V1TI to this mode iterator.
33182 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33183 (*vsx_le_perm_store_<mode>): Likewise.
33184 (pre-reload splitter for VSX stores): Likewise.
33185 (post-reload splitter for VSX stores): Likewise.
33186 (*vsx_xxpermdi2_le_<mode>): Likewise.
33187 (*vsx_lxvd2x2_le_<mode>): Likewise.
33188 (*vsx_stxvd2x2_le_<mode>): Likewise.
33190 2017-03-09 Michael Eager <eager@eagercon.com>
33192 Correct failures with --enable-checking=yes,rtl.
33194 * config/microblaze/microblaze.c (microblaze_expand_shift):
33195 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33196 test for const0_rtx.
33197 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33198 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33200 2017-03-09 Richard Biener <rguenther@suse.de>
33202 PR tree-optimization/79977
33203 * graphite-scop-detection.c (scop_detection::merge_sese):
33204 Handle the case of extra exits to blocks dominating the entry.
33206 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33208 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33211 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33213 PR rtl-optimization/79949
33214 * lra-constraints.c (process_alt_operands): Check memory when
33215 trying to predict a cycle. Print about the overall increase.
33217 2017-03-09 Richard Biener <rguenther@suse.de>
33219 PR middle-end/79971
33220 * gimple-expr.c (useless_type_conversion_p): Preserve
33221 TYPE_SATURATING for fixed-point types.
33223 2017-03-09 Richard Biener <rguenther@suse.de>
33226 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33227 alignment of BLKmode params.
33229 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33232 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33233 (VALL_NO_V2Q): Likewise.
33235 * config/aarch64/aarch64-simd.md
33236 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33238 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33239 VALL_NO_V2Q mode iterator.
33240 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33242 2017-03-09 Martin Liska <mliska@suse.cz>
33244 PR tree-optimization/79631
33245 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33246 tree_int_cst_sign_bit just for INTEGER constants.
33248 2017-03-09 Martin Liska <mliska@suse.cz>
33252 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33255 2017-03-09 Marek Polacek <polacek@redhat.com>
33258 * tree.c (inchash::add_expr): Handle TREE_VEC.
33260 2017-03-09 Martin Liska <mliska@suse.cz>
33263 (chkp_narrow_size_and_offset): New function.
33264 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33265 (void chkp_parse_bit_field_ref): New function.
33266 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33267 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33269 2017-03-09 Martin Liska <mliska@suse.cz>
33272 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33273 (chkp_find_bounds_1): Remove gcc_unreachable.
33275 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33278 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33279 BUILT_IN_SYNC*, determine the access type from the size suffix and
33280 always build a MEM_REF with that type. Handle forgotten
33281 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33284 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33285 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33286 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33287 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33288 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33289 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33290 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33291 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33292 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33293 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33294 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33295 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33296 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33297 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33298 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33299 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33300 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33301 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33302 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33303 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33304 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33305 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33306 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33307 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33308 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33309 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33310 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33311 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33312 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33313 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33314 definitions outside of __OPTIMIZE__ guarded section.
33317 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33318 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33319 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33320 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33323 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33325 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33326 ("vfenez<mode>"): Add missing constraints.
33328 2017-03-08 Martin Sebor <msebor@redhat.com>
33331 * config/nds32/nds32.c (nds32_option_override):
33332 Fix misspelled diagnostic.
33334 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33337 * gimplify.c (gimplify_omp_for): Replace index var in outer
33338 taskloop statement with an artificial variable and add
33339 OMP_CLAUSE_PRIVATE clause for it.
33341 2017-03-08 Richard Biener <rguenther@suse.de>
33343 PR tree-optimization/79955
33344 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33345 for accesses that are completely outside of the variable.
33347 2017-03-08 Andrew Haley <aph@redhat.com>
33349 PR tree-optimization/79943
33350 * tree-ssa-loop-split.c (compute_new_first_bound): When
33351 calculating the new upper bound, (END-BEG) should be added, not
33354 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33356 * config/avr/avr.md (setmemhi): Make sure match_dup
33357 operand number comes before match_scratch.
33359 2017-03-08 Richard Biener <rguenther@suse.de>
33361 PR tree-optimization/79920
33362 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33363 with ncopies == 1 to ...
33364 (vect_transform_slp_perm_load): ... here. Properly compute
33365 all element loads by iterating VF times over the group. Do
33366 not handle ncopies (computed in a broken way) in
33367 vect_create_mask_and_perm.
33369 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33372 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33373 is a uniform vector, use uniform_vector_p return value instead of
33374 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33376 2017-03-07 Marek Polacek <polacek@redhat.com>
33378 PR middle-end/79809
33379 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33380 (alloca_call_type): Likewise.
33382 2017-03-07 Martin Liska <mliska@suse.cz>
33384 * gcov.c (process_args): Put comment to correct location.
33386 2017-03-07 Martin Liska <mliska@suse.cz>
33388 PR middle-end/68270
33389 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33390 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33391 (chkp_narrow_bounds_for_field): Likewise.
33392 (chkp_parse_array_and_component_ref): Pass one more argument to
33395 2017-03-07 Richard Biener <rguenther@suse.de>
33397 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33400 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33402 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33403 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33405 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33408 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33409 to end of description.
33410 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33412 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33414 PR rtl-optimization/79901
33415 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33417 (*avx512f_<code><mode>3<mask_name>): ... this.
33418 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33419 iterator instead of VI8_AVX2_AVX512BW.
33421 PR rtl-optimization/79901
33422 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33423 min/max expander, expand it using expand_vec_cond_expr.
33426 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33429 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33432 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33433 to void * for PCH reasons.
33434 * dwarf2out.c (output_loc_operands, output_die): Cast
33435 v.val_vec.array to unsigned char *.
33437 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33440 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33443 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33445 PR rtl-optimization/79571
33446 * lra-constraints.c (process_alt_operands): Calculate static
33447 reject and subtract it from overall when only addresses will be
33450 2017-03-06 Julia Koval <julia.koval@intel.com>
33453 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33454 incoming stack boundary to 128 for 64-bit targets.
33456 2017-03-06 Richard Biener <rguenther@suse.de>
33458 PR tree-optimization/79894
33459 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33460 to NULL after folding it.
33462 2017-03-06 Richard Biener <rguenther@suse.de>
33464 PR tree-optimization/79824
33465 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33466 check disabling peeling for gaps.
33468 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33470 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33471 attributes): Document gettimeofday.
33473 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33475 * config/s390/s390.c (s390_option_override_internal): Set
33476 PARAM_MIN_VECT_LOOP_BOUND
33478 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33480 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33481 * config/s390/s390.md: Likewise.
33483 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33486 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33487 (<avx2_avx512>_perm<mode>): Rename to ...
33488 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33490 (<avx512>_perm<mode>_mask): Rename to ...
33491 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33493 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33494 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33495 instead of VI8F_256_512.
33496 (avx512f_perm<mode>): New define_expand.
33497 (avx512f_perm<mode>_mask): Likewise.
33498 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33499 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33501 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33503 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33504 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33506 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33508 2017-03-06 Martin Liska <mliska@suse.cz>
33511 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33512 when having a SSA NAME w/o VAR_DECL assigned to it.
33514 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33516 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33517 msa_dpsub_<su>_d): Fix MODE for vec_select.
33519 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33521 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33523 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33525 2017-03-06 Richard Biener <rguenther@suse.de>
33527 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33528 * plugin.c (register_plugin_info): Likewise.
33529 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33531 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33533 * config/i386/sse.md (sse_storehps, sse_storelps,
33534 avx_<castmode><avxsizesuffix>_<castmode>,
33535 avx512f_<castmode><avxsizesuffix>_<castmode>,
33536 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33537 in condition that at least one operand is not a MEM.
33539 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33541 PR middle-end/79805
33542 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33543 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33545 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33546 gimple_call_nothrow_p flag based on whether original builtin can throw.
33547 If it can, emit following stmts on the fallthrough edge.
33548 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33549 don't create new bb if inserting just debug stmts on the edge, try to
33550 insert them on the fallthru bb or just reset debug stmts.
33552 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33555 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33556 restore recog_data (including the operand rtxes inside it) around
33557 the call to get_insn_template.
33559 2017-03-03 Martin Sebor <msebor@redhat.com>
33561 PR tree-optimization/79699
33562 * context.c (context::~context): Free MPFR caches to avoid
33563 a memory leak on program exit.
33565 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33567 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33568 Use wide_int::ulow () instead of .elt (0).
33570 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33572 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33573 (*pushxf): Limit oF constraint to 32bit targets and add oC
33574 constraint for 64bit targets.
33575 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33576 (*pushdf): Change rmF constraint to rmC.
33578 2017-03-03 Martin Liska <mliska@suse.cz>
33580 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33581 Remove unused variable.
33583 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33586 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33587 is a memory operand, increase num_memory.
33588 (ix86_expand_args_builtin): Likewise.
33590 2017-03-03 Jan Hubicka <jh@suse.cz>
33593 * ipa-devirt.c (maybe_record_node): Properly handle
33594 __cxa_pure_virtual visibility.
33596 2017-03-03 Martin Liska <mliska@suse.cz>
33598 PR tree-optimization/79803
33599 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33601 (pass_loop_prefetch::execute): Disabled optimization if an
33602 assumption about L1 cache size is not met.
33604 2017-03-03 Martin Liska <mliska@suse.cz>
33606 PR rtl-optimization/79574
33607 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33608 (hash_scan_set): Likewise.
33609 (dump_hash_table): Likewise.
33610 (hoist_code): Likewise.
33612 2017-03-03 Richard Biener <rguenther@suse.de>
33614 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33615 in place of uhigh ().
33616 (fixed_convert_from_real): Likewise.
33618 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33621 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33623 2017-03-03 Richard Biener <rguenther@suse.de>
33625 PR middle-end/79818
33626 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33627 TYPE_OVERFLOW_UNDEFINED check.
33629 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33631 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33633 (vector_ae_<mode>_p): Likewise.
33634 (vector_nez_<mode>_p): Likewise.
33635 (vector_ne_v2di_p): Likewise.
33636 (vector_ae_v2di_p): Likewise.
33637 (vector_ne_<mode>_p): Likewise.
33638 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33640 (vsx_tsqrt<mode>2_fe): Likewise.
33642 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33645 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33647 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33649 PR rtl-optimization/79780
33650 * cprop.c (one_cprop_pass): When second and further conditional trap
33651 in a single basic block is turned into an unconditional trap, turn it
33652 into a deleted note to avoid RTL verification failures.
33654 2017-03-02 Richard Biener <rguenther@suse.de>
33656 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33658 2017-03-02 Richard Biener <rguenther@suse.de>
33660 PR tree-optimization/79345
33662 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33663 param and abort the walk, returning -1 if it is hit.
33664 (walk_aliased_vdefs): Take a limit param and pass it on.
33665 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33666 defaulting to 0 and return a signed int.
33667 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33668 (check_defs): New helper.
33669 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33670 about uninitialized memory.
33671 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33672 bogus uninitialized warning.
33673 (fixed_convert_from_real): Likewise.
33675 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33677 PR tree-optimization/66768
33678 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33679 iv_use if base object can't be determined.
33681 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33683 PR tree-optimization/79345
33684 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33685 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33686 (get_pattern_stats): Initialize it.
33687 * genemit.c (gen_expand): Verify match_scratch numbers come after
33688 match_operand/match_dup numbers.
33689 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33690 match_scratch numbers.
33691 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33693 * config/s390/s390.md (trunctdsd2): Likewise.
33695 2017-03-02 Richard Biener <rguenther@suse.de>
33697 * wide-int.h (wide_int_storage::operator=): Implement in terms
33700 2017-03-02 Richard Biener <rguenther@suse.de>
33702 PR tree-optimization/79777
33703 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33704 the to insert expression to sth existing.
33706 2017-03-01 Martin Sebor <msebor@redhat.com>
33708 PR middle-end/79692
33709 * gimple-ssa-sprintf.c
33710 (directive::known_width_and_precision): New function.
33711 (format_integer): Use it.
33712 (get_mpfr_format_length): Consider the full range of precision
33713 when computing %g output with the # flag. Set the likely byte
33714 count to 3 rather than 1 when precision is indeterminate.
33715 (format_floating): Correct the lower bound of precision.
33717 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33719 * doc/invoke.texi: Document default code model for 64-bit Linux.
33721 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33724 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33725 udiv rather than div since input pattern is unsigned.
33727 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33729 * config/i386/i386.c (print_reg): Warn for values of
33730 unsupported size in integer register.
33732 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33735 * config/rs6000/predicates.md (current_file_function_operand): Do
33736 not allow self calls to be local if the function is replaceable.
33738 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33741 * config/rs6000/altivec.h (vec_ctz and others): Change the
33742 preprocessor macro that controls conditional compilation from
33743 _ARCH_PWR9 to __POWER9_VECTOR__.
33744 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33745 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33746 control (instead of _ARCH_PWR9 control) so that template
33747 definition uses power9-specific function.
33748 (vec_any_eq): Likewise.
33749 (vec_all_ne): Change macro definition to use a power9-specific
33750 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33751 _ARCH_PWR9 control).
33752 (vec_any_eq) Likewise.
33753 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33754 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33755 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33756 support for xvcmpnedp instruction.
33757 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33758 macro expansion so that Power9 implementation of vec_all_ne does
33759 not use the AltiVec predicate framework.
33760 (VCMPNEH_P): Likewise.
33761 (VCMPNEW_P): Likewise.
33762 (VCMPNED_P): Likewise.
33763 (VCMPNEFP_P): Likewise.
33764 (VCMPNEDP_P): Likewise.
33765 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33766 implementation of vec_any_eq to not use AltiVec predicate
33768 (VCMPAEH_P): Likewise.
33769 (VCMPAEW_P): Likewise.
33770 (VCMPAED_P): Likewise.
33771 (VCMPAEFP_P): Likewise.
33772 (VCMPAEDP_P): Likewise.
33773 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33774 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33775 not use the AltiVec predicate framework.
33776 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33777 of vec_any_eq to not use AltiVec predicate framework.
33778 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33779 support for predefined __POWER9_VECTOR__ macro to indicate that
33780 Power9 instruction selection is enabled.
33781 (altivec_overloaded_builtins): Remove extraneous
33782 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33783 function argument types RS6000_BTI_bool_V16QI and
33784 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33785 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33786 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33787 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33788 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33789 Power9 for implementations of vec_cmpne. Change the signature for
33790 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33791 (representing vec_all_ne) to remove the previously described first
33792 argument of type RS6000_BTI_INTSI, as this was an artifact of
33793 reliance on the AltiVec predicate framework, which is no longer
33794 used in the implementation of these functions. Add
33795 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33796 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33797 since, unlike the AltiVec predicate framework implementation, we
33798 do not share function descriptors between vec_alle and vec_anyeq.
33799 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33800 set of modes that receive special treatment even when
33801 TARGET_P9_VECTOR is true. The special treatment emits code that
33802 does not depend on Power9 instructions.
33803 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33804 define_expand to not rely on AltiVec predicate framework.
33805 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33807 (vector_ne_v2di_p): Change this define_expand to not rely on
33808 AltiVec predicate framework.
33809 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33811 (vector_ne_<mode>_p): Change this define_expand to not rely on
33812 AltiVec predicate framework.
33813 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33815 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33816 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33817 define_insn pattern.
33818 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33819 define_insn pattern because the xvcmpne<VSs>. instruction is not
33821 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33822 instruction is not supported.
33824 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33826 * config/nvptx/nvptx.c: Include intl.h.
33828 2017-03-01 Martin Jambor <mjambor@suse.cz>
33831 * ipa-prop.h (ipa_bits): Removed field known.
33832 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33833 to pointers. Adjusted their comments to warn about their sharing.
33834 (ipcp_transformation_summary): Change bits to a vector of pointers.
33835 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33836 (ipa_get_ipa_bits_for_value): Declare.
33837 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33838 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33839 (ipa_bits_hash_table): Likewise.
33840 (ipa_vr_ggc_hash_traits): Likewise.
33841 (ipa_vr_hash_table): Likewise.
33842 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33843 being pointers and vr_known being removed.
33844 (ipa_set_jf_unknown): Likewise.
33845 (ipa_get_ipa_bits_for_value): New function.
33846 (ipa_set_jfunc_bits): Likewise.
33847 (ipa_get_value_range): New overloaded functions.
33848 (ipa_set_jfunc_vr): Likewise.
33849 (ipa_compute_jump_functions_for_edge): Use the above functions to
33850 construct bits and vr parts of jump functions.
33851 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33852 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33854 (ipcp_grow_transformations_if_necessary): Also allocate
33855 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33857 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33858 them. Fix too long lines.
33859 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33860 vr_known being removed.
33861 (ipa_read_jump_function): Use new setter functions to construct bits
33862 and vr parts of jump functions or set them to NULL.
33863 (write_ipcp_transformation_info): Adjust for bits being pointers.
33864 (read_ipcp_transformation_info): Likewise.
33865 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33867 Include gt-ipa-prop.h.
33868 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33870 (ipcp_store_bits_results): Likewise.
33871 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33872 Do not write to existing jump functions but use a temporary instead.
33874 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33877 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33878 attempt to use its first operand as BIT_FIELD_REF base.
33880 2017-03-01 Richard Biener <rguenther@suse.de>
33882 PR middle-end/79721
33883 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33884 interpolating formula in wrapping arithmetic.
33885 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33887 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33889 PR tree-optimization/79734
33890 * tree-vect-generic.c (expand_vector_condition): Optimize
33891 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33892 Handle VEC_COND_EXPR where comparison has different inner width from
33893 type's inner width.
33895 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33897 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33898 markup, and similar issues. Remove @opindex entries for things
33899 that aren't options. Add missing -mmpy-option entries.
33901 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33903 PR tree-optimization/79737
33904 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33905 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33906 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33907 instead of byte_size. Formatting fix.
33908 (shift_bytes_in_array_right): Formatting fix.
33910 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33913 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33914 condition on optimize for the leaf function test.
33916 2017-02-28 Martin Liska <mliska@suse.cz>
33919 * read-rtl-function.c (function_reader::handle_unknown_directive):
33920 Bail out when one uses -flto.
33922 2017-02-28 Martin Liska <mliska@suse.cz>
33924 * common.opt: Replace space with tabular for options of <number>
33926 * config/i386/i386.opt: Show <number> value for
33927 -mlarge-data-threshold.
33928 * opts.c (print_filtered_help): Do not display number in hexadecimal
33931 2017-02-28 Martin Liska <mliska@suse.cz>
33933 * common.opt: Fix --help=option -Q for options which are of
33936 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33938 * config/i386/i386.c (print_reg): Error out for values
33939 of 8-bit size in invalid integer register.
33941 2017-02-28 Martin Sebor <msebor@redhat.com>
33943 PR tree-optimization/79691
33944 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33946 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33949 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33950 gcc_unreachable with output_operand_lossage.
33952 2017-02-28 Richard Biener <rguenther@suse.de>
33954 PR tree-optimization/79740
33955 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33957 (visit_nary_op): Insert the nary into the hashtable if we
33958 pattern-matched sth.
33959 * tree-ssa-pre.c (eliminate_insert): Robustify.
33961 2017-02-28 Richard Biener <rguenther@suse.de>
33963 PR middle-end/79731
33964 * fold-const.c (decode_field_reference): Reject out-of-bound
33967 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33969 * config/i386/i386.c: Include intl.h.
33970 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33971 instead of just cond ? "..." : "...".
33972 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33973 * coverage.c (read_counts_file): Likewise.
33974 * omp-offload.c: Include intl.h.
33975 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33976 of just cond ? "..." : "...".
33977 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33978 of just cond ? "..." : "...".
33980 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33983 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33985 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33986 'tune for' CPU name.
33987 * config/arm/arm-cpu-data.h: Regenerated.
33989 2017-02-28 Richard Biener <rguenther@suse.de>
33991 PR tree-optimization/79732
33992 * tree-inline.c (expand_call_inline): Do not shadow var.
33994 2017-02-28 Richard Biener <rguenther@suse.de>
33996 PR tree-optimization/79723
33997 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33998 address-space properly.
34000 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34002 * doc/optinfo.texi (Optimization groups): Fix option used for
34004 * doc/invoke.texi (-fopt-info): Document "omp".
34005 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34006 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34007 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34008 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34009 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34011 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34013 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34014 OPTGROUP_OMP with "omp".
34016 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34019 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34020 for arithmetic shift of unsigned V2DI.
34022 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34024 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34025 arc/linux.h headers.
34026 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34027 (LINK_SPEC): Likewise.
34028 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34029 (EXTRA_SPECS): Likewise.
34030 (STARTFILE_SPEC): Likewise.
34031 (ENDFILE_SPEC): Likewise.
34032 (LIB_SPEC): Likewise.
34033 (TARGET_SDATA_DEFAULT): Likewise.
34034 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34035 (MULTILIB_DEFAULTS): Likewise.
34036 (DWARF2_UNWIND_INFO): Likewise.
34037 * config/arc/big.h: New file.
34038 * config/arc/elf.h: Likewise.
34039 * config/arc/linux.h: Likewise.
34040 * config/arc/t-uClibc: Remove.
34042 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34044 PR tree-optimization/77536
34045 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34046 (tree_transform_and_unroll_loop): Use above function to compute the
34047 estimated niter of unrolled loop and use it when scaling profile.
34048 Also use count info rather than frequency if it's non-zero.
34049 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34050 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34051 (vect_transform_loop): Call above function.
34053 2017-02-27 Richard Biener <rguenther@suse.de>
34055 PR tree-optimization/45397
34056 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34057 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34058 (visit_nary_op): Add pattern matching for CSEing sign-changed
34059 or truncated operations with wider ones.
34061 2017-02-27 Richard Biener <rguenther@suse.de>
34063 PR tree-optimization/79690
34064 * tree-vect-stmts.c (vectorizable_store): Use vector type
34065 built from the DR with address-space.
34067 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34069 * doc/invoke.texi (Optimize Options): Refine the description
34070 of asan-use-after-return.
34072 2017-02-25 Alan Modra <amodra@gmail.com>
34074 PR rtl-optimization/79584
34075 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34076 base, not ad->base_term, the reg within base. Remove assertion
34077 that ad->base == ad->base_term. Replace gen_int_mode using
34078 bogus mode with const0_rtx.
34080 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34082 PR middle-end/79396
34083 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34084 FMA_EXPR like tcc_binary or tcc_unary.
34086 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34089 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34091 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34092 (output_loc_operands): Handle DW_OP_call_ref and
34093 DW_OP_GNU_variable_value.
34094 (struct variable_value_struct): New type.
34095 (struct variable_value_hasher): Likewise.
34096 (variable_value_hash): New variable.
34097 (string_types): Remove.
34098 (copy_loc_descr): New function.
34099 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34100 (prepend_loc_descr_to_each): New function.
34101 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34102 instead of add_loc_descr_to_each if the first argument is single
34103 location list and the second has multiple.
34104 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34105 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34106 when looking for variable value which doesn't have other location info.
34107 (loc_list_from_tree): Formatting fix.
34108 (gen_array_type_die): Simplify DW_AT_string_length handling.
34109 (adjust_string_types): Remove.
34110 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34111 string_types. Call resolve_variable_values.
34112 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34113 (resolve_addr_in_expr): Likewise. Add A argument.
34114 (copy_deref_exprloc): Remove deref argument. Adjust for the
34115 original expression being DW_OP_GNU_variable_value with optionally
34116 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34117 optionally after it.
34118 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34119 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34120 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34121 (variable_value_hasher::hash, variable_value_hasher::equal): New
34123 (resolve_variable_value_in_expr, resolve_variable_value,
34124 resolve_variable_values, note_variable_value_in_expr,
34125 note_variable_value): New functions.
34126 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34129 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34132 * opts.h (handle_generated_option): Add GENERATED_P argument.
34133 * opts-common.c (handle_option): Adjust function comment.
34134 (handle_generated_option): Add GENERATED_P argument, pass it to
34136 (control_warning_option): Pass false to handle_generated_option
34138 * opts.c (maybe_default_option): Pass true to handle_generated_option
34140 * optc-gen.awk: Likewise.
34142 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34144 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34145 a REG, look at the REG it is a SUBREG of.
34146 (splitter for cmpeqsi_t): Ditto.
34148 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34150 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34151 the special USEs with the pattern of the insn, not the insn itself.
34153 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34156 * doc/invoke.texi: Document -mload-store-pairs.
34158 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34159 Sandra Loosemore <sandra@codesourcery.com>
34161 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34162 argument isn't a CONST_INT.
34163 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34164 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34165 (nios2_validate_compare): Bypass alternate compare logic if *op2
34166 is not a CONST_INT.
34167 (ldstwm_operation_p): Return false if first_base is not a REG or
34168 if first_offset is not a CONST_INT.
34170 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34172 * config/cris/cris.md: Use correct operand in a define_peephole2.
34174 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34176 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34178 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34180 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34181 this_insn if it is an INSN or JUMP_INSN.
34182 (force_offsettable): Look at base, not at addr.
34183 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34184 on things that aren't necessarily CONST_INTs.
34186 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34188 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34189 -mfpmath=sse is the default also for x86-32 targets with SSE2
34190 instruction set when @option{-ffast-math} is enabled
34192 2017-02-24 Jeff Law <law@redhat.com>
34194 PR rtl-optimizatoin/79286
34195 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34198 2017-02-24 Richard Biener <rguenther@suse.de>
34200 PR tree-optimization/79389
34201 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34204 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34206 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34207 function comment to reflect reality.
34208 (loop_exits_before_overflow): Fix typo in function description.
34210 2017-02-24 Richard Biener <rguenther@suse.de>
34212 PR tree-optimization/79389
34213 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34214 properly that a threading opportunity exists. Detect conditional
34215 copy/constant propagation opportunities.
34217 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34219 * config/visium/visium.md (type): Add trap.
34220 (b): New mode attribute.
34221 (*btst): Rename into...
34222 (*btst<mode>): ...this and adjust.
34223 (*cbranchsi4_btst_insn): Rename into...
34224 (*cbranch<mode>4_btst_insn): ...this and adjust.
34225 (trap): New define_insn.
34227 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34229 PR tree-optimization/79389
34230 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34231 (noce_reversed_cond_code): New function.
34232 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34233 reversed_comparison_code. Formatting fix.
34234 (noce_try_store_flag): Test rev_cond != NULL in addition to
34235 reversed_comparison_code.
34236 (noce_try_store_flag_constants): Likewise.
34237 (noce_try_store_flag_mask): Likewise.
34238 (noce_try_addcc): Use rev_cond if non-NULL instead of
34239 reversed_comparison_code.
34240 (noce_try_cmove_arith): Likewise. Formatting fixes.
34241 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34242 (noce_find_if_block): Initialize rev_cond.
34243 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34244 instead of false as last argument never attempt to reverse it
34247 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34249 PR tree-optimization/79663
34250 * tree-predcom.c (combine_chains): Process refs in reverse order
34251 only for ZERO length chains, and add explaining comment.
34253 2017-02-23 Jeff Law <law@redhat.com>
34255 PR tree-optimization/79578
34256 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34257 in call to operand_equal_p.
34259 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34262 * config/i386/cpuid.h: Fix another undefined behavior.
34264 2017-02-23 Richard Biener <rguenther@suse.de>
34266 PR tree-optimization/79683
34267 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34268 vector types for data-refs.
34270 2017-02-23 Martin Liska <mliska@suse.cz>
34272 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34274 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34276 PR middle-end/79665
34277 * internal-fn.c (get_range_pos_neg): Moved to ...
34278 * tree.c (get_range_pos_neg): ... here. No longer static.
34279 * tree.h (get_range_pos_neg): New prototype.
34280 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34281 are known to be in between 0 and signed maximum inclusive, try to
34282 expand both unsigned and signed divmod and use the cheaper one from
34285 2017-02-22 Jeff Law <law@redhat.com>
34287 PR tree-optimization/79578
34288 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34289 to compare base operands.
34291 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34294 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34295 gpc_reg_operand instead of fpr_reg_operand.
34297 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34299 * config/mips/mips.c (mips_return_in_memory): Force FP
34300 vector types to be returned in memory for o32 ABI.
34302 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34304 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34305 instead of DW_TAG_member for static data member declarations and don't
34306 set no_linkage_name for static inline data members.
34307 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34310 2017-02-22 Martin Liska <mliska@suse.cz>
34312 * doc/invoke.texi: Replace inequality signs with square brackets
34315 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34318 * lra-constraints.c (simplify_operand_subreg): Handle
34319 WORD_REGISTER_OPERATIONS targets.
34321 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34324 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34325 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34326 elimination by swapping fld*.
34328 2017-02-22 Richard Biener <rguenther@suse.de>
34330 PR tree-optimization/79673
34331 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34332 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34333 irrelevant address-space qualifiers and avoiding a
34334 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34336 2017-02-22 Richard Biener <rguenther@suse.de>
34338 PR tree-optimization/79666
34339 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34340 to not symbolically negate if that may introduce undefined
34343 2017-02-22 Martin Liska <mliska@suse.cz>
34346 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34347 * data-streamer-out.c (streamer_write_gcov_count_stream):
34349 * value-prof.c (stream_out_histogram_value): Make assert more
34350 precise based on type of counter.
34352 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34355 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34356 Use nonimmediate_operand instead of memory_operand for operand 1.
34357 (float-extend standard_x87sse_constant_load splitter): Ditto.
34359 2017-02-21 Jeff Law <law@redhat.com>
34361 PR tree-optimization/79621
34362 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34363 blocks with edges to themselves.
34365 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34368 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34369 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34370 Use gimple_call_builtin_p.
34373 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34374 on temporarily removed DEBUG_INSNs.
34376 PR tree-optimization/79649
34377 * tree-loop-distribution.c (classify_partition): Give up on
34378 non-generic address space loads/stores.
34380 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34382 * doc/loop.texi (Loop manipulation): Remove nonexistent
34383 tree_ssa_loop_version from the documentation.
34384 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34386 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34389 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34390 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34391 * config/rs6000/rs6000.c: Include except.h.
34392 (rs6000_expand_split_stack_prologue): Call
34393 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34395 2017-02-21 Martin Jambor <mjambor@suse.cz>
34398 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34399 have been analyzed.
34401 2017-02-21 Martin Jambor <mjambor@suse.cz>
34403 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34404 for backward compatibility only.
34405 * doc/invoke.texi (Option Summary): Remove all references to
34406 -fipa-cp-alignment.
34408 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34412 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34414 * lra-constraints.c (curr_insn_transform): Handle
34415 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34417 2017-02-21 Martin Liska <mliska@suse.cz>
34419 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34421 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34423 PR translation/79638
34424 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34426 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34429 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34430 (arm_function_ok_for_sibcall): Return false for an indirect call by
34431 descriptor if all the argument registers are used.
34432 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34433 alignment of the function.
34435 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34437 PR tree-optimization/61441
34438 * simplify-rtx.c (simplify_const_unary_operation): For
34439 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34440 the sNaN unmodified.
34442 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34444 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34445 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34446 instead of SYSTEM_HEADER_DIR.
34448 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34449 Martin Liška <mliska@suse.cz>
34451 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34452 Fix typos and grammar, use active voice, and clarify.
34454 2017-02-20 Marek Polacek <polacek@redhat.com>
34456 PR middle-end/79537
34457 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34460 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34462 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34465 * config/i386/i386.c (ix86_expand_builtin): Handle
34466 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34467 ix86_builtins_isa[fcode].isa as a requirement of those
34468 flags and any other flag in the bitmask.
34469 (ix86_init_mmx_sse_builtins): Use 0 instead of
34470 ~OPTION_MASK_ISA_64BIT as mask.
34471 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34472 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34473 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34474 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34476 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34479 * lra-constraints.c (split_reg): Check requested split mode
34480 is supported by the register.
34482 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34484 * lra-constraints.c (simplify_operand_subreg): Remove early
34487 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34490 * lra-constraints.c (curr_insn_transform): Tighten condition
34491 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34493 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34496 * lra-constraints.c (curr_insn_transform): Handle
34497 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34499 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34502 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34504 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34506 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34509 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34512 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34514 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34517 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34519 * final.c (last_columnnum, override_columnnum): New variables.
34520 (final_start_function): Set last_columnnum, pass it to begin_prologue
34521 hook and pass 0 to dwarf2out_begin_prologue.
34522 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34523 to source_line debug hook.
34524 (notice_source_line): Compute last_columnnum and for debug_column_info
34525 return true on column changes.
34526 * debug.h (struct gcc_debug_hooks): Add column argument to
34527 source_line and begin_prologue hooks.
34528 (debug_nothing_int_charstar_int_bool): Remove prototype.
34529 (debug_nothing_int_int_charstar,
34530 debug_nothing_int_int_charstar_int_bool): New prototypes.
34531 (dwarf2out_begin_prologue): Add column argument.
34532 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34533 begin_prologue hooks.
34534 (debug_nothing_int_charstar_int_bool): Remove.
34535 (debug_nothing_int_int_charstar,
34536 debug_nothing_int_int_charstar_int_bool): New functions.
34537 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34538 through to dwarf2out_source_line.
34539 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34540 (dwarf2out_source_line): Add column argument, emit it if requested.
34541 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34543 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34544 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34545 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34546 through to dwarf2out_begin_prologue.
34547 (vmsdbgout_source_line): Add column argument, pass it through to
34548 dwarf2out_source_line.
34549 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34550 dbxout_source_line caller.
34551 (dbxout_source_line): Add column argument.
34553 * common.opt (gno-column-info, gcolumn-info): New options.
34554 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34555 (check_die): Also test for multiple DW_AT_decl_column attributes.
34556 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34557 DW_AT_decl_column if requested.
34558 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34560 (gen_variable_die): Likewise.
34561 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34562 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34565 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34566 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34567 (ix86_handle_option): Handle OPT_m3dnowa.
34568 * doc/invoke.texi (-m3dnowa): Document.
34569 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34570 -m3dnowa instead of -m3dnow -march=athlon.
34573 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34574 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34576 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34579 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34580 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34581 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34582 generator for vsx_xxpermdi_<mode>_be.
34583 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34584 force big-endian semantics.
34585 (vsx_xxpermdi_<mode>_be): New define_expand with same
34586 implementation as previous version of vsx_xxpermdi_<mode>.
34588 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34590 PR tree-optimization/79327
34591 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34592 variable, its initialization and use.
34594 2017-02-17 Julia Koval <julia.koval@intel.com>
34596 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34597 (OPTION_MASK_ISA_PKU_UNSET): New.
34598 (ix86_handle_option): Handle -mrdpid.
34599 * config/i386/cpuid.h (bit_RDPID): New.
34600 * config/i386/driver-i386.c (host_detect_local_cpu):
34601 Detect RDPID feature.
34602 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34603 * config/i386/i386-c.c (ix86_target_macros_internal):
34605 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34606 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34607 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34608 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34609 * config/i386/i386.md (define_insn "rdpid"): New.
34610 * config/i386/i386.opt Add -mrdpid.
34611 * config/i386/immintrin.h (_rdpid_u32): New.
34613 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34615 PR rtl-optimization/79541
34616 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34617 instead of transforming it into USE.
34619 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34621 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34622 If HONOR_SNANS (SFmode) force the input to a register.
34623 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34624 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34625 an frsp or similar insn.
34627 2017-02-17 Martin Liska <mliska@suse.cz>
34629 PR rtl-optimization/79577
34630 * params.def (selsched-max-sched-times): Increase minimum to 1.
34632 2017-02-17 Martin Liska <mliska@suse.cz>
34634 PR rtl-optimization/79574
34635 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34637 2017-02-17 Martin Liska <mliska@suse.cz>
34639 PR tree-optimization/79529
34640 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34641 ssa_defined_default_def_p to handle cases which are implicitly
34643 * tree-ssa.c (ssa_defined_default_def_p): New function.
34644 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34645 which are implicitly defined.
34646 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34648 2017-02-17 Richard Biener <rguenther@suse.de>
34650 PR middle-end/79576
34651 * params.def (max-ssa-name-query-depth): Limit to 10.
34653 2017-02-17 Richard Biener <rguenther@suse.de>
34655 PR tree-optimization/79552
34656 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34659 2017-02-17 Richard Biener <rguenther@suse.de>
34662 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34664 2017-02-17 Marek Polacek <polacek@redhat.com>
34666 PR middle-end/79536
34667 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34668 (fold_negate_expr): New wrapper.
34670 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34672 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34673 Correct terminology and de-emphasize pre-standard behavior.
34675 2017-02-16 Alan Modra <amodra@gmail.com>
34677 PR rtl-optimization/79286
34678 * ira.c (def_dominates_uses): New function.
34679 (update_equiv_regs): Don't create an equivalence for insns that
34680 may trap where the register def does not dominate the use.
34682 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34684 PR rtl-optimization/78127
34685 * lra.c (lra): Call lra_eliminate before finish the loop after
34688 2017-02-16 Richard Biener <rguenther@suse.de>
34690 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34692 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34693 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34694 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34695 (isl_val_int_from_wi): New function.
34696 (extract_affine_gmp): Rename to ...
34697 (extract_affine_wi): ... this, take a widest_int.
34698 (extract_affine_int): Just wrap extract_affine_wi.
34699 (add_param_constraints): Use isl_val_int_from_wi.
34700 (add_loop_constraints): Likewise, and extract_affine_wi.
34702 2017-02-15 Jeff Law <law@redhat.com>
34704 PR middle-end/79521
34705 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34706 ira_init_register_move_cost_if_necessary.
34708 2017-02-15 Martin Sebor <msebor@redhat.com>
34710 PR middle-end/32003
34711 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34712 removed in a prior commit.
34714 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34716 PR tree-optimization/79347
34717 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34718 counters during peeling.
34720 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34722 * Makefile.in (site.exp): Remove "set ISLVER".
34724 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34727 * real.c (real_from_integer): Call real_convert even for decimal.
34729 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34732 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34734 2017-02-14 Andrew Pinski <apinski@cavium.com>
34736 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34737 cores and change the partno/implementer to be correct.
34738 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34739 the 'B" as the implementer.
34740 * config/aarch64/aarch64-tune.md: Regenerate.
34742 2017-02-14 Carl Love <cel@us.ibm.com>
34744 * config/rs6000/rs6000.c: Add case statement entry to make the
34745 xvcvuxdsp built-in argument unsigned.
34746 * config/rs6000/vsx.md: Fix the source and return operand types so they
34747 match the instruction definitions from the ISA document. Fix typo
34748 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34751 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34754 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34755 member early_clobber_alts.
34756 * lra-lives.c (reg_early_clobber_p): New.
34757 (process_bb_lives): Use it.
34758 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34759 (debug_operand_data): Initialize early_clobber_alts.
34760 (setup_operand_alternative): Set up early_clobber_alts.
34761 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34762 alternatives to new_insn_reg.
34763 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34765 (lra_update_insn_regno_info): Pass the new arg.
34767 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34769 PR middle-end/79505
34770 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34771 (new_oacc_loop_raw): Don't clear already cleared fields.
34774 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34775 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34776 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34778 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34781 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34783 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34786 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34787 the extra instruction to the right place to store 128-bit constant
34790 2017-02-14 Martin Sebor <msebor@redhat.com>
34792 PR middle-end/79448
34793 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34794 warning for strings of unknown length.
34796 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34798 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34800 2017-02-14 Jeff Law <law@redhat.com>
34803 * ira-costs.c (scan_one_insn): Initialize register move costs
34804 for pseudos seen in USE/CLOBBER insns.
34806 PR tree-optimization/79095
34807 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34808 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34809 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34810 if the operands are known to be not equal, then the resulting range
34812 (intersect_ranges): If the new range is ~[0,0] and the old range is
34813 wide, then prefer ~[0,0].
34814 * tree-vrp.c (overflow_comparison_p_1): New function.
34815 (overflow_comparison_p): New function.
34816 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34817 if NAME is used in an overflow test.
34818 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34819 overflow check that can be expressed as an equality test, then adjust
34820 ops to be that equality test.
34822 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34824 * config/s390/s390-builtin-types.def: Remove flags argument.
34825 * config/s390/s390.c (s390_init_builtins): Likewise.
34827 2017-02-14 Martin Liska <mliska@suse.cz>
34829 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34830 vector. Fix trailing white spaces.
34832 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34834 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34837 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34839 PR rtl-optimization/68664
34840 * config/arm/arm.c (arm_sched_can_speculate_insn):
34841 New function. Declare prototype.
34842 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34844 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34846 PR rtl-optimization/68664
34847 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34849 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34851 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34853 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34854 max skip bytes for function, loop and jump.
34856 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34858 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34859 ABS_EXPR for gimple dump.
34861 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34864 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34866 PR tree-optimization/79408
34867 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34868 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34869 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34870 also if rhs1 is INTEGER_CST.
34872 2017-02-14 Richard Biener <rguenther@suse.de>
34874 PR middle-end/79432
34875 * tree-into-ssa.c (insert_phi_nodes): When the function can
34876 have abnormal edges rewrite SSA names with broken use-def
34877 dominance out of SSA and register them for PHI insertion.
34879 2017-02-13 Martin Sebor <msebor@redhat.com>
34881 PR middle-end/79496
34882 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34883 clearing info.nowrite flag when snprintf size argument is a range.
34885 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34887 * cprop.c (cprop_jump): Add missing space in string literal.
34888 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34889 (get_constraint_for_component_ref): Likewise.
34890 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34891 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34892 * lra-constraints.c (process_alt_operands): Likewise.
34893 * ipa-inline.c (inline_small_functions): Likewise.
34894 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34895 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34896 * trans-mem.c (diagnose_tm_1_op): Likewise.
34897 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34898 (grid_parallel_clauses_gridifiable): Likewise.
34900 * config/nvptx/mkoffload.c (process): Add space in between
34903 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34904 "MOD4_SSE_REGS" and "ALL_REGS".
34906 * spellcheck.c (test_data): Add , in between "foo" and "food".
34908 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34911 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34912 boundary crossing check and subsequent code generation agree.
34914 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34916 * config/aarch64/aarch64.c (has_memory_op): Delete.
34917 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34920 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34922 PR rtl-optimization/79388
34923 PR rtl-optimization/79450
34924 * combine.c (distribute_notes): When removing TEM_INSN for which
34925 corresponding dest has last value recorded, invalidate that last
34928 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34930 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34931 of explicit '@'. Add missing assembly comment marker on branch costs
34934 2017-02-13 Nathan Sidwell <nathan@acm.org>
34936 * gengtype-lex.l (<in_struct>): Add '/'.
34938 2017-02-13 Martin Liska <mliska@suse.cz>
34941 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34943 2017-02-13 Richard Biener <rguenther@suse.de>
34945 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34947 * configure: Re-generate.
34948 * config.in: Likewise.
34949 * graphite-dependences.c: Simplify as if
34950 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34951 * graphite-isl-ast-to-gimple.c: Likewise.
34952 * graphite-optimize-isl.c: Likewise.
34953 * graphite-poly.c: Likewise.
34954 * graphite-sese-to-poly.c: Likewise.
34955 * graphite.h: Likewise.
34956 * toplev.c: Include isl/version.h and use isl_version () for
34957 printing the ISL version.
34958 * doc/install.texi: Update ISL requirement.
34960 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34962 * doc/standards.texi (Standards): Update reference to
34965 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34967 * doc/extend.texi (Named Address Spaces): sourceware.org now
34969 * doc/install.texi (Binaries): Ditto.
34972 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34974 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34975 standard terminology "stringize"/"stringizing" throughout.
34976 * doc/cppinternals.texi: Likewise.
34978 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34980 * doc/extend.texi: Fix some spelling mistakes and typos.
34981 * doc/invoke.texi: Likewise.
34983 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34986 * params.def (inline-min-speedup) Change from 10 to 8.
34988 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34990 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34993 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34996 * ipa-inline-analysis.c (get_minimal_bb): New function.
34997 (record_modified): Use it.
34998 (remap_edge_change_prob): Handle also ancestor functions.
35000 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35002 * doc/contrib.texi (Contributors): Remove broken link into
35003 the Mauve CVS repository.
35005 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35007 PR middle-end/79454
35008 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35009 result computation whenever lhs doesn't have vector mode, not
35010 just when it has BLKmode.
35012 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35014 * doc/makefile.texi (profiledbootstrap): Refer to the
35015 installation instructions only in textual form.
35017 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35020 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35022 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35024 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35025 (Specific): Update mingw-w64 reference.
35027 (Specific): Remove broken link to Renesas RX processor.
35029 2017-02-10 Richard Biener <rguenther@suse.de>
35031 * toplev.c (process_options): Do not mention obsolete graphite
35032 options when printing sorry message about missing graphite support.
35033 Mention -floop-nest-optimize.
35035 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35037 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35038 (vtst_p16): Likewise.
35039 (vtstq_p8): Likewise.
35040 (vtstq_p16): Likewise.
35042 (vtstq_p64): Likewise.
35043 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35044 (vset_lane_p64): New.
35045 (vsetq_lane_p64): New.
35047 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35049 PR tree-optimization/79411
35050 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35051 stmt operands are SSA_NAMEs used in abnormal phis.
35052 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35055 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35058 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35061 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35063 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35065 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35067 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35071 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35072 not arbitrary TREE_CONSTANT.
35075 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35076 "omp declare target link" attribute unless is_global_var.
35077 * omp-offload.c (find_link_var_op): Likewise.
35079 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35080 Chung-Lin Tang <cltang@codesourcery.com>
35082 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35084 (gimplify_adjust_omp_clauses): Don't delete TILE.
35085 (gimplify_omp_for): Deal with TILE.
35086 * internal-fn.c (expand_GOACC_TILE): New function.
35087 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35089 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35090 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35092 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35093 avoid DIV for outermost collapse var.
35094 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35095 Remove out of date comments, fix whitespace.
35096 * omp-general.c (omp_extract_for_data): Deal with tiling.
35097 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35098 adjust OLF_DIM_BASE value.
35099 (struct omp_for_data): Add tiling field.
35100 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35101 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35102 for auto loops. Remove default auto determining, moved to
35103 oacc_loop_fixed_partitions.
35104 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35105 stmts, add e_mask field.
35106 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35107 (oacc_thread_numbers): Use oacc_dim_call.
35108 (oacc_xform_tile): New.
35109 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35110 (finish_oacc_loop): Adjust for ifns vector.
35111 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35112 add case for GOACC_TILE fns.
35113 (oacc_loop_xform_loop): Delete.
35114 (oacc_loop_process): Iterate over call list directly, and add
35115 handling for GOACC_TILE fns.
35116 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35118 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35119 vector partitioning to outer loops. Assign 2 partitions to loops
35120 when available. Add TILE handling.
35121 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35122 (execite_oacc_device_lower): Process GOACC_TILE fns,
35123 ignore unknown specs.
35124 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35125 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35126 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35128 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35130 * configure.ac (ACX_BUGURL): Update.
35131 * configure: Regenerate.
35133 2017-02-09 Richard Biener <rguenther@suse.de>
35135 PR tree-optimization/69823
35136 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35137 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35139 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35141 * config/arc/arc-c.def: Add __NPS400__ definition.
35142 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35143 (TARGET_NPS400): Define.
35145 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35147 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35149 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35150 pointer, arch_info.
35151 (arc_cpu_types): Fill the arch_info field with a pointer into the
35152 arc_arch_types table.
35153 (arc_selected_cpu): Declare.
35154 * config/arc/arc.c (arc_selected_cpu): Make global.
35155 (arc_selected_arch): Delete.
35156 (arc_base_cpu): Delete.
35157 (arc_override_options): Remove references to deleted variables,
35158 update access to arch information.
35159 (ARC_OPT): Update access to arch information.
35160 (ARC_OPTX): Likewise.
35161 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35162 (TARGET_ARC600): Update access to arch information.
35163 (TARGET_ARC601): Likewise.
35164 (TARGET_ARC700): Likewise.
35165 (TARGET_EM): Likewise.
35166 (TARGET_HS): Likewise.
35167 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35170 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35173 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35174 condition/operands for integer GE/LE/GEU/LEU operations.
35176 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35178 PR translation/79397
35179 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35182 2017-02-08 Martin Jambor <mjambor@suse.cz>
35185 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35186 whether allocation happened.
35187 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35188 nothing was allocated.
35190 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35192 PR tree-optimization/79408
35193 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35194 constant, but SSA_NAME with a known integer range, use the minimum
35195 of that range instead of op1 to determine if modulo can be replaced
35196 with its first operand.
35198 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35200 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35202 2017-02-08 Richard Biener <rguenther@suse.de>
35204 PR tree-optimization/71824
35205 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35206 Check all loops contained in the merged region.
35208 2017-02-07 Andrew Pinski <apinski@cavium.com>
35210 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35212 2017-02-07 Andrew Pinski <apinski@cavium.com>
35214 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35215 (thunderxt88): Likewise.
35216 (thunderxt81): Disable LSE and change v8.1 to v8.
35217 (thunderxt83): Likewise.
35219 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35220 Richard Biener <rguenther@suse.de>
35222 PR middle-end/79399
35223 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35224 type from int to size_t.
35225 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35227 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35229 PR rtl-optimization/79386
35230 * cprop.c (bypass_conditional_jumps): Initialize
35231 bypass_last_basic_block already before splitting bbs after
35232 unconditional traps...
35233 (bypass_conditional_jumps): ... rather than here.
35236 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35237 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35238 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35239 fix -masm=intel patterns.
35241 2017-02-07 Richard Biener <rguenther@suse.de>
35243 PR tree-optimization/79256
35244 PR middle-end/79278
35245 * builtins.c (get_object_alignment_2): Use min_align_of_type
35246 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35247 and ADJUST_FIELD_ALIGN.
35249 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35251 * doc/tm.texi: Regenerate.
35252 * stor-layout.c (layout_decl): Adjust.
35253 (update_alignment_for_field): Likewise.
35254 (place_field): Likewise.
35255 (min_align_of_type): Likewise.
35256 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35257 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35258 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35259 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35260 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35261 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35262 * config/i386/i386.c (x86_field_alignment): Likewise.
35263 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35264 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35265 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35266 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35267 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35268 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35272 2017-01-30 Richard Biener <rguenther@suse.de>
35274 PR tree-optimization/79256
35275 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35276 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35279 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35281 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35282 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35283 builtins to SImode and emit a zero-extend, if necessary.
35285 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35287 * docs/invoke.texi (RISC-V Options): Alphabetize.
35289 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35291 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35294 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35296 * config/riscv/riscv.c: New file.
35297 * common/config/riscv/riscv-common.c: Likewise.
35298 * config.gcc: Likewise.
35299 * config/riscv/constraints.md: Likewise.
35300 * config/riscv/elf.h: Likewise.
35301 * config/riscv/generic.md: Likewise.
35302 * config/riscv/linux.h: Likewise.
35303 * config/riscv/multilib-generator: Likewise.
35304 * config/riscv/peephole.md: Likewise.
35305 * config/riscv/pic.md: Likewise.
35306 * config/riscv/predicates.md: Likewise.
35307 * config/riscv/riscv-builtins.c: Likewise.
35308 * config/riscv/riscv-c.c: Likewise.
35309 * config/riscv/riscv-ftypes.def: Likewise.
35310 * config/riscv/riscv-modes.def: Likewise.
35311 * config/riscv/riscv-opts.h: Likewise.
35312 * config/riscv/riscv-protos.h: Likewise.
35313 * config/riscv/riscv.h: Likewise.
35314 * config/riscv/riscv.md: Likewise.
35315 * config/riscv/riscv.opt: Likewise.
35316 * config/riscv/sync.md: Likewise.
35317 * config/riscv/t-elf-multilib: Likewise.
35318 * config/riscv/t-linux: Likewise.
35319 * config/riscv/t-linux-multilib: Likewise.
35320 * config/riscv/t-riscv: Likewise.
35321 * configure.ac: Likewise.
35322 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35323 Waterman as RISC-V maintainers.
35324 * doc/install.texi: Add RISC-V entries.
35325 * doc/invoke.texi: Add RISC-V options section.
35326 * doc/md.texi: Add RISC-V constraints section.
35327 * configure: Regenerated.
35329 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35332 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35333 false values to be constant vectors with all 0 or all 1 bits set.
35334 (vcondu<mode><mode>): Likewise.
35335 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35337 (fpmask_comparison_operator): Update comment.
35338 (vecint_comparison_operator): New predicate.
35339 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35340 vector conditionals when the true and false values are constant
35341 vectors with all 0 bits or all 1 bits set.
35343 2017-02-06 Martin Sebor <msebor@redhat.com>
35345 PR tree-optimization/79376
35346 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35348 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35350 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35351 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35352 to simplify split condition.
35354 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35356 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35357 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35360 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35362 PR rtl-optimization/68664
35363 * target.def (can_speculate_insn): New hook.
35364 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35365 * doc/tm.texi: Regenerate.
35366 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35367 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35368 (rs6000_sched_can_speculate_insn): New function.
35370 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35372 PR tree-optimization/79284
35373 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35374 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35375 vectorizable_mask_load_store, vectorizable_operation,
35376 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35377 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35378 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35379 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35380 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35381 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35382 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35383 is_gimple_assign (stmt). Replace another such test with
35384 is_gimple_assign (stmt).
35386 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35389 * config/avr/avr.c (rtl-iter.h): Include it.
35390 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35391 (avr_legitimate_combined_insn): ...and implementation.
35393 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35395 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35396 * config/s390/s390.c (s390_const_operand_ok)
35397 (s390_canonicalize_comparison, s390_extract_part)
35398 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35399 (s390_contiguous_bitmask_p, s390_rtx_costs)
35400 (legitimize_pic_address): Likewise.
35401 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35402 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35403 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35404 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35405 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35407 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35409 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35410 REGNO($0) == REGNO($1).
35412 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35414 * config/s390/linux.h(SIZE_TYPE): Add comment.
35416 2017-02-06 Julian Brown <julian@codesourcery.com>
35417 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35418 Virendra Pathak <virendra.pathak@broadcom.com>
35420 * config/aarch64/aarch64-cores.def: Change the scheduler
35422 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35423 * config/aarch64/thunderx2t99.md: New file.
35425 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35427 * doc/standards.texi (Go Language): Update link to language
35430 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35432 * tree-eh.c (lower_resx): Sanitize profile.
35433 (cleanup_empty_eh_move_lp): Likewise.
35435 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35438 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35440 * cfgloopmanip.h (loop_version): Update prototype.
35441 * modulo-sched.c (sms_schedule): Update call of loop_version.
35442 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35443 * tree-parloops.c (gen_parallel_loop): Likewise.
35444 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35445 * tree-ssa-loop-split.c (split_loop): Likewise.
35446 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35447 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35449 2017-02-05 Martin Liska <mliska@suse.cz>
35452 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35454 (print_operand_address): Initialize a struct to zero.
35456 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35458 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35459 garbage collector only in textual form.
35461 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35463 * doc/extend.texi (x86 specific memory model extensions for
35464 transactional memory): Simplify a phrase.
35466 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35469 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35470 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35471 (atomic_storedi_1): Likewise.
35473 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35475 PR tree-optimization/79338
35476 * tree-parloops.c (gather_scalar_reductions): Don't call
35477 vect_analyze_loop_form for loop->inner before destroying loop's
35480 2017-02-03 Martin Sebor <msebor@redhat.com>
35482 PR tree-optimization/79327
35483 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35484 when precision has resulted in leading zeros.
35485 (format_integer): Adjust the likely counter to assume an unknown
35486 argument that may be zero is non-zero.
35488 2017-02-03 Jason Merrill <jason@redhat.com>
35491 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35492 avoid copying non-taken branch.
35494 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35496 PR tree-optimization/79340
35497 * tree-vect-loop.c (vectorizable_reduction): Release
35498 vec_defs elements after safe_splicing them into other vectors.
35501 PR tree-optimization/79327
35502 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35503 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35505 (format_integer): Use wide_int_to_tree instead of build_int_cst
35506 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35507 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35508 of shortest and longest sequence.
35510 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35512 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35513 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35515 2017-02-03 Walter Lee <walt@tilera.com>
35518 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35519 after initial stackframe link reg save.
35520 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35522 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35525 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35526 wu for stxssp alternative.
35528 2017-02-03 Martin Sebor <msebor@redhat.com>
35530 PR tree-optimization/79352
35531 * gimple-fold.c (get_range_strlen): Add argument.
35532 (get_range_strlen): Change return type to bool.
35533 (get_maxval_strlen): Pass in a dummy argument.
35534 * gimple-fold.h (get_range_strlen): Change return type to bool.
35535 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35536 * tree.h (array_at_struct_end_p): Add argument.
35537 * tree.c (array_at_struct_end_p): Handle it.
35539 2017-02-03 Martin Liska <mliska@suse.cz>
35542 * multiple_target.c (create_dispatcher_calls): Redirect edge
35543 from a caller of a dispatcher.
35544 (expand_target_clones): Make the clones local.
35545 (ipa_target_clone): Do both target clones and resolvers.
35546 (ipa_dispatcher_calls): Remove the pass.
35547 (pass_dispatcher_calls::gate): Likewise.
35548 (make_pass_dispatcher_calls): Likewise.
35549 * passes.def (pass_target_clone): Put as very first IPA early
35552 2017-02-03 Martin Liska <mliska@suse.cz>
35554 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35555 in case of a function with ifunc attribute.
35557 2017-02-03 Martin Liska <mliska@suse.cz>
35559 * cgraph.c (cgraph_node::dump): Dump function version info.
35560 * symtab.c (symtab_node::dump_base): Add missing new line.
35562 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35564 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35565 (ifcombine_ifandif): Use it.
35567 2017-02-03 Martin Liska <mliska@suse.cz>
35569 * doc/invoke.texi: Document default value for
35570 use-after-scope-direct-emission-threshold.
35572 2017-02-03 Martin Liska <mliska@suse.cz>
35574 PR tree-optimization/79339
35575 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35576 (format_floating): Likewise.
35578 2017-02-03 Martin Liska <mliska@suse.cz>
35581 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35583 (ipa_node_params_t::remove): Likewise.
35584 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35585 initialization from removed ipa_node_params_t::insert.
35586 (ipa_node_params::~ipa_node_params): Move from removed
35587 ipa_node_params_t::release.
35588 * symbol-summary.h (symbol_summary::m_released): New member.
35589 Do not release a summary twice. Do not allow to call finalizer
35590 for types of a summary that live in GGC memory.
35592 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35594 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35597 2017-02-02 Martin Sebor <msebor@redhat.com>
35599 PR middle-end/79275
35600 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35601 (format_string): Tighten up the range of output for non-constant
35602 strings and correct the expected range for wide non-constant strings.
35604 2017-02-02 Martin Sebor <msebor@redhat.com>
35606 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35608 PR middle-end/32003
35609 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35611 (-fdump-tree-@var): Add to index and document how to come up
35612 with pass-specific option and dump file names.
35613 (-fdump-passes): Clarify where to look for output.
35615 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35617 PR middle-end/77445
35618 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35619 statistics of the analyzed path; allow threading for speed when
35620 any of BBs along the path are optimized for speed.
35622 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35624 PR middle-end/78468
35625 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35626 settings of the virtual registers.
35629 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35631 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35632 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35635 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35637 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35638 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35640 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35642 * config/s390/s390.md: Add missing comments with the expanded
35644 * config/s390/vector.md: Likewise.
35645 * config/s390/vx-builtins.md: Likewise.
35647 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35650 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35651 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35652 conditions on a single line.
35654 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35656 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35657 __S390_VX__ to __VX__.
35659 2017-02-01 Andrew Pinski <apinski@cavium.com>
35661 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35662 stmt_info to record_stmt_cost.
35663 (vect_get_known_peeling_cost): Pass stmt_info if known to
35665 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35666 cpu_vector_cost field into
35667 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35668 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35669 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35670 splitting of scalar_stmt_cost and vec_stmt_cost.
35671 (thunderx_vector_cost): Likewise.
35672 (cortexa57_vector_cost): LIkewise.
35673 (exynosm1_vector_cost): Likewise.
35674 (xgene1_vector_cost): Likewise.
35675 (thunderx2t99_vector_cost): Improve after the splitting of the two
35677 (aarch64_builtin_vectorization_cost): Update for the splitting of
35678 scalar_stmt_cost and vec_stmt_cost.
35680 2017-02-01 Torvald Riegel <triegel@redhat.com>
35681 Richard Henderson <rth@redhat.com>
35683 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35684 conditional on existance of a fast atomic load.
35685 * optabs-query.c (can_atomic_load_p): New function.
35686 * optabs-query.h (can_atomic_load_p): Declare it.
35687 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35688 no fast atomic load is available for the particular size of access.
35689 (expand_atomic_compare_and_swap): Likewise.
35690 (expand_atomic_load): Likewise.
35691 (expand_atomic_store): Likewise.
35692 (expand_atomic_fetch_op): Likewise.
35693 * testsuite/lib/target-supports.exp
35694 (check_effective_target_sync_int_128): Remove x86 because it provides
35695 no fast atomic load.
35696 (check_effective_target_sync_int_128_runtime): Likewise.
35698 2017-02-01 Richard Biener <rguenther@suse.de>
35700 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35701 (graphite_transform_loops): Provide opt-info for optimized nests.
35702 * tree-parloop.c (parallelize_loops): Provide opt-info for
35703 parallelized loops.
35705 2017-02-01 Richard Biener <rguenther@suse.de>
35707 PR middle-end/79315
35708 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35709 was not set before.
35711 2017-02-01 Richard Biener <rguenther@suse.de>
35713 PR tree-optimization/71824
35714 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35715 Verify the loops are valid in the merged SESE region.
35716 (scop_detection::can_represent_loop_1): Check analyzing the
35717 evolution of the number of iterations in the region succeeds.
35719 2017-01-31 Ian Lance Taylor <iant@golang.org>
35721 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35722 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35724 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35726 PR preprocessor/79210
35727 * input.c (get_substring_ranges_for_loc): Replace line_width
35728 assertion with error-handling.
35730 2017-01-31 Richard Biener <rguenther@suse.de>
35732 PR tree-optimization/77318
35733 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35734 (create_pw_aff_from_tree): Take loop parameter.
35735 (add_condition_to_pbb): Pass loop of the condition to
35736 create_pw_aff_from_tree.
35738 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35740 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35741 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35743 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35747 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35749 (convert_int_to_float128): Likewise.
35750 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35751 (convert_int_to_float128): Likewise.
35752 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35753 (UNSPEC_IEEE128_CONVERT): Likewise.
35754 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35755 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35756 Use local variables for IBM extended format.
35757 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35758 (fix_trunc<mode>si2_fprs): Likewise.
35759 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35760 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35761 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35762 to know that we can now have integers of all sizes in vector
35764 (fix<uns>_<mode>di2_hw): Likewise.
35765 (float<uns>_<mode>si2_hw): Likewise.
35766 (fix_<mode>si2_hw): Likewise.
35767 (fixuns_<mode>si2_hw): Likewise.
35768 (float<uns>_<mode>di2_hw): Likewise.
35769 (float_<mode>di2_hw): Likewise.
35770 (float_<mode>si2_hw): Likewise.
35771 (floatuns_<mode>di2_hw): Likewise.
35772 (floatuns_<mode>si2_hw): Likewise.
35773 (xscvqp<su>wz_<mode>): Delete, no longer used.
35774 (xscvqp<su>dz_<mode>): Likewise.
35775 (xscv<su>dqp_<mode>): Likewise.
35776 (ieee128_mfvsrd_64bit): Likewise.
35777 (ieee128_mfvsrd_32bit): Likewise.
35778 (ieee128_mfvsrwz): Likewise.
35779 (ieee128_mtvsrw): Likewise.
35780 (ieee128_mtvsrd_64bit): Likewise.
35781 (ieee128_mtvsrd_32bit): Likewise.
35783 2017-01-31 Martin Liska <mliska@suse.cz>
35786 * ipa-prop.c (ipa_free_all_node_params): Call release method
35787 instead of ~sumbol_summary to not to trigger double times
35790 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35792 PR tree-optimization/71691
35793 * bitmap.h (class auto_bitmap): New.
35794 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35795 is_maybe_undefined instead of ssa_undefined_value_p.
35797 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35799 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35800 __S390_ARCH_LEVEL__ to __ARCH__.
35802 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35804 PR tree-optimization/79267
35805 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35806 if should_remove_lhs_p is true.
35808 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35811 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35812 (add_alignment_attribute): New.
35813 (base_type_die): Add alignment attribute.
35814 (subrange_type_die): Likewise.
35815 (modified_type_die): Likewise.
35816 (gen_array_type_die): Likewise.
35817 (gen_descr_array_type_die: Likewise.
35818 (gen_enumeration_type_die): Likewise.
35819 (gen_subprogram_die): Likewise.
35820 (gen_variable_die): Likewise.
35821 (gen_field_die): Likewise.
35822 (gen_ptr_to_mbr_type_die): Likewise.
35823 (gen_struct_or_union_type_die): Likewise.
35824 (gen_subroutine_type_die): Likewise.
35825 (gen_typedef_die): Likewise.
35826 (base_type_cmp): Compare alignment attribute.
35828 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35831 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35832 (setb_unsigned) New pattern for setb with CCUNS.
35833 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35834 subfc./subfe sequence to avoid overflow problems. Generate a
35835 shorter sequence with cmpld/setb for power9.
35836 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35837 for generating subfc. instruction.
35838 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35839 now uses this instruction.
35841 2017-01-30 Ian Lance Taylor <iant@google.com>
35844 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35845 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35847 2017-01-30 Martin Sebor <msebor@redhat.com>
35849 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35850 Move constant to the right of a relational operator.
35851 (get_mpfr_format_length, format_character, format_string): Ditto.
35852 (should_warn_p, maybe_warn): Same.
35854 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35856 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35859 * asan.c (get_translation_unit_decl): Remove function.
35860 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35862 2017-01-30 Martin Liska <mliska@suse.cz>
35864 PR gcov-profile/79259
35865 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35866 -fprofile-generate.
35868 2017-01-30 Martin Liska <mliska@suse.cz>
35871 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35872 Initialize variables with NULL value.
35874 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35877 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35879 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35881 2017-01-30 Richard Biener <rguenther@suse.de>
35883 PR tree-optimization/79276
35884 * tree-vrp.c (process_assert_insertions): Properly adjust common
35885 when removing a duplicate.
35887 2017-01-30 Richard Biener <rguenther@suse.de>
35889 PR tree-optimization/79256
35890 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35891 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35893 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35895 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35898 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35899 ("*r<noxa>sbg_<mode>_sll_bitmask")
35900 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35901 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35902 Use contiguous_bitmask_nowrap_operand.
35904 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35907 * config/rs6000/altivec.h (vec_xl): Revise #define.
35908 (vec_xst): Likewise.
35910 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35912 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35914 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35916 PR rtl-optimization/79194
35917 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35918 traps before call to bypass_conditional_jumps.
35920 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35922 PR tree-optimization/71374
35923 * lra-constraints.c (check_conflict_input_operands): New.
35924 (match_reload): Use it.
35926 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35929 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35930 account to calculate conflict_set.
35932 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35934 PR rtl-optimization/78559
35935 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35936 other_insn in combine.
35938 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35940 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35941 uint16_type_node for BT_UINT16.
35943 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35945 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35946 "RTL Tests" to menu.
35947 (GIMPLE Tests): New node.
35948 (RTL Tests): New node.
35950 2017-01-27 Richard Biener <rguenther@suse.de>
35952 PR tree-optimization/79245
35953 * tree-loop-distribution.c (distribute_loop): Apply cost
35954 modeling also to detected patterns.
35956 2017-01-27 Richard Biener <rguenther@suse.de>
35958 PR tree-optimization/71433
35959 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35960 (compare_assert_loc): New function.
35961 (process_assert_insertions): Sort and optimize assert locations
35962 to remove duplicates and push down identical assertions on
35963 edges to their destination block.
35965 2017-01-27 Richard Biener <rguenther@suse.de>
35967 PR tree-optimization/79244
35968 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35969 out SSA names even if abnormal.
35971 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35973 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35974 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35975 instead of MPFR_RNDN.
35977 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35980 * arm.c (arm_option_override): Don't call build_target_option_node
35981 until after doing all option overrides.
35982 (arm_valid_target_attribute_tree): Likewise.
35984 2017-01-27 Martin Liska <mliska@suse.cz>
35986 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35987 for {cd}tors and C++ {cd}tors.
35989 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35991 * config/s390/s390.md ("*setmem_long_and")
35992 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35994 2017-01-26 Martin Sebor <msebor@redhat.com>
35996 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35999 2017-01-26 Martin Sebor <msebor@redhat.com>
36001 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36002 HAVE_DFmode before using XFmode or DFmode.
36003 (parse_directive): Avoid using the z length modifier to avoid
36004 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36006 PR middle-end/78703
36007 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36008 to accept adjustment as an array.
36009 (get_int_range): New function.
36010 (struct directive): Make width and prec arrays.
36011 (directive::set_width, directive::set_precision): Call get_int_range.
36012 (format_integer, format_floating): Handle width and precision ranges.
36013 (format_string, parse_directive): Same.
36015 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36018 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36019 comdat_type_p set, just clone them, but keep the children in the
36023 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36024 which have direct callers with -fvar-tracking-assignments enabled
36026 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36027 inside of type units.
36029 2017-01-26 Martin Sebor <msebor@redhat.com>
36031 PR middle-end/78703
36032 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36034 (struct format_result): Replace number_chars, number_chars_min,
36035 and number_chars_max with a single member of struct result_range.
36037 (format_result::operator+=): Adjust.
36038 (struct fmtresult): Remove bounded. Handle likely and unlikely
36040 (fmtresult::adjust_for_width_or_precision): New function.
36041 (fmtresult:type_max_digits): New function.
36042 (bytes_remaining): Handle likely and unlikely counters.
36043 (min_bytes_remaining): Remove.
36044 (format_percent): Simplify.
36045 (format_integer, format_floating): Set likely and unlikely counters.
36046 (get_string_length, format_character, format_string): Same.
36047 (format_plain, should_warn_p): New function.
36048 (maybe_warn): Call should_warn_p. Update diagnostic messages
36049 and handle those for all directives, including plain strings.
36050 (format_directive): Handle likely and unlikely counters.
36051 Remove unnecessary quoting from diagnostics. Add an informational
36053 (add_bytes): Remove.
36054 (pass_sprintf_length::compute_format_length): Simplify.
36055 (try_substitute_return_value): Handle likely and unlikely counters.
36057 2017-01-26 Carl Love <cel@us.ibm.com>
36059 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36060 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36062 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36065 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36066 endianess for subregs into account.
36067 * lra-constraints.c (lra_constraints): Do risky transformations
36068 always on the first iteration.
36069 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36070 last_call_used_reg_set.
36071 (process_bb_lives): Define and use last_call_used_reg_set.
36072 * lra.c (lra): Always continue after lra_constraints on the first
36075 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36077 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36079 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36081 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36083 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36084 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36085 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36086 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36087 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36088 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36089 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36090 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36091 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36093 2017-01-26 Marek Polacek <polacek@redhat.com>
36096 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36097 for the third operand.
36099 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36101 PR middle-end/79236
36102 * omp-low.c (struct omp_context): Add simt_stmt field.
36103 (scan_omp_for): Return omp_context *.
36104 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36105 context to the _simt_ SIMD stmt.
36106 (lower_omp_for): For combined SIMD with sibling _simt_
36107 SIMD, make sure to use the same decls in _looptemp_
36108 clauses as in the sibling.
36110 2017-01-26 David Sherwood <david.sherwood@arm.com>
36112 PR middle-end/79212
36113 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36116 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36119 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36120 emit fld b; fld a; if possible.
36122 * brig-builtins.def: Update copyright years.
36123 * config/arm/arm_acle_builtins.def: Update copyright years.
36125 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36128 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36129 constraint instead of o for the stxsd instruction.
36131 2017-01-25 Carl Love <cel@us.ibm.com>
36133 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36134 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36136 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36138 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36140 2017-01-25 Richard Biener <rguenther@suse.de>
36142 PR tree-optimization/69264
36143 * target.def (vector_alignment_reachable): Improve documentation.
36144 * doc/tm.texi: Regenerate.
36145 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36147 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36148 earlier changes with respect to TYPE_USER_ALIGN.
36149 (vector_alignment_reachable_p): Likewise. Improve dumping.
36151 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36154 * config/arm/arm.md (xordi3): Force constant operand into a register
36157 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36159 * doc/invoke.texi (-fstore-merging): Correct default optimization
36160 levels at which it is enabled.
36161 (-O): Move -fstore-merging from list to...
36164 2017-01-25 Richard Biener <rguenther@suse.de>
36167 * omp-expand.c: Include debug.h.
36168 (expand_omp_taskreg): Make sure to generate early debug before
36169 outlining anything from a function.
36170 (expand_omp_target): Likewise.
36171 (grid_expand_target_grid_body): Likewise.
36173 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36176 * asan.c (get_translation_unit_decl): New function.
36177 (asan_add_global): Extract modules file name from globals
36178 TRANSLATION_UNIT_DECL name.
36180 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36183 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36184 for long calls with APCS frame and VFP.
36186 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36188 * cfg.c (original_copy_tables_initialized_p): New function.
36189 * cfg.h (original_copy_tables_initialized_p): New decl.
36190 * cfgrtl.c (relink_block_chain): Guard the call to
36191 free_original_copy_tables with a call to
36192 original_copy_tables_initialized_p.
36193 * cgraph.h (symtab_node::native_rtl_p): New decl.
36194 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36195 (symtab_node::needed_p): Don't assert for early assembly output
36196 for __RTL functions.
36197 (cgraph_node::finalize_function): Set "force_output" for __RTL
36199 (cgraph_node::analyze): Bail out early for __RTL functions.
36200 (analyze_functions): Update assertion to support __RTL functions.
36201 (cgraph_node::expand): Bail out early for __RTL functions.
36202 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36204 * function.h (struct function): Update comment for field
36206 * gimple-expr.c: Include "tree-pass.h".
36207 (gimple_has_body_p): Return false for __RTL functions.
36208 * Makefile.in (OBJS): Add run-rtl-passes.o.
36209 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36211 (gcc::pass_manager::get_clean_slate): New accessor.
36212 * passes.c: Include "insn-addr.h".
36213 (should_skip_pass_p): Add logging. Update logic for running
36214 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36215 property-provider override so it is only done for gimple passes.
36217 (skip_pass): New function.
36218 (execute_one_pass): Call skip_pass when skipping passes.
36219 * read-md.c (md_reader::read_char): Support filtering
36220 the input to a subset of line numbers.
36221 (md_reader::md_reader): Initialize fields
36222 m_first_line and m_last_line.
36223 (md_reader::read_file_fragment): New function.
36224 * read-md.h (md_reader::read_file_fragment): New decl.
36225 (md_reader::m_first_line): New field.
36226 (md_reader::m_last_line): New field.
36227 * read-rtl-function.c (function_reader::create_function): Only
36228 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36229 curr_properties. Set DECL_INITIAL to a dummy block.
36230 (read_rtl_function_body_from_file_range): New function.
36231 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36233 * run-rtl-passes.c: New file.
36234 * run-rtl-passes.h: New file.
36236 2017-01-24 Jeff Law <law@redhat.com>
36238 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36241 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36243 PR tree-optimization/79159
36244 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36245 (record_nonwrapping_iv): Improve boundary using above function if no
36246 value range information.
36248 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36249 Martin Jambor <mjambor@suse.cz>
36251 * brig-builtins.def: New file.
36252 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36253 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36254 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36255 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36256 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36257 * builtin-types.def (BT_INT8): New.
36258 (BT_INT16): Likewise.
36259 (BT_UINT8): Likewise.
36260 (BT_UINT16): Likewise.
36261 (BT_FN_ULONG): Likewise.
36262 (BT_FN_UINT_INT): Likewise.
36263 (BT_FN_UINT_ULONG): Likewise.
36264 (BT_FN_UINT_LONG): Likewise.
36265 (BT_FN_UINT_PTR): Likewise.
36266 (BT_FN_ULONG_PTR): Likewise.
36267 (BT_FN_INT8_FLOAT): Likewise.
36268 (BT_FN_INT16_FLOAT): Likewise.
36269 (BT_FN_UINT32_FLOAT): Likewise.
36270 (BT_FN_UINT16_FLOAT): Likewise.
36271 (BT_FN_UINT8_FLOAT): Likewise.
36272 (BT_FN_UINT64_FLOAT): Likewise.
36273 (BT_FN_UINT16_UINT32): Likewise.
36274 (BT_FN_UINT32_UINT16): Likewise.
36275 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36276 (BT_FN_INT_PTR_INT): Likewise.
36277 (BT_FN_UINT_PTR_UINT): Likewise.
36278 (BT_FN_LONG_PTR_LONG): Likewise.
36279 (BT_FN_ULONG_PTR_ULONG): Likewise.
36280 (BT_FN_VOID_UINT64_UINT64): Likewise.
36281 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36282 (BT_FN_INT8_INT8_INT8): Likewise.
36283 (BT_FN_INT16_INT16_INT16): Likewise.
36284 (BT_FN_INT_INT_INT): Likewise.
36285 (BT_FN_UINT_FLOAT_UINT): Likewise.
36286 (BT_FN_FLOAT_UINT_UINT): Likewise.
36287 (BT_FN_ULONG_UINT_UINT): Likewise.
36288 (BT_FN_ULONG_UINT_PTR): Likewise.
36289 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36290 (BT_FN_UINT_UINT_UINT): Likewise.
36291 (BT_FN_VOID_UINT_PTR): Likewise.
36292 (BT_FN_UINT_UINT_PTR: Likewise.
36293 (BT_FN_UINT32_UINT64_PTR): Likewise.
36294 (BT_FN_INT_INT_UINT_UINT): Likewise.
36295 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36296 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36297 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36298 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36299 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36300 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36301 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36302 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36303 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36304 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36305 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36306 * doc/frontends.texi: List BRIG FE.
36307 * doc/install.texi (Testing): Add BRIG tesring requirements.
36308 * doc/invoke.texi (Overall Options): Mention BRIG.
36309 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36311 2017-01-24 Richard Biener <rguenther@suse.de>
36313 PR translation/79208
36314 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36316 2017-01-24 Martin Jambor <mjambor@suse.cz>
36319 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36320 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36321 and known_contexts.
36323 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36325 PR middle-end/79123
36326 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36327 casts from signed to unsigned really don't have a range.
36329 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36331 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36332 GMP_RNDx for compatiblity.
36334 2017-01-24 Martin Liska <mliska@suse.cz>
36337 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36338 that would prevent us to call alloca with -1 as argument.
36340 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36342 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36343 Avoid -Wformat-security warning.
36345 2017-01-23 Andrew Pinski <apinski@cavium.com>
36347 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36350 2017-01-23 Martin Sebor <msebor@redhat.com>
36352 PR middle-end/78703
36353 * gimple-ssa-sprintf.c (warn_level): New global.
36354 (format_integer): Use it here and throughout the rest of the file.
36355 Use the same switch to compute sign as base.
36356 (maybe_warn): New function.
36357 (format_directive): Factor out warnings into maybe_warn.
36358 Add debugging output. Use warn_level.
36359 (add_bytes): Use warn_level.
36360 (pass_sprintf_length::compute_format_length): Add debugging output.
36361 (try_substitute_return_value): Same.
36362 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36364 PR middle-end/78703
36365 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36366 (struct fmtresult, format_integer, format_floating): Adjust.
36367 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36368 (get_string_length, format_string,format_directive): Same.
36369 (pass_sprintf_length::compute_format_length): Same.
36370 (try_substitute_return_value): Simplify slightly.
36372 PR middle-end/78703
36373 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36374 (fmtresult::operator+=): Outlined.
36375 (struct fmtresult): Add ctors.
36376 (struct conversion_spec): Rename...
36377 (struct directive): ...to this. Add and remove data members.
36378 (directive::set_width, directive::set_precision): New functions.
36379 (format_percent): Use fmtresult ctor.
36380 (get_width_and_precision): Remove.
36381 (format_integer): Make naming changes. Avoid computing width and
36383 (format_floating): Same. Adjust indentation.
36384 (format_character, format_none): New functions.
36385 (format_string): Moved character handling to format_character.
36386 (format_directive): Remove arguments, change return type.
36387 (parse_directive): New function.
36388 (pass_sprintf_length::compute_format_length): Move directive
36389 parsing to parse_directive.
36391 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36393 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36394 (assign_assembler_name_if_needed): ... this.
36395 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36396 (assign_assembler_name_if_needed): ... this.
36397 (free_lang_data_in_cgraph): Adjust callers.
36398 * cgraphunit.c (cgraph_node::analyze): Likewise.
36399 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36402 2017-01-23 Richard Biener <rguenther@suse.de>
36404 PR tree-optimization/79088
36405 PR tree-optimization/79188
36406 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36407 resetting loop bounds after last path deletion. Reset loop
36408 bounds of the target loop, make code match the comments.
36409 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36410 Make sure loops need no fixups.
36412 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36414 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36415 exponent support with double type for first argument.
36416 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36417 type returned by __builtin_vec_extract_sig,
36418 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36419 functions from "vector int" to "vector unsigned int" or from
36420 "vector long long int" to "vector unsigned long long int".
36421 Changed type returned by __builtin_vec_extract_exp,
36422 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36423 functions from "vector int" to "vector unsigned int" or from
36424 "vector long long int" to "vector unsigned long long int".
36425 Changed return type of __builtin_vec_test_data_class,
36426 __builtin_vec_test_data_class_sp, and
36427 __builtin_vec_test_data_class_dp from "vector int" to
36428 "vector bool int" or from "vector long long int" to "vector bool
36429 long long int" and changed second argument type from "unsigned
36430 int" to "int". Added new overloaded function forms "vector float
36431 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36432 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36433 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36434 double, vector unsigned long long int)" and "vector double
36435 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36436 long int)". Changed return type of
36437 __builtin_scalar_test_data_class and
36438 __builtin_scalar_test_data_class_sp and
36439 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36440 int" and changed second argument from "unsigned int" to "int".
36441 Changed type returned by __builtin_scalar_test_neg,
36442 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36443 from "int" to "bool int". Added new overloaded function form
36444 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36445 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36446 exponent double-precision with floating point first argument.
36447 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36448 documentation of scalar_test_data_class, scalar_test_neg,
36449 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36450 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36451 vec_test_data_class built-in functions to reflect refinements in
36452 their type signatures.
36454 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36456 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36458 (aarch64_elf_asm_destructor): Likewise.
36460 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36462 PR rtl-optimization/78634
36463 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36464 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36465 * ifcvt.c (noce_try_cmove): Add missing cost check.
36467 PR rtl-optimization/71724
36468 * combine.c (if_then_else_cond): Look for situations where it is
36469 beneficial to undo the work of one of the recursive calls.
36471 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36473 PR tree-optimization/70754
36474 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36475 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36476 combined stmt before it if not NULL.
36477 (combine_chains): Process refs reversely and compute dominance point
36480 2017-01-23 Martin Liska <mliska@suse.cz>
36482 PR tree-optimization/79196
36483 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36484 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36486 (strlen_optimize_stmt): Call the renamed function.
36488 2017-01-23 Michael Matz <matz@suse.de>
36490 PR tree-optimization/78384
36491 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36493 2017-01-23 Richard Biener <rguenther@suse.de>
36495 PR tree-optimization/79186
36496 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36497 both incoming edges before moving an assert.
36499 2017-01-23 Martin Jambor <mjambor@suse.cz>
36501 * ipa-prop.c (load_from_param_1): Removed.
36502 (load_from_unmodified_param): Bits from load_from_param_1 put back
36504 (load_from_param): Removed.
36505 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36506 with stmt. Reverted back to use of load_from_unmodified_param.
36508 2017-01-23 Martin Jambor <mjambor@suse.cz>
36511 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36512 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36513 field a pointer to garbage collected vector, mark lattices and
36514 ipcp_orig_node with GTY((skip)).
36515 (ipa_get_param_count): Adjust to descriptors being a pointer.
36516 (ipa_get_param): Likewise.
36517 (ipa_get_type): Likewise.
36518 (ipa_get_param_move_cost): Likewise.
36519 (ipa_set_param_used): Likewise.
36520 (ipa_get_controlled_uses): Likewise.
36521 (ipa_set_controlled_uses): Likewise.
36522 (ipa_is_param_used): Likewise.
36523 (ipa_node_params_t): Move into garbage collector. New methods insert
36525 (ipa_node_params_sum): Annotate wth GTY(()).
36526 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36528 (ipa_load_from_parm_agg): Adjust declaration.
36529 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36530 * ipa-profile.c (ipa_profile): Likewise.
36531 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36532 (ipa_populate_param_decls): Make descriptors parameter garbage
36534 (ipa_dump_param): Adjust to descriptors being a pointer.
36535 (ipa_alloc_node_params): Likewise.
36536 (ipa_initialize_node_params): Likewise.
36537 (load_from_param_1): Make descriptors parameter garbage collected.
36538 (load_from_unmodified_param): Likewise.
36539 (load_from_param): Likewise.
36540 (ipa_load_from_parm_agg): Likewise.
36541 (ipa_node_params::~ipa_node_params): Removed.
36542 (ipa_free_all_node_params): Remove call to delete operator.
36543 (ipa_node_params_t::insert): New.
36544 (ipa_node_params_t::remove): Likewise.
36545 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36546 copy known_csts and known_contexts vectors.
36547 (ipa_read_node_info): Adjust to descriptors being a pointer.
36548 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36550 (ipcp_transform_function): Make descriptors variable garbage
36553 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36555 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36556 * config/i386/avx512dqintrin.h: Ditto.
36557 * config/i386/avx512fintrin.h: Ditto.
36558 * config/i386/i386.c: Handle new builtins.
36559 * config/i386/i386-builtin.def: Add new builtins.
36560 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36561 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36563 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36564 Martin Liska <mliska@suse.cz>
36566 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36567 * asan.c (asan_expand_poison_ifn): Support stores and use
36568 appropriate ASAN report function.
36569 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36570 * internal-fn.def (ASAN_POISON_USE): Declare.
36571 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36572 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36573 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36574 ASAN_POISON calls w/o LHS.
36575 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36576 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36577 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36578 * gimplify.c (asan_poison_variables): Add attribute
36579 use_after_scope_memory to variables that really needs to live
36581 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36582 having the attribute.
36584 2017-01-23 Martin Liska <mliska@suse.cz>
36586 * asan.c (create_asan_shadow_var): New function.
36587 (asan_expand_poison_ifn): Likewise.
36588 * asan.h (asan_expand_poison_ifn): New declaration.
36589 * internal-fn.c (expand_ASAN_POISON): Likewise.
36590 * internal-fn.def (ASAN_POISON): New builtin.
36591 * sanopt.c (pass_sanopt::execute): Expand
36592 asan_expand_poison_ifn.
36593 * tree-inline.c (copy_decl_for_dup_finish): Make function
36595 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36596 * tree-ssa.c (is_asan_mark_p): New function.
36597 (execute_update_addresses_taken): Rewrite local variables
36598 (identified just by use-after-scope as addressable) into SSA.
36600 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36602 * doc/install.texi (Specific): opensource.apple.com uses https
36603 now. Remove trailing slash.
36605 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36607 * README.Portability: Remove note on an Irix compatibility issue.
36609 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36611 * gcov.c (INCLUDE_ALGORITHM): Define.
36612 (INCLUDE_VECTOR): Define.
36613 No longer include <vector> and <algorithm> directly.
36615 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36617 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36619 * doc/invoke.texi (Code Gen Options): Ditto.
36621 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36624 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36626 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36628 rtl-optimization/79125
36629 * cprop.c (local_cprop_pass): Handle cases where we make an
36630 unconditional trap.
36632 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36636 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36637 read from, for big endian.
36639 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36641 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36642 register pauth builtins for LP64 only.
36644 2017-01-20 Marek Polacek <polacek@redhat.com>
36647 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36650 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36652 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36654 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36655 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36656 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36658 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36661 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36662 in big-endian mode when they are not a single duplicated value.
36664 2017-01-20 Richard Biener <rguenther@suse.de>
36666 * BASE-VER: Bump to 7.0.1.
36668 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36670 * omp-low.c (omplow_simd_context): New struct. Use it...
36671 (lower_rec_simd_input_clauses): ...here and...
36672 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36673 references to idx, lane, max_vf, is_simt.
36675 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36677 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36680 2017-01-20 Martin Jambor <mjambor@suse.cz>
36682 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36683 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36685 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36686 (GTFILES): Rename hsa.c to hsa-common.c.
36687 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36688 * hsa-dump.c: Likewise.
36689 * hsa-gen.c: Likewise.
36690 * hsa-regalloc.c: Likewise.
36691 * ipa-hsa.c: Likewise.
36692 * omp-expand.c: Likewise.
36693 * omp-low.c: Likewise.
36694 * toplev.c: Likewise.
36696 2017-01-20 Marek Polacek <polacek@redhat.com>
36699 * doc/invoke.texi: Document -Wduplicated-branches.
36700 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36701 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36702 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36703 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36704 return 0 only when not OEP_LEXICOGRAPHIC.
36705 (fold_build_cleanup_point_expr): Use the expression
36706 location when building CLEANUP_POINT_EXPR.
36707 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36708 * tree.c (add_expr): Handle error_mark_node.
36710 2017-01-20 Martin Liska <mliska@suse.cz>
36713 * tree-profile.c (init_ic_make_global_vars): Do not call
36715 (gimple_init_gcov_profiler): Likewise.
36717 2017-01-20 Martin Liska <mliska@suse.cz>
36720 * cgraph.h (maybe_create_reference): Remove argument and
36722 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36724 * ipa-cp.c (create_specialized_node): Likewise.
36725 * symtab.c (symtab_node::maybe_create_reference): Handle
36726 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36728 2017-01-20 Martin Liska <mliska@suse.cz>
36730 * read-rtl-function.c (function_reader::create_function): Use
36731 build_decl instread of build_decl_stat.
36733 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36735 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36736 * config/i386/avx512dqintrin.h: Ditto.
36737 * config/i386/avx512fintrin.h: Ditto.
36738 * config/i386/i386-builtin-types.def: Add new types.
36739 * config/i386/i386.c: Handle new types.
36740 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36741 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36742 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36743 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36744 (__builtin_ia32_kshiftridi): New.
36745 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36747 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36751 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36752 define to rs6000_init_stack_protect_guard.
36753 (rs6000_init_stack_protect_guard): New function.
36755 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36756 Yunqiang Su <yunqiang.su@imgtec.com>
36758 * config.gcc (supported_defaults): Add madd4.
36759 (with_madd4): Add validation.
36760 (all_defaults): Add madd4.
36761 * config/mips/mips.opt (mmadd4): New option.
36762 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36764 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36766 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36767 (ISA_HAS_FUSED_MADD4): Likewise.
36768 * doc/invoke.texi (-mmadd4): Document the new option.
36769 * doc/install.texi (--with-madd4): Document the new option.
36771 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36773 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36774 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36775 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36776 (aarch64_init_pauth_hint_builtins): New.
36777 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36778 (aarch64_expand_builtin): Expand new builtins.
36780 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36782 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36783 * combine-stack-adj.c (no_unhandled_cfa): Handle
36784 REG_CFA_TOGGLE_RA_MANGLE.
36785 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36786 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36787 info for return address signing.
36788 (aarch64_expand_epilogue): Likewise.
36790 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36792 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36793 * config/aarch64/aarch64-protos.h
36794 (aarch64_return_address_signing_enabled): New declaration.
36795 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36797 (aarch64_expand_prologue): Sign return address before it's pushed onto
36799 (aarch64_expand_epilogue): Authenticate return address fetched from
36801 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36802 (aarch64_attributes): New function attributes for "sign-return-address".
36803 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36804 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36805 ("*do_return"): Generate combined instructions according to key index.
36806 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36807 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36809 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36810 * config/aarch64/aarch64.opt (msign-return-address=): New.
36811 * doc/extend.texi (AArch64 Function Attributes): Documents
36812 "sign-return-address=".
36813 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36815 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36817 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36818 overall option summary.
36820 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36822 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36823 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36824 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36825 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36827 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36829 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36830 -mpower9-minmax by default for -mcpu=power9.
36831 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36832 128-bit floating point.
36834 2017-01-20 Alan Modra <amodra@gmail.com>
36836 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36837 optimizing for size.
36839 2017-01-20 Alan Modra <amodra@gmail.com>
36842 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36843 for strcmp and strncmp from corresponding builtin decl.
36845 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36847 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36848 instead of i386/rtems-64.h.
36849 * config/i386/rtems-64.h: Remove.
36851 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36855 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36857 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36859 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36861 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36862 Change int to HOST_WIDE_INT.
36863 * config/aarch64/aarch64-protos.h
36864 (aarch64_simd_gen_const_vector_dup): Likewise.
36865 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36867 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36869 * langhooks-def.h (lhd_type_for_size): New decl.
36870 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36871 * langhooks.c (lhd_type_for_size): New function, taken from
36874 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36876 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36877 define_bypass for CR latency.
36878 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36879 (power9-alu2): Add define_bypass for CR latency.
36881 (power9-mul): Update insn latency.
36882 (power9-mul-compare): Update insn latency, bypass latency and remove
36885 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36887 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36889 * config/aarch64/aarch64.md
36890 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36891 aarch64_nopcrelative_literal_loads.
36892 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36894 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36896 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36897 TARGET_LOONGSON_3A.
36898 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36900 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36903 * config.gcc (supported_defaults): Add lxc1-sxc1.
36904 (with_lxc1_sxc1): Add validation.
36905 (all_defaults): Add lxc1-sxc1.
36906 * config/mips/mips.opt (mlxc1-sxc1): New option.
36907 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36909 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36910 __mips_no_lxc1_sxc1.
36911 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36912 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36913 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36915 2017-01-19 Richard Biener <rguenther@suse.de>
36917 PR tree-optimization/72488
36918 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36919 sure to restore SSA info.
36920 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36922 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36924 PR rtl-optimization/79121
36925 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36926 of the inner type when shifting an extended value.
36928 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36931 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36932 interposable aliases.
36934 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36937 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36938 Use the evmergelohi instruction.
36939 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36940 (mov_sitf_e500_subreg8_2_be): Likewise.
36941 (mov_sitf_e500_subreg12_2_le): Likewise.
36942 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36943 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36944 (mov_sitf_e500_subreg8_2_le): Likewise.
36945 (mov_sitf_e500_subreg12_2_be): Likewise.
36947 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36949 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36950 attribute from vecsimple to vecperm.
36951 (altivec_vbpermq2): Likewise.
36953 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36956 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36958 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36959 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36960 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36961 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36962 case where N arg is SIZE_MAX.
36963 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36964 (cmpstrsi): Add pattern.
36966 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36968 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36969 __builtin_vec_revb builtins.
36970 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36971 built-in functions to support generation of the ISA 3.0 XXBR<x>
36972 vector byte reverse instructions.
36973 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36974 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36975 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36976 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36977 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36978 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36979 (P9V_BUILTIN_VEC_REVB): Likewise.
36980 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36981 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36982 (p9_xxbrq_v16qi): Likewise.
36983 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36984 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36985 (p9_xxbrh_v8hi): Likewise.
36986 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36987 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36988 vec_revb built-in functions.
36990 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36992 PR rtl-optimization/78952
36993 * config/i386/i386.md (any_extract): New code iterator.
36994 (*insvqi_2): Use any_extract for source operand.
36995 (*insvqi_3): Use any_shiftrt for source operand.
36997 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36999 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37001 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37003 2017-01-18 Matthias Klose <doko@ubuntu.com>
37005 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37007 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37009 * config/rs6000/altivec.h (vec_bperm): Change #define.
37010 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37011 (altivec_vbpermq2): New define_insn.
37012 (altivec_vbpermd): Likewise.
37013 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37014 function interface.
37015 (VBPERMD): Likewise.
37016 (VBPERM): New polymorphic function interface.
37017 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37018 Add entries for P9V_BUILTIN_VEC_VBPERM.
37019 * doc/extend.texi: Add interfaces for vec_bperm.
37021 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37023 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37024 first letter of error messages.
37025 (s390_resolve_overloaded_builtin): Likewise.
37026 * config/s390/s390.c (s390_expand_builtin): Likewise.
37027 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37028 (s390_valid_target_attribute_inner_p): Likewise.
37029 * config/s390/s390.md ("tabort"): Likewise.
37031 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37033 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37034 (ISA_AVOID_DIV_HILO): New macro.
37035 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37036 (ISA_HAS_DDIV): Likewise.
37038 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37040 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37042 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37044 * doc/invoke.texi (fabi-version): Spelling fix.
37046 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37049 * doc/invoke.texi (fabi-version): Mention mangling fix for
37052 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37055 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37057 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37060 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37061 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37063 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37065 (stack_protect_test): Ditto.
37066 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37067 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37069 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37070 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37071 -mstack-protector-guard-offset=.
37072 (RS/6000 and PowerPC Options): Ditto.
37074 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37076 * config/i386/i386.h (MASK_CLASS_P): New define.
37077 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37078 there are no registers from different register sets also when
37079 mask registers are used. Update function comment.
37080 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37081 to (*k/*r) and (*k/*km) alternatives.
37083 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37085 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37086 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37087 (EH_RETURN_HANDLER_RTX): New define.
37088 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37089 Force frame pointer in EH return functions.
37090 (aarch64_expand_epilogue): Add barrier for eh_return.
37091 (aarch64_final_eh_return_addr): Remove.
37092 (aarch64_eh_return_handler_rtx): New function.
37093 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37095 (aarch64_eh_return_handler_rtx): New prototype.
37097 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37099 * config/rs6000/altivec.h (vec_rlmi): New #define.
37100 (vec_vrlnm): Likewise.
37101 (vec_rlnm): Likewise.
37102 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37103 (UNSPEC_VRLNM): Likewise.
37104 (VIlong): New mode iterator.
37105 (altivec_vrl<VI_char>mi): New define_insn.
37106 (altivec_vrl<VI_char>nm): Likewise.
37107 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37109 (VRLDNM): Likewise.
37110 (RLNM): New polymorphic function entry.
37111 (VRLWMI): New monomorphic function entry.
37112 (VRLDMI): Likewise.
37113 (RLMI): New polymorphic function entry.
37114 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37115 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37116 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37119 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37122 * dwarf2out.c (field_byte_offset): Restore the
37123 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37124 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37125 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37128 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37131 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37133 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37136 * dwarf2out.c (add_data_member_location_attribute): For constant
37137 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37138 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37139 DW_AT_byte_size attributes.
37141 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37143 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37144 after forcing to constant memory when the code model is medium.
37146 2017-01-17 Julia Koval <julia.koval@intel.com>
37149 * config/i386/avx512fintrin.h
37150 (_mm512_i32gather_ps): Change __addr type to void const*.
37151 (_mm512_mask_i32gather_ps): Ditto.
37152 (_mm512_i32gather_pd): Ditto.
37153 (_mm512_mask_i32gather_pd): Ditto.
37154 (_mm512_i64gather_ps): Ditto.
37155 (_mm512_mask_i64gather_ps): Ditto.
37156 (_mm512_i64gather_pd): Ditto.
37157 (_mm512_mask_i64gather_pd): Ditto.
37158 (_mm512_i32gather_epi32): Ditto.
37159 (_mm512_mask_i32gather_epi32): Ditto.
37160 (_mm512_i32gather_epi64): Ditto.
37161 (_mm512_mask_i32gather_epi64): Ditto.
37162 (_mm512_i64gather_epi32): Ditto.
37163 (_mm512_mask_i64gather_epi32): Ditto.
37164 (_mm512_i64gather_epi64): Ditto.
37165 (_mm512_mask_i64gather_epi64): Ditto.
37166 (_mm512_i32scatter_ps): Change __addr type to void*.
37167 (_mm512_mask_i32scatter_ps): Ditto.
37168 (_mm512_i32scatter_pd): Ditto.
37169 (_mm512_mask_i32scatter_pd): Ditto.
37170 (_mm512_i64scatter_ps): Ditto.
37171 (_mm512_mask_i64scatter_ps): Ditto.
37172 (_mm512_i64scatter_pd): Ditto.
37173 (_mm512_mask_i64scatter_pd): Ditto.
37174 (_mm512_i32scatter_epi32): Ditto.
37175 (_mm512_mask_i32scatter_epi32): Ditto.
37176 (_mm512_i32scatter_epi64): Ditto.
37177 (_mm512_mask_i32scatter_epi64): Ditto.
37178 (_mm512_i64scatter_epi32): Ditto.
37179 (_mm512_mask_i64scatter_epi32): Ditto.
37180 (_mm512_i64scatter_epi64): Ditto.
37181 (_mm512_mask_i64scatter_epi64): Ditto.
37182 * config/i386/avx512pfintrin.h
37183 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37184 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37185 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37186 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37187 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37188 (_mm512_prefetch_i32scatter_ps): Ditto.
37189 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37190 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37191 (_mm512_prefetch_i64scatter_pd): Ditto.
37192 (_mm512_prefetch_i64scatter_ps): Ditto.
37193 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37194 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37195 * config/i386/avx512vlintrin.h
37196 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37197 (_mm_mmask_i32gather_ps): Ditto.
37198 (_mm256_mmask_i32gather_pd): Ditto.
37199 (_mm_mmask_i32gather_pd): Ditto.
37200 (_mm256_mmask_i64gather_ps): Ditto.
37201 (_mm_mmask_i64gather_ps): Ditto.
37202 (_mm256_mmask_i64gather_pd): Ditto.
37203 (_mm_mmask_i64gather_pd): Ditto.
37204 (_mm256_mmask_i32gather_epi32): Ditto.
37205 (_mm_mmask_i32gather_epi32): Ditto.
37206 (_mm256_mmask_i32gather_epi64): Ditto.
37207 (_mm_mmask_i32gather_epi64): Ditto.
37208 (_mm256_mmask_i64gather_epi32): Ditto.
37209 (_mm_mmask_i64gather_epi32): Ditto.
37210 (_mm256_mmask_i64gather_epi64): Ditto.
37211 (_mm_mmask_i64gather_epi64): Ditto.
37212 (_mm256_i32scatter_ps): Change __addr type to void*.
37213 (_mm256_mask_i32scatter_ps): Ditto.
37214 (_mm_i32scatter_ps): Ditto.
37215 (_mm_mask_i32scatter_ps): Ditto.
37216 (_mm256_i32scatter_pd): Ditto.
37217 (_mm256_mask_i32scatter_pd): Ditto.
37218 (_mm_i32scatter_pd): Ditto.
37219 (_mm_mask_i32scatter_pd): Ditto.
37220 (_mm256_i64scatter_ps): Ditto.
37221 (_mm256_mask_i64scatter_ps): Ditto.
37222 (_mm_i64scatter_ps): Ditto.
37223 (_mm_mask_i64scatter_ps): Ditto.
37224 (_mm256_i64scatter_pd): Ditto.
37225 (_mm256_mask_i64scatter_pd): Ditto.
37226 (_mm_i64scatter_pd): Ditto.
37227 (_mm_mask_i64scatter_pd): Ditto.
37228 (_mm256_i32scatter_epi32): Ditto.
37229 (_mm256_mask_i32scatter_epi32): Ditto.
37230 (_mm_i32scatter_epi32): Ditto.
37231 (_mm_mask_i32scatter_epi32): Ditto.
37232 (_mm256_i32scatter_epi64): Ditto.
37233 (_mm256_mask_i32scatter_epi64): Ditto.
37234 (_mm_i32scatter_epi64): Ditto.
37235 (_mm_mask_i32scatter_epi64): Ditto.
37236 (_mm256_i64scatter_epi32): Ditto.
37237 (_mm256_mask_i64scatter_epi32): Ditto.
37238 (_mm_i64scatter_epi32): Ditto.
37239 (_mm_mask_i64scatter_epi32): Ditto.
37240 (_mm256_i64scatter_epi64): Ditto.
37241 (_mm256_mask_i64scatter_epi64): Ditto.
37242 (_mm_i64scatter_epi64): Ditto.
37243 (_mm_mask_i64scatter_epi64): Ditto.
37244 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37245 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37246 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37247 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37248 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37249 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37250 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37251 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37252 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37253 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37254 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37255 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37256 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37257 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37258 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37259 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37260 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37261 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37262 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37263 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37264 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37265 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37266 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37267 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37268 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37269 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37270 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37271 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37272 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37273 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37274 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37275 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37276 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37277 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37278 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37279 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37280 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37281 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37282 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37283 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37284 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37285 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37286 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37287 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37288 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37289 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37290 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37291 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37292 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37293 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37294 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37295 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37296 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37297 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37298 definitions accordingly.
37300 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37301 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37304 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37307 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37310 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37311 hard regs for inner regno.
37313 2017-01-17 Martin Liska <mliska@suse.cz>
37316 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37317 assumption and add comment.
37319 2017-01-17 Nathan Sidwell <nathan@acm.org>
37321 * ipa-visibility.c (localize_node): New function, broken out of ...
37322 (function_and_variable_visibility): ... here. Call it.
37324 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37326 PR middle-end/77445
37327 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37328 correctly set frequency of oudgoing edge.
37329 (duplicate_thread_path): Fix profile updating.
37331 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37334 * configure.ac: Add GCC_BASE_VER.
37335 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37336 version from BASE-VER file.
37337 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37338 (gcc.o): Depend on $(BASEVER).
37339 * common.opt (dumpfullversion): New option.
37340 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37341 * doc/invoke.texi: Document -dumpfullversion.
37342 * doc/install.texi: Document --with-gcc-major-version-only.
37343 * configure: Regenerated.
37345 2017-01-17 Richard Biener <rguenther@suse.de>
37347 PR tree-optimization/71433
37348 * tree-vrp.c (register_new_assert_for): Merge same asserts
37349 on all incoming edges.
37350 (process_assert_insertions_for): Handle insertions at the
37353 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37355 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37356 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37358 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37361 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37364 2017-01-17 Alan Modra <amodra@gmail.com>
37367 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37368 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37369 symbolic stack limit when pic.
37371 2017-01-16 Martin Sebor <msebor@redhat.com>
37373 PR tree-optimization/78608
37374 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37376 2017-01-16 Jeff Law <law@redhat.com>
37379 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37380 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37381 for several include directories that may be relative to sysroot.
37382 * config/i386/x-mingw32 (gplus_includedir): Define.
37383 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37384 (native_system_includedir): Likewise.
37385 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37386 override if TARGET_SYSTEM_ROOT is defined.
37387 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37389 PR tree-optimization/79090
37390 PR tree-optimization/33562
37391 PR tree-optimization/61912
37392 PR tree-optimization/77485
37393 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37394 and computed trims into the dump file.
37396 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37398 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37400 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37403 * gimplify.c (gimplify_init_constructor): If want_value and
37404 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37408 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37409 sequence. Formatting fixes.
37410 (doloop_optimize): Formatting fixes.
37413 * gcc.c (debug_level_greater_than_spec_func): New function.
37414 (static_spec_functions): Add debug-level-gt spec function.
37415 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37417 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37418 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37419 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37420 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37421 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37422 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37424 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37426 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37427 QImode fixups to general and mask registers only.
37429 2017-01-16 Carl Love <cel@us.ibm.com>
37431 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37432 for built-in functions
37433 vector signed char vec_nabs (vector signed char)
37434 vector signed short vec_nabs (vector signed short)
37435 vector signed int vec_nabs (vector signed int)
37436 vector signed long long vec_nabs (vector signed long long)
37437 vector float vec_nabs (vector float)
37438 vector double vec_nabs (vector double)
37439 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37441 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37442 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37443 * doc/extend.texi: Update the documentation file for the new built-in
37446 2017-01-16 Martin Sebor <msebor@redhat.com>
37448 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37451 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37453 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37454 UNSPEC_VSX__XXSPLTD to require special splat handling.
37456 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37459 * system.h: Poison strndup.
37461 2017-01-16 Alan Modra <amodra@gmail.com>
37464 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37467 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37469 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37471 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37473 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37474 call recog here. Assert that INSN_CODE (insn) is non-negative.
37476 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37479 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37481 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37482 in the currently scheduled RTL fragment.
37484 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37486 PR rtl-optimization/78751
37487 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37490 2017-01-14 Jeff Law <law@redhat.com>
37492 PR tree-optimization/79090
37493 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37494 variable length stores.
37495 (compute_trims): Delete dead assignment to *trim_tail.
37496 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37499 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37501 PR rtl-optimization/78626
37502 PR rtl-optimization/78727
37503 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37504 of a block, and split such blocks after everything else is finished.
37506 2017-01-14 Alan Modra <amodra@gmail.com>
37509 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37510 target legitimate_combined_insn.
37511 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37512 (rs6000_legitimate_combined_insn): New function.
37513 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37515 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37516 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37517 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37519 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37521 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37523 2017-01-13 Jeff Law <law@redhat.com>
37525 PR tree-optimization/33562
37526 PR tree-optimization/61912
37527 PR tree-optimization/77485
37528 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37530 (delete_dead_assignment): Likewise.
37531 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37532 statement to delete_dead_call and delete_dead_assignment.
37534 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37537 * substring-locations.c (format_warning_va): Strengthen case 1 so
37538 that both endpoints of the substring must be within the format
37539 range for just the substring to be printed.
37541 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37543 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37544 * config/i386/i386.c (ix86_target_string): Add missing options
37545 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37546 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37547 flags_other and ix86_target_other to flags2_other. Display unknown
37549 (ix86_valid_target_attribute_inner_p): Add missing options and
37550 reorder options by implied ISAs, as in ix86_target_string.
37552 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37554 * hash-table.h (hash_table::too_empty_p): New function.
37555 (hash_table::expand): Use it.
37556 (hash_table::traverse): Likewise.
37557 (hash_table::empty_slot): Use sizeof (value_type) instead of
37558 sizeof (PTR) to convert bytes to elements. Shrink the table
37559 if the current size is excessive for the current number of
37562 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37564 * ira-costs.c (record_reg_classes): Break from the inner loop
37565 early once alt_fail is known to be true. Update outer loop
37566 handling accordingly.
37568 2017-01-13 Jeff Law <law@redhat.com>
37570 * tree-ssa-dse.c (decrement_count): New function.
37571 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37572 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37573 when we know the partially dead statement is a mem* function.
37575 PR tree-optimization/61912
37576 PR tree-optimization/77485
37577 * tree-ssa-dse.c: Include expr.h.
37578 (maybe_trim_constructor_store): New function.
37579 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37581 PR tree-optimization/33562
37582 PR tree-optimization/61912
37583 PR tree-optimization/77485
37584 * doc/invoke.texi: Document new dse-max-object-size param.
37585 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37586 * tree-ssa-dse.c: Include params.h.
37587 (dse_store_status): New enum.
37588 (initialize_ao_ref_for_dse): New, partially extracted from
37590 (valid_ao_ref_for_dse, normalize_ref): New.
37591 (setup_live_bytes_from_ref, compute_trims): Likewise.
37592 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37593 (maybe_trim_partially_dead_store): Likewise.
37594 (maybe_trim_complex_store): Likewise.
37595 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37596 Track what bytes live from the original store. Return tri-state
37597 for dead, partially dead or live.
37598 (dse_dom_walker): Add constructor, destructor and new private members.
37599 (delete_dead_call, delete_dead_assignment): New extracted from
37601 (dse_optimize_stmt): Make a member of dse_dom_walker.
37602 Use initialize_ao_ref_for_dse.
37604 PR tree-optimization/33562
37605 PR tree-optimization/61912
37606 PR tree-optimization/77485
37607 * sbitmap.h (bitmap_count_bits): Prototype.
37608 (bitmap_clear_range, bitmap_set_range): Likewise.
37609 * sbitmap.c (bitmap_clear_range): New function.
37610 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37612 2017-01-13 Martin Liska <mliska@suse.cz>
37615 * function.c (set_cfun): Add new argument force.
37616 * function.h (set_cfun): Likewise.
37617 * ipa-inline-transform.c (inline_call): Use the function when
37618 strict alising from is dropped for function we inline to.
37620 2017-01-13 Richard Biener <rguenther@suse.de>
37622 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37623 for dumping GIMPLE INTEGER_CSTs.
37625 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37627 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37628 to 201112L since C++17.
37630 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37633 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37634 if -fsanitize=kernel-address is present.
37636 2017-01-13 Richard Biener <rguenther@suse.de>
37638 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37639 as _Literal ( type ) number in case usual suffixes do not
37640 preserve all information.
37642 2017-01-13 Richard Biener <rguenther@suse.de>
37644 PR tree-optimization/77283
37645 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37646 and ssa-iterators.h.
37647 (is_feasible_trace): Implement a cost model based on joiner
37650 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37653 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37654 char or short to __float128/_Float128 directly.
37656 2017-01-12 Martin Sebor <msebor@redhat.com>
37658 to -Wformat-overflow.
37659 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37660 (min_bytes_remaining): Same.
37661 (get_string_length): Same.
37662 (format_string): Same.
37663 (format_directive): Same.
37665 (pass_sprintf_length::handle_gimple_call): Same.
37667 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37669 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37670 info.nowrite calls with no lhs that can't throw. Return bool
37671 whether gsi_remove has been called or not.
37672 (pass_sprintf_length::handle_gimple_call): Return bool whether
37673 try_substitute_return_value called gsi_remove. Formatting fix.
37674 (pass_sprintf_length::execute): Don't use gsi_remove if
37675 handle_gimple_call returned true.
37678 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37679 be removed due to side-effects, don't remove following barrier nor
37680 turn the successor edge into fallthru edge.
37682 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37685 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37686 element-reversing loads and stores as not swappable.
37688 2017-01-12 Nathan Sidwell <nathan@acm.org>
37689 Nicolai Stange <nicstange@gmail.com>
37691 * combine.c (try_combine): Don't ignore result of overlap checking
37692 loop. Combine overlap & asm check into single loop.
37694 2017-01-12 Richard Biener <rguenther@suse.de>
37696 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37697 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37699 2017-01-12 Richard Biener <rguenther@suse.de>
37701 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37702 and TS_TARGET_OPTION directly derive from TS_BASE.
37703 * tree-core.h (tree_optimization_option): Derive from tree_base.
37704 (tree_target_option): Likewise.
37706 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37708 * config/i386/i386.c (memory_address_length): Increase len
37709 only when rip_relative_addr_p returns false.
37711 2017-01-11 Julia Koval <julia.koval@intel.com>
37713 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37714 (OPTION_MASK_ISA_SGX_SET): New.
37715 (ix86_handle_option): Handle OPT_msgx.
37716 * config.gcc: Added sgxintrin.h.
37717 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37718 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37719 * config/i386/i386.c (ix86_target_string): Add -msgx.
37721 (ix86_option_override_internal): Handle new options.
37722 (ix86_valid_target_attribute_inner_p): Add sgx.
37723 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37724 * config/i386/i386.opt: Add msgx.
37725 * config/i386/sgxintrin.h: New file.
37726 * config/i386/x86intrin.h: Add sgxintrin.h.
37728 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37731 * fold-const.c (maybe_nonzero_address): Return 1 for function
37733 (tree_single_nonzero_warnv_p): Don't handle function local objects
37737 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37740 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37743 * opts.c: Include "spellcheck.h"
37744 (struct string_fragment): New struct.
37745 (struct edit_distance_traits<const string_fragment &>): New
37747 (get_closest_sanitizer_option): New function.
37748 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37750 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37752 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37754 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37755 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37756 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37757 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37758 for initial die_offset if dwarf_split_debug_info.
37759 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37760 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37762 (output_skeleton_debug_sections): Formatting fix. Use
37763 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37764 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37766 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37768 * config/arm/cortex-a53.md: Add bypasses for
37769 cortex_a53_r2f_cvt.
37770 (cortex_a53_r2f): Only use for transfers.
37771 (cortex_a53_f2r): Likewise.
37772 (cortex_a53_r2f_cvt): Add reservation for conversions.
37773 (cortex_a53_f2r_cvt): Likewise.
37775 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37777 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37778 to all inlined functions, change static to extern.
37780 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37783 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37785 (arm_assemble_integer): Likewise.
37787 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37789 * config.gcc: Use new awk script to check CPU, FPU and architecture
37790 parameters for --with-... options.
37791 * config/arm/parsecpu.awk: New file
37792 * config/arm/arm-cpus.in: New file.
37793 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37795 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37797 * config/arm/t-arm: Update dependency rules.
37798 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37799 of processing .def files.
37800 * config/arm/genopt.sh: Deleted.
37801 * config/arm/gentune.sh: Deleted.
37802 * config/arm/arm-cores.def: Deleted.
37803 * config/arm/arm-arches.def: Deleted.
37804 * config/arm/arm-fpus.def: Deleted.
37805 * config/arm/arm-tune.md: Regenerated.
37806 * config/arm/arm-tables.opt: Regenerated.
37807 * config/arm/arm-cpu.h: New generated file.
37808 * config/arm/arm-cpu-data.h: New generated file.
37809 * config/arm/arm-cpu-cdata.h: New generated file.
37811 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37814 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37816 (input_varpool_node): Unpack dynamically_initialized bit.
37818 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37820 PR rtl-optimization/79032
37821 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37822 the alignment of the adjusted memory reference against that of MODE,
37823 instead of the alignment of the original memory reference.
37825 2017-01-11 Martin Jambor <mjambor@suse.cz>
37827 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37829 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37830 decorated functions.
37832 2017-01-11 Richard Biener <rguenther@suse.de>
37834 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37835 set range/nonnull info for PHI results. Do not set it on
37836 stmts marked for removal.
37838 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37840 * expr.c (store_field): In the bitfield case, fetch the return value
37841 from the registers before applying a single big-endian adjustment.
37842 Always do a final load for a BLKmode value not larger than a word.
37844 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37847 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37848 that we correctly handle column numbers greater than
37849 LINE_MAP_MAX_COLUMN_NUMBER.
37851 2017-01-10 Martin Sebor <msebor@redhat.com>
37853 PR middle-end/78245
37854 * gimple-ssa-sprintf.c (get_destination_size): Call
37855 {init,fini}object_sizes.
37856 * tree-object-size.c (addr_object_size): Adjust.
37857 (pass_through_call): Adjust.
37858 (pass_object_sizes::execute): Adjust.
37859 * tree-object-size.h (fini_object_sizes): Declare.
37861 2017-01-10 Martin Sebor <msebor@redhat.com>
37863 PR tree-optimization/78775
37864 * builtins.c (get_size_range): Move...
37865 * calls.c: ...to here.
37866 (alloc_max_size): Accept zero argument.
37867 (operand_signed_p): Remove.
37868 (maybe_warn_alloc_args_overflow): Call get_size_range.
37869 * calls.h (get_size_range): Declare.
37871 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37873 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37874 from TI's devices.csv file as of September 2016.
37875 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37877 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37879 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37880 * doc/invoke.texi: Likewise.
37881 * doc/md.texi: Likewise.
37882 * doc/objc.texi: Likewise.
37884 2017-01-10 Joshua Conner <joshconner@google.com>
37886 * config/arm/fuchsia-elf.h: New file.
37887 * config/fuchsia.h: New file.
37888 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37889 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37891 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37893 2016-01-10 Richard Biener <rguenther@suse.de>
37895 PR tree-optimization/79034
37896 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37897 Propagate out degenerate PHIs in the joiner.
37899 2017-01-10 Martin Liska <mliska@suse.cz>
37901 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37902 (sort_congruence_classes_by_decl_uid): Likewise.
37903 (sort_congruence_class_groups_by_decl_uid): Likewise.
37904 (sem_item_optimizer::merge_classes): Sort class, groups in these
37905 classes and members in the groups by DECL_UID of declarations.
37906 This would make merge operations stable.
37908 2017-01-10 Martin Liska <mliska@suse.cz>
37910 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37911 usage of m_classes_vec.
37912 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37913 (sem_item_optimizer::get_group_by_hash): Likewise.
37914 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37915 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37916 (sem_item_optimizer::verify_classes): Likewise.
37917 (sem_item_optimizer::process_cong_reduction): Likewise.
37918 (sem_item_optimizer::dump_cong_classes): Likewise.
37919 (sem_item_optimizer::merge_classes): Likewise.
37920 * ipa-icf.h (congruence_class_hash): Rename from
37921 congruence_class_group_hash. Remove declaration of m_classes_vec.
37923 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37925 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37926 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37927 * config.gcc: Add avx512vpopcntdqintrin.h.
37928 * config/i386/avx512vpopcntdqintrin.h: New.
37929 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37930 * config/i386/i386-builtin-types.def: Add new types.
37931 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37932 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37933 __builtin_ia32_vpopcountq_v8di_mask): New.
37934 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37935 __AVX512VPOPCNTDQ__.
37936 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37937 (PTA_AVX512VPOPCNTDQ): Define.
37938 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37939 TARGET_AVX512VPOPCNTDQ_P): Define.
37940 * config/i386/i386.opt: Add mavx512vpopcntdq.
37941 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37942 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37944 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37946 PR middle-end/77484
37947 * predict.def (PRED_CALL): Set to 67.
37949 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37951 * expr.c (store_field): In the bitfield case, if the value comes from
37952 a function call and is of an aggregate type returned in registers, do
37953 not modify the field mode; extract the value in all cases if the mode
37954 is BLKmode and the size is not larger than a word.
37956 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37959 * config/i386/cpuid.h: Fix undefined behavior.
37961 2017-01-04 Jeff Law <law@redhat.com>
37963 PR tree-optimization/79007
37964 PR tree-optimization/67955
37965 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37966 conservative for pt.null when flag_non_call_exceptions is on.
37968 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37970 PR translation/79019
37971 PR translation/79020
37972 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37973 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37974 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37976 * config/avr/avr.opt (maccumulate-args): Likewise.
37977 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37978 * common.opt (freport-bug): Likewise.
37979 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37980 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37981 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37982 translatable string.
37983 * config/i386/i386.c (function_value_32): Likewise.
37984 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37985 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37987 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37988 * common/config/msp430/msp430-common.c (msp430_handle_option):
37990 * symtab.c (symtab_node::verify_base): Likewise.
37991 * opts.c (set_debug_level): Likewise.
37992 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37993 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37994 missing whitespace to translatable strings.
37995 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37996 * config/sh/superh.h: Likewise.
37997 * config/i386/xopintrin.h: Likewise.
37998 * config/i386/znver1.md: Likewise.
37999 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38000 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38001 * double-int.h (struct double_int): Likewise.
38002 * double-int.c (div_and_round_double): Likewise.
38003 * wide-int.cc: Likewise.
38004 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38005 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38006 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38007 (crossjumps_occurred): ... this.
38008 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38011 PR tree-optimization/78899
38012 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38013 returning bool return struct loop *, NULL for failure and the new
38015 (versionable_outer_loop_p): Don't version outer loop if it has
38016 dont_vectorized bit set.
38017 (tree_if_conversion): When versioning outer loop, ensure
38018 tree_if_conversion is performed also on the inner loop of the
38019 non-vectorizable outer loop copy.
38020 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38021 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38022 prevent vectorization of it.
38023 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38024 the outer loop vectorization of the non-scalar version is attempted
38025 before vectorization of the inner loop in scalar version. If
38026 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38027 vectorization of its inner loop.
38028 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38029 has 2 inner loops, rename also on edges from bb whose single pred
38030 is outer_loop->header. Fix typo in function comment.
38032 2017-01-09 Martin Sebor <msebor@redhat.com>
38035 * asan.c (asan_emit_stack_protection): Increase local buffer size
38036 to avoid snprintf truncation warning.
38038 2017-01-09 Andrew Pinski <apinski@cavium.com>
38040 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38041 to reference thunderx2t99 for the tuning structure
38042 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38044 (thunderx2t99_extra_costs): This.
38045 * config/aarch64/aarch64-tune.md: Regenerate.
38046 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38047 (vulcan_addrcost_table): This.
38048 (vulcan_regmove_cost): Rename to ...
38049 (thunderx2t99_regmove_cost): This.
38050 (vulcan_vector_cost): Rename to ...
38051 (thunderx2t99_vector_cost): this.
38052 (vulcan_branch_cost): Rename to ...
38053 (thunderx2t99_branch_cost): This.
38054 (vulcan_tunings): Rename to ...
38055 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38056 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38058 2017-01-09 Martin Jambor <mjambor@suse.cz>
38062 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38063 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38064 (propagate_vr_accross_jump_function): Use the above function for all
38065 value range computations for pass-through jump functions and type
38066 converasion from explicit value range values.
38067 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38068 parameters from TYPE_ARG_TYPES.
38069 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38070 (ipa_write_node_info): Stream type of the actual argument.
38071 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38073 2017-01-09 Martin Liska <mliska@suse.cz>
38076 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38077 (lookup_compiler): Do not show error message with have_E.
38079 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38081 PR tree-optimization/78938
38082 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38083 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38084 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38085 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38088 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38090 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38093 2017-01-09 Richard Biener <rguenther@suse.de>
38095 PR tree-optimization/78997
38096 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38097 name condition properly.
38099 2017-01-09 Richard Biener <rguenther@suse.de>
38102 * dwarf2out.c (is_cxx): New overload with context.
38103 (is_naming_typedef_decl): Use it.
38105 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38107 * invoke.texi (Option Summary): Correct spacing in option lists
38108 and add line breaks to fix over-long lines.
38110 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38112 PR middle-end/17660
38114 * extend.texi (Common Variable Attributes): Add xref to GCC
38115 Internals manual to explain mode attribute keywords.
38117 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38120 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38121 and Preprocessor Options.
38122 (Options for Linking): Document -pthread here....
38123 (RS/6000 and PowerPC Options): ...not here.
38124 (Solaris 2 Options): ...or here.
38125 * doc/cppopts.texi: Document -pthread.
38127 2017-01-08 Martin Sebor <msebor@redhat.com>
38129 PR middle-end/77708
38130 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38131 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38132 New member functions.
38133 (format_directive): Used them.
38135 (pass_sprintf_length::handle_gimple_call): Same.
38136 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38137 to avoid truncation for any argument.
38138 (extract_affine_mul): Same.
38139 * tree.c (get_file_function_name): Same.
38141 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38143 PR middle-end/77484
38144 * predict.def (PRED_INDIR_CALL): Set to 86.
38146 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38148 PR preprocessor/54124
38149 * doc/cppopts.texi: Reformat -d subtable to list the full name
38150 of the options. Add cross-reference to the docs for the general
38151 compiler -d options.
38152 * doc/invoke.texi (Developer Options): Add cross-reference to the
38153 preprocessor-specific -d option documentation.
38155 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38157 PR preprocessor/13498
38158 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38159 redudant material, and reflect new command-line options.
38160 (System Headers): Likewise.
38162 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38164 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38165 -isystem, and -idirafter. Copy-edit.
38166 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38167 default for -ftrack-macro-expansion. Delete obsolete and
38168 badly-formatted implementation details about -fdebug-cpp output.
38169 * doc/cppwarnopts.texi: Copy-edit.
38171 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38174 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38175 that the transition from a max line width >= 1<<10 to narrower
38176 lines works correctly.
38178 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38180 * doc/options.texi (PerFunction): New.
38181 * opt-functions.awk (switch_flags): Map both Optimization and
38182 PerFunction to CL_OPTIMIZATION.
38183 * opth-gen.awk: Test for PerFunction flag along with
38185 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38186 it only when the latter is present. Skip those that don't in
38187 the hash function generator.
38188 * common.opt (fvar-tracking): Mark as PerFunction instead of
38190 (fvar-tracking-assignments): Likewise.
38191 (fvar-tracking-assignments-toggle): Likewise.
38192 (fvar-tracking-uninit): Likewise.
38194 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38196 PR translation/79018
38197 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38200 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38203 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38204 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38205 TARGET_LONG_JUMP_TABLE_OFFSETS.
38206 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38207 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38208 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38209 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38210 * config/m68k/m68k.md (tablejump expander): Likewise.
38211 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38212 TARGET_LONG_JUMP_TABLE_OFFSETS.
38213 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38214 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38216 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38217 David Holsgrove <david.holsgrove@xilinx.com>
38219 * common/config/microblaze/microblaze-common.c
38220 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38221 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38223 * config/microblaze/microblaze.c (microblaze_must_save_register)
38224 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38226 (microblaze_eh_return): New function.
38227 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38228 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38229 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38230 * config/microblaze/microblaze.md (eh_return): New pattern.
38232 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38234 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38235 GCC_DIAGNOSTIC_STRINGIFY): Define.
38237 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38239 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38241 * config/arm/arm.md (<mcrr>): New.
38243 * config/arm/arm.c (arm_arch5te): New.
38244 (arm_option_override): Set arm_arch5te.
38245 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38247 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38248 (arm_mcrr_qualifiers): ... this. New.
38249 (MRRC_QUALIFIERS): Define to...
38250 (arm_mrrc_qualifiers): ... this. New.
38251 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38253 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38254 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38255 (MRRCI, mrrc, MRRC): New.
38256 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38257 VUNSPEC_MRRC2): New.
38259 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38261 * config/arm/arm.md (<mcr>): New.
38263 * config/arm/arm.c (arm_coproc_builtin_available): Add
38264 support for mcr, mrc, mcr2 and mrc2.
38265 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38266 (arm_mcr_qualifiers): ... this. New.
38267 (MRC_QUALIFIERS): Define to ...
38268 (arm_mrc_qualifiers): ... this. New.
38269 (MCR_QUALIFIERS): Define to ...
38270 (arm_mcr_qualifiers): ... this. New.
38271 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38273 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38274 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38275 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38276 VUNSPEC_MRC2): New.
38278 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38280 * config/arm/arm.md (*ldc): New.
38284 * config/arm/arm.c (arm_coproc_builtin_available): Add
38285 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38286 (arm_coproc_ldc_stc_legitimate_address): New.
38287 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38288 'qualifier_const_pointer'.
38289 (LDC_QUALIFIERS): Define to...
38290 (arm_ldc_qualifiers): ... this. New.
38291 (STC_QUALIFIERS): Define to...
38292 (arm_stc_qualifiers): ... this. New.
38293 * config/arm/arm-protos.h
38294 (arm_coproc_ldc_stc_legitimate_address): New.
38295 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38296 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38297 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38298 stc2, stcl, stc2l): New.
38299 * config/arm/constraints.md (Uz): New.
38300 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38301 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38302 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38303 VUNSPEC_STC2L): New.
38305 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38307 * config/arm/arm.md (<cdp>): New.
38308 * config/arm/arm.c (neon_const_bounds): Rename this ...
38309 (arm_const_bounds): ... this.
38310 (arm_coproc_builtin_available): New.
38311 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38312 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38313 (CDP_QUALIFIERS): Define to...
38314 (arm_cdp_qualifiers): ... this. New.
38316 (arm_expand_builtin_args): Add case for 6 arguments.
38317 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38318 (arm_const_bounds): ... this.
38319 (arm_coproc_builtin_available): New.
38320 * config/arm/arm_acle.h (__arm_cdp): New.
38322 * config/arm/arm_acle_builtins.def (cdp): New.
38324 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38325 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38326 'arm_const_bounds'.
38327 * config/arm/types.md (coproc): New.
38328 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38329 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38330 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38331 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38333 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38335 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38336 (UBINOP_QUALIFIERS): New.
38338 (acle_builtin_data): New. Change comment.
38339 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38340 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38341 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38342 arm_acle_builtins.def.
38343 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38344 (arm_init_acle_builtins): New.
38345 (CRC32_BUILTIN): Remove.
38346 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38347 crc32cb, crc32ch and crc32cw.
38348 (arm_init_crc32_builtins): Remove.
38349 (arm_init_builtins): Use arm_init_acle_builtins rather
38350 than arm_init_crc32_builtins.
38351 (arm_expand_acle_builtin): New.
38352 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38353 * config/arm/arm_acle_builtins.def: New.
38355 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38357 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38358 (arm_builtin_datum): ... this.
38359 (arm_init_neon_builtin): Rename to ...
38360 (arm_init_builtin): ... this. Add a new parameters PREFIX
38361 and USE_SIG_IN_NAME.
38362 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38363 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38364 'arm_builtin_datum'.
38365 (arm_init_vfp_builtins): Likewise.
38366 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38367 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38368 (arm_expand_neon_args): Rename to ...
38369 (arm_expand_builtin_args): ... this. Rename builtin_arg
38370 enum values and differentiate between ARG_BUILTIN_MEMORY
38371 and ARG_BUILTIN_NEON_MEMORY.
38372 (arm_expand_neon_builtin_1): Rename to ...
38373 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38374 values, arm_expand_builtin_args and add bool parameter NEON.
38375 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38376 (arm_expand_vfp_builtin): Likewise.
38377 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38379 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38381 PR middle-end/77484
38382 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38383 * predict.c (tree_estimate_probability_bb): Reverse direction of
38384 polymorphic call predictor.
38386 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38388 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38389 (determine_pass_name_match): ...this new function and...
38390 (should_skip_pass_p): ...this new function.
38392 2017-01-06 Nathan Sidwell <nathan@acm.org>
38394 * ipa-visibility.c (function_and_variable_visibility): Reformat
38395 comments and long lines. Remove extrneous if.
38396 * symtab.c (symtab_node::make_decl_local): Fix code format.
38397 (symtab_node::set_section_for_node): Fix comment typo.
38399 2017-01-06 Martin Liska <mliska@suse.cz>
38402 * lra-constraints.c: Rename invariant to lra_invariant.
38403 * predict.c (set_even_probabilities): Initialize e to NULL.
38405 2017-01-05 Martin Sebor <msebor@redhat.com>
38407 PR tree-optimization/78910
38408 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38409 (format_integer): Correct off-by-one error in the handling
38410 of precision with negative numbers in signed conversions..
38412 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38414 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38416 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38418 PR tree-optimization/71016
38419 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38420 factor_out_conditional_conversion. Formatting fix.
38421 (factor_out_conditional_conversion): Add cond_stmt argument.
38422 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38423 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38426 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38428 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38429 read-rtl-function.o, and selftest-rtl.o.
38430 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38431 (selftest::aarch64_test_loading_full_dump): New function.
38432 (selftest::aarch64_run_selftests): New function.
38433 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38434 selftest::aarch64_run_selftests.
38435 * config/i386/i386.c
38436 (selftest::ix86_test_loading_dump_fragment_1): New function.
38437 (selftest::ix86_test_loading_call_insn): New function.
38438 (selftest::ix86_test_loading_full_dump): New function.
38439 (selftest::ix86_test_loading_unspec): New function.
38440 (selftest::ix86_run_selftests): Call the new functions.
38441 * emit-rtl.c (maybe_set_max_label_num): New function.
38442 * emit-rtl.h (maybe_set_max_label_num): New decl.
38443 * function.c (instantiate_decls): Guard call to
38444 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38445 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38447 * gensupport.c (gen_reader::gen_reader): Pass "false"
38448 for new "compact" param of rtx_reader.
38449 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38450 rather than an empty string for NULL strings.
38451 * read-md.c: Potentially include config.h rather than bconfig.h.
38452 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38453 (have_error): New global, copied from errors.c.
38454 (md_reader::read_name): Rename to...
38455 (md_reader::read_name_1): ...this, adding "out_loc" param,
38456 and converting "missing name or number" to returning false, rather
38458 (md_reader::read_name): Reimplement in terms of read_name_1.
38459 (md_reader::read_name_or_nil): New function.
38460 (md_reader::read_string): Handle "(nil)" by returning NULL.
38461 (md_reader::md_reader): Add new param "compact".
38462 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38463 (md_reader::read_file): New method.
38464 * read-md.h (md_reader::md_reader): Add new param "compact".
38465 (md_reader::read_file): New method.
38466 (md_reader::is_compact): New accessor.
38467 (md_reader::read_name): Convert return type from void to file_location.
38468 (md_reader::read_name_or_nil): New decl.
38469 (md_reader::read_name_1): New decl.
38470 (md_reader::m_compact): New field.
38471 (noop_reader::noop_reader): Pass "false" for new "compact" param
38473 (rtx_reader::rtx_reader): Add new "compact" param.
38474 (rtx_reader::read_rtx_operand): Make virtual and convert return
38475 type from void to rtx.
38476 (rtx_reader::read_until): New decl.
38477 (rtx_reader::handle_any_trailing_information): New virtual function.
38478 (rtx_reader::postprocess): New virtual function.
38479 (rtx_reader::finalize_string): New virtual function.
38480 (rtx_reader::m_in_call_function_usage): New field.
38481 (rtx_reader::m_reuse_rtx_by_id): New field.
38482 * read-rtl-function.c: New file.
38483 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38484 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38485 (selftest::verify_three_block_rtl_cfg): New decl.
38486 * read-rtl-function.h: New file.
38487 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38488 For host, include function.h, memmodel.h, and emit-rtl.h.
38489 (one_time_initialization): New function.
38490 (struct compact_insn_name): New struct.
38491 (compact_insn_names): New array.
38492 (find_code): Handle insn codes in compact dumps.
38493 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38494 (bind_subst_iter_and_attr): Likewise.
38495 (add_condition_to_string): Likewise.
38496 (add_condition_to_rtx): Likewise.
38497 (apply_attribute_uses): Likewise.
38498 (add_current_iterators): Likewise.
38499 (apply_iterators): Likewise.
38500 (initialize_iterators): Guard usage of apply_subst_iterator with
38501 #ifdef GENERATOR_FILE.
38502 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38503 (md_reader::read_mapping): Likewise.
38504 (add_define_attr_for_define_subst): Likewise.
38505 (add_define_subst_attr): Likewise.
38506 (read_subst_mapping): Likewise.
38507 (check_code_iterator): Likewise.
38508 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38510 (one_time_initialization): New function.
38511 (rtx_reader::read_until): New method.
38512 (read_flags): New function.
38513 (parse_reg_note_name): New function.
38514 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38515 Handle reuse_rtx ids.
38516 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38517 Add parsing support for RTL dumps, mirroring the special-cases in
38518 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38519 values, and calling handle_any_trailing_information.
38520 (rtx_reader::read_rtx_operand): Convert return type from void
38521 to rtx, returning return_rtx. Handle case 'e'. Call
38522 finalize_string on XSTR and XTMPL fields.
38523 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38524 "(nil)" values were omitted. Call the postprocess vfunc on the
38526 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38527 class ctor. Initialize m_in_call_function_usage. Call
38528 one_time_initialization.
38529 * rtl-tests.c (selftest::test_uncond_jump): Call
38530 set_new_first_and_last_insn.
38531 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38532 * selftest-rtl.c: New file.
38533 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38534 (selftest::get_insn_by_uid): New decl.
38535 * selftest-run-tests.c (selftest::run_tests): Call
38536 read_rtl_function_c_tests.
38537 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38538 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38541 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38543 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38544 operands in a special way. Assert that pos+len <= mode precision.
38546 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38548 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38549 3 argument Alias with unlimited for the negative form.
38550 (fno-vect-cost-model): Removed.
38552 2017-01-05 Martin Liska <mliska@suse.cz>
38554 * hsa-gen.c (gen_hsa_divmod): New function.
38555 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38557 2017-01-05 Martin Liska <mliska@suse.cz>
38560 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38563 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38565 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38566 small constant length operands.
38568 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38570 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38571 between loop iterations.
38573 2017-01-05 Martin Liska <mliska@suse.cz>
38576 * gimplify.c (gimplify_decl_expr): Compare to
38577 asan_poisoned_variables instread of checking flags.
38578 (gimplify_target_expr): Likewise.
38579 (gimplify_expr): Likewise.
38580 (gimplify_function_tree): Conditionally initialize
38581 asan_poisoned_variables.
38583 2017-01-04 Jeff Law <law@redhat.com>
38585 PR tree-optimizatin/78812
38586 * rtl.h (contains_mem_rtx_p): Prototype.
38587 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38588 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38589 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38590 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38591 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38593 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38595 * input.c (assert_char_at_range): Default-initialize actual_range.
38597 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38599 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38600 to match the caller.
38602 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38604 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38605 insns after final jump in test to emit dummy move.
38607 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38609 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38610 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38612 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38614 * multiple_target.c (create_dispatcher_calls): Init e_next.
38615 * tree-ssa-loop-split.c (split_loop): Init border.
38616 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38619 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38624 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38625 (altivec_register_operand): Do not return true if the operand
38626 contains a SUBREG mixing SImode and SFmode.
38627 (vsx_register_operand): Likewise.
38628 (vsx_reg_sfsubreg_ok): New predicate.
38629 (vfloat_operand): Do not return true if the operand contains a
38630 SUBREG mixing SImode and SFmode.
38631 (vint_operand): Likewise.
38632 (vlogical_operand): Likewise.
38633 (gpc_reg_operand): Likewise.
38634 (int_reg_operand): Likewise.
38635 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38636 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38637 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38639 (rs6000_emit_move_si_sf_subreg): New helper function.
38640 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38641 fixup SUBREGs involving SImode and SFmode.
38642 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38643 numbers for the new peephole2 optimization.
38644 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38645 the GLIBC math library that do AND/IOR/XOR operations on single
38646 precision floating point.
38647 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38648 target macros to say whether we need to avoid SUBREGs mixing
38650 (TARGET_ALLOW_SF_SUBREG): Likewise.
38651 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38652 (UNSPEC_SI_FROM_SF): Likewise.
38653 (iorxor): Change spacing.
38654 (and_ior_xor): New iterator for AND, IOR, and XOR.
38655 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38656 (movdi_from_sf_zero_ext): Likewise.
38657 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38658 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38659 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38660 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38661 (fms<mode>4): Likewise.
38662 (fnma<mode>4): Likewise.
38663 (fnms<mode>4): Likewise.
38664 (nfma<mode>4): Likewise.
38665 (nfms<mode>4): Likewise.
38667 2017-01-04 Marek Polacek <polacek@redhat.com>
38670 * doc/invoke.texi: Document -Wpointer-compare.
38672 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38674 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38677 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38678 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38681 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38684 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38685 documentation of the powerpc_popcntb_ok attribute.
38686 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38687 code to issue warning messages if a requested CPU configuration is
38688 not supported by the binary (assembler and loader) toolchain.
38689 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38690 made to define a built-in function that has been disabled.
38691 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38692 made to define a built-in function that has been disabled.
38693 (altivec_init_builtins): Add comment explaining why definition
38694 of the DST built-in functions is not preceded by an assertion
38695 check. Add assertions to prevent ICE if attempts are made to
38696 define an altivec predicate or an abs* built-in function that has
38698 (htm_init_builtins): Add comment explaining why definition of the
38699 htm built-in functions is not preceded by an assertion check.
38701 2017-01-04 Jeff Law <law@redhat.com>
38703 PR tree-optimizatin/67955
38704 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38705 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38706 the points-to solution does not include pt_null. Use DECL_PT_UID
38709 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38711 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38712 Use gen_int_mode instead of gen_lopwart for const_int operands.
38714 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38716 PR tree-optimization/71563
38717 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38718 out of range value - has low bits known to be zero.
38720 2017-01-04 Alan Modra <amodra@gmail.com>
38722 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38723 * configure: Regenerate.
38724 * config.in: Regenerate.
38726 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38729 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38730 a substring of the message, but strcmp with the whole message. Ifdef
38731 ENABLE_NLS, translate the message first using dgettext.
38733 2017-01-03 Jeff Law <law@redhat.com>
38735 PR tree-optimizatin/78856
38736 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38737 (mark_threaded_blocks): Remove code to truncate thread paths that
38738 cross multiple loop headers. Instead invalidate the cached loop
38739 iteration information and handle case of a thread path walking
38740 into an irreducible region.
38742 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38745 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38746 assertions. Add support for doing the signbit if the IEEE 128-bit
38747 floating point value is in a GPR.
38748 * config/rs6000/rs6000.md (Fsignbit): Delete.
38749 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38750 Update the length attribute if the value is in a GPR.
38751 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38752 the sign or zero extension instruction, since the value is always 0/1.
38753 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38756 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38757 extracting SImode to a GPR register so that we can generate a
38758 store, limit the vector to be in a traditional Altivec register
38759 for the vextuwrx instruction.
38761 2017-01-03 Ian Lance Taylor <iant@google.com>
38763 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38765 2017-01-03 Martin Sebor <msebor@redhat.com>
38767 PR tree-optimization/78696
38768 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38769 precision. Use MPFR for %f for greater fidelity. Correct handling
38771 (pass_sprintf_length::compute_format_length): Set width and precision
38772 specified by asrerisk to void_node for vararg functions.
38773 (try_substitute_return_value): Adjust dump output.
38775 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38777 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38779 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38781 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38782 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38783 -mlra/-mno-lra was passed to the compiler.
38785 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38787 PR rtl-optimization/65618
38788 * emit-rtl.c (try_split): Move initialization of "before" and
38789 "after" to just before the call to emit_insn_after_setloc.
38791 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38793 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38795 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38797 * dwarf2out.c (gen_enumeration_type_die): When
38798 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38800 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38802 PR tree-optimization/78965
38803 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38804 Change first argument from const call_info & to call_info &. For %n
38805 set info.nowrite to false.
38807 PR middle-end/78901
38808 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38809 possibly throwing calls.
38811 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38812 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38813 and fns handling, rather than in a separate case SSA_NAME.
38815 2017-01-02 Jeff Law <law@redhat.com>
38817 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38818 fixes for first_period and second_period variables.
38820 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38823 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38824 (*insvqi_1): New insn pattern.
38825 (*insvqi_1_mem_rex64): Ditto.
38826 (*insvqi_2): Ditto.
38827 (*insvqi_3): Rename from *insvqi.
38829 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38831 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38833 * doc/cfg.texi (Edges): Remove reference to Java.
38834 (Maintaining the CFG): Ditto.
38836 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38838 PR middle-end/77674
38839 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38840 transparent aliases.
38842 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38844 PR middle-end/77484
38845 * predict.def (PRED_CALL): Update hitrate.
38846 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38847 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38848 into direct/indirect/polymorphic variants.
38850 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38852 Update copyright years.
38854 * gcc.c (process_command): Update copyright notice dates.
38855 * gcov-dump.c (print_version): Ditto.
38856 * gcov.c (print_version): Ditto.
38857 * gcov-tool.c (print_version): Ditto.
38858 * gengtype.c (create_file): Ditto.
38859 * doc/cpp.texi: Bump @copying's copyright year.
38860 * doc/cppinternals.texi: Ditto.
38861 * doc/gcc.texi: Ditto.
38862 * doc/gccint.texi: Ditto.
38863 * doc/gcov.texi: Ditto.
38864 * doc/install.texi: Ditto.
38865 * doc/invoke.texi: Ditto.
38867 Copyright (C) 2017 Free Software Foundation, Inc.
38869 Copying and distribution of this file, with or without modification,
38870 are permitted in any medium without royalty provided the copyright
38871 notice and this notice are preserved.