1 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3 * doc/invoke.texi: Document -Wcast-function-type.
4 * recog.h (stored_funcptr): Change signature.
5 * tree-dump.c (dump_node): Avoid warning.
6 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
8 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
11 * gimple-fold.c (get_range_strlen): Add the handling of non-member
14 2017-12-14 David Malcolm <dmalcolm@redhat.com>
16 PR tree-optimization/83312
17 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
18 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
19 "val" param, and to cope with arbitrary basic blocks.
20 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
21 handle NULL_TREE for "val", dropping "bb" param.
22 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
23 drop "bb" param. Handle NULL_TREE for "val".
24 (find_case_label_for_value): Make "switch_stmt" param const.
25 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
27 (vrp_prop::check_all_array_refs): Reimplement as...
28 (check_array_bounds_dom_walker::before_dom_children): ...this new
29 vfunc. Replace linear search through BB block list, excluding
30 those with non-executable in-edges via dominator walk.
32 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34 * config/arm/arm.opt (mverbose-cost-dump): New option.
35 * config/arm/arm.c (arm_rtx_costs): Use it.
37 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
40 * reload1.c (emit_input_reload_insns): Skip debug markers.
42 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
44 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
48 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
49 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
50 * config/sh/sh.c: Skip debug insns besides notes.
51 * config/sh/sh.md: Likewise.
52 * config/sh/sh_treg_combine.cc: Likewise.
53 * config/sh/sync.md: Likewise.
55 2017-12-14 Tom de Vries <tom@codesourcery.com>
57 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
58 Add item for weak_undefined.
60 2017-12-14 Richard Biener <rguenther@suse.de>
62 PR tree-optimization/67842
63 * sese.h (bb_in_region): Remove #if 0'ed code.
65 2017-12-14 Richard Biener <rguenther@suse.de>
67 PR tree-optimization/83326
68 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
69 may_be_zero parameter and handle it by not marking the first
70 peeled copy as not exiting the loop.
71 (try_peel_loop): Likewise.
72 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
73 to handle the case of constant or zero iterations and perform
74 loop header copying on-the-fly.
76 2017-12-14 Richard Biener <rguenther@suse.de>
78 PR tree-optimization/83418
79 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
80 Instead of asserting we don't get unfolded comparisons deal with
83 2017-12-14 Jakub Jelinek <jakub@redhat.com>
86 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
87 basic blocks. Assert debug bind insns don't appear outside of bbs,
88 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
91 PR tree-optimization/83198
92 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
93 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
94 value if arg is a REAL_CST with incompatible type.
96 2017-12-14 Sudakshina Das <sudi.das@arm.com>
97 Bin Cheng <bin.cheng@arm.com>
100 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
102 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
105 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
107 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
108 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
110 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
111 Alan Hayward <alan.hayward@arm.com>
112 David Sherwood <david.sherwood@arm.com>
114 * poly-int.h: New file.
115 * poly-int-types.h: Likewise.
116 * coretypes.h: Include them.
117 (POLY_INT_CONVERSION): Define.
118 * target.def (estimated_poly_value): New hook.
119 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
120 * doc/tm.texi: Regenerate.
121 * doc/poly-int.texi: New file.
122 * doc/gccint.texi: Include it.
123 * doc/rtl.texi: Describe restrictions on subreg modes.
124 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
125 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
126 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
127 * targhooks.h (default_estimated_poly_value): Declare.
128 * targhooks.c (default_estimated_poly_value): New function.
129 * target.h (estimated_poly_value): Likewise.
130 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
131 (wi::unary_traits): Delete.
132 (wi::binary_traits::signed_shift_result_type): Define for
133 offset_int << HOST_WIDE_INT, etc.
134 (generic_wide_int::operator <<=): Define for all types and use
135 wi::lshift instead of <<.
136 (wi::hwi_with_prec): Add a default constructor.
137 (wi::ints_for): New class.
138 (operator <<): Define for all wide-int types.
139 (operator /): New function.
140 (operator %): Likewise.
141 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
142 (ASSERT_MAYBE_NE_AT): New macros.
144 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
145 Dominik Vogt <vogt@linux.vnet.ibm.com>
148 * emit-rtl.c (init_emit): Remove ??? comment.
149 * explow.c (get_dynamic_stack_size): Take known alignment of stack
150 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
151 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
152 alignment of 3 virtual registers to BITS_PER_WORD.
154 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
156 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
158 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
159 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
161 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
164 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
166 2017-12-13 Jakub Jelinek <jakub@redhat.com>
168 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
169 with the exception of debug begin stmt markers appear before
173 * final.c (rest_of_handle_final): Call variable_tracking_main only
174 if !flag_var_tracking.
176 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
177 Jakub Jelinek <jakub@redhat.com>
181 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
182 labels debug stmts that can only appear after labels.
184 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
186 PR rtl-optimization/82398
187 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
188 EXPR_USEFULNESS in priority comparison.
190 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
192 PR rtl-optimization/83393
193 * combine.c (move_deaths): If reg_stat points to a too new insn in
194 last_death, do not use it: find the proper insn instead.
196 2017-12-12 Jeff Law <law@redhat.com>
198 PR tree-optimization/83298
199 PR tree-optimization/83362
200 PR tree-optimization/83383
201 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
202 push_value_range a public interface. Add new argument to
203 record_ranges_from_stmt.
204 * gimple-ssa-evrp-analyze.c
205 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
206 Update comments. Handle recording temporary equivalences.
207 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
208 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
209 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
210 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
211 gimple-ssa-evrp-analyze.h.
212 (record_temporary_equivalences_from_phis): Add new argument. When
213 the PHI arg is an SSA_NAME, set the result's range to the range
215 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
217 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
218 Pass it down to children as needed.
219 (thread_outgoing_edges): Likewise.
220 (thread_across_edge): Likewise. Push/pop range state as needed.
221 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
223 2017-12-12 Julia Koval <julia.koval@intel.com>
225 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
226 (PTA_CANNONLAKE): Remove PTA_CLWB.
228 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
231 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
232 (vcondv2div2df): Likewise.
233 (vconduv2dfv2di): Likewise.
235 2017-12-12 Jakub Jelinek <jakub@redhat.com>
237 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
238 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
239 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
240 ATTR_MATHFN_FPROUNDING_ERRNO.
242 2017-12-12 Richard Biener <rguenther@suse.de>
244 PR tree-optimization/83385
245 * graphite-scop-detection.c (get_order, order): Remove.
246 (bb_to_rpo): New global.
248 (build_scops): Sort pbbs in RPO order.
250 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
252 * combine.c (simplify_set): Do not transform subregs to zero_extends
253 if the destination mode is a vector mode.
255 2017-12-12 Jakub Jelinek <jakub@redhat.com>
257 PR tree-optimization/83359
258 * tree-cfg.h (fold_loop_internal_call): Declare.
259 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
260 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
261 (find_loop_dist_alias): New function.
262 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
263 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
266 PR tree-optimization/80631
267 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
268 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
269 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
270 hardcoding zero as the value if COND_EXPR is never true. For
271 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
272 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
273 hardcoding MAX_EXPR as the reduction operation.
274 (is_nonwrapping_integer_induction): Allow negative step.
275 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
276 vect_create_epilog_for_reduction, if no value is suitable, don't
277 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
279 2017-12-12 Richard Biener <rguenther@suse.de>
281 PR tree-optimization/81889
282 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
283 range info from the non-wrapping IV instead of just the range
286 2017-12-12 Julia Koval <julia.koval@intel.com>
288 * config.gcc: Add vaesintrin.h.
289 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
291 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
292 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
294 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
295 * config/i386/immintrin.h: Include vaesintrin.h.
296 * config/i386/sse.md (vaesdec_<mode>): New pattern.
297 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
298 _mm_aesdec_epi128): New intrinsics.
300 2017-12-12 Julia Koval <julia.koval@intel.com>
302 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
303 OPTION_MASK_ISA_VAES_UNSET): New.
304 (ix86_handle_option): Handle -mvaes.
305 * config/i386/cpuid.h: Define bit_VAES.
306 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
307 * config/i386/i386-c.c (__VAES__): New.
308 * config/i386/i386.c (ix86_target_string): Add -mvaes.
309 (ix86_valid_target_attribute_inner_p): Ditto.
310 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
311 * config/i386/i386.opt: Add -mvaes.
312 * doc/invoke.texi: Ditto.
314 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
316 * debug.h (gcc_debug_hooks): Add inline_entry.
317 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
318 * debug.c (do_nothing_debug_hooks): Likewise.
319 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
320 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
321 (dwarf2_lineno_debug_hooks): Likewise.
323 * common.opt (gstatement-frontiers): New, setting
324 debug_nonbind_markers_p.
325 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
326 * toplev.c (process_options): Autodetect value for debug statement
328 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
329 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
331 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
332 markers. Integrate source bind into debug stmt expand loop.
333 (pass_expand::execute): Check debug marker limit. Avoid deep
334 TER and expand debug locations for debug bind insns only.
335 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
337 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
338 * final.c (reemit_insn_block_notes): Take current block from
339 nonbind markers. Declare note where it's first set.
340 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
341 begin stmt markers if enabled.
342 (notice_source_line): Handle nonbind markers. Fail if their
343 location is unknown or that of builtins.
344 (rest_of_handle_final): Convert begin stmt markers to notes if
345 var-tracking didn't run.
346 (rest_of_clean_state): Skip begin stmt markers.
347 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
349 * function.c (allocate_struct_function): Set begin_stmt_markers.
350 * function.h (struct function): Add debug_marker_count counter
351 and debug_nonbind_markers flag.
352 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
353 * gimple-low.c (lower_function_body): Adjust
354 debug_nonbind_markers.
355 (lower_stmt): Drop or skip gimple debug stmts.
356 (lower_try_catch): Skip debug stmts.
357 * gimple.c (gimple_build_debug_begin_stmt): New.
358 (gimple_copy): Increment debug_marker_count if copying one.
359 * gimple.h (gimple_build_debug_begin_stmt): Declare.
360 * gimplify.c (rexpr_location): New.
361 (rexpr_has_location): New.
362 (warn_switch_unreachable_r): Handle gimple debug stmts.
363 (shortcut_cond_r): Call expr_location.
365 (find_goto_label): New.
366 (shortcut_cond_expr): Call expr_has_location, expr_location, and
368 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
370 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
371 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
372 (LANG_HOOKS_INITIALIZER): ... this.
373 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
374 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
376 (update_ebb_live_info): Skip debug insn markers.
377 * lra.c (debug_insn_static_data): Rename to...
378 (debug_bind_static_data): ... this.
379 (debug_marker_static_data): New.
380 (lra_set_insn_recog_data): Select one of the above depending
382 (lra_update_isn_regno_info): Don't assume debug insns have
384 (push_insns): Skip debug insns.
385 * lto-streamer-in.c (input_function): Drop debug stmts
386 depending on active options. Adjust debug_nonbind_markers.
387 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
388 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
389 begin stmt marker notes.
390 (print_insn): Likewise.
391 * recog.c (extract_insn): Recognize rtl for debug markers.
392 * rtl.def (DEBUG_MARKER): New.
393 * tree-inline.c: Include params.h.
394 (remap_gimple_stmt): Handle nonbind markers.
395 (maybe_move_debug_stmts_to_successors): Likewise.
396 (copy_debug_stmt): Likewise.
397 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
398 markers regardless of no side effects.
399 (tsi_link_before): Don't update container's side effects when adding
401 (tsi_link_after): Likewise.
402 (expr_first): Skip begin stmt markers.
403 (expr_last): Likewise.
404 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
405 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
406 Disregard nonbind markers.
407 * tree.c (make_node_stat): Don't set side effects for begin stmt
409 (build1_stat): Likewise.
410 * tree.def (DEBUG_BEGIN_STMT): New.
411 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
412 * var-tracking.c (delete_debug_insns): Renamed to...
413 (delete_vta_debug_insns): ... this.
414 (reemit_marker_as_note): New.
415 (vt_initialize): Reemit markers.
416 (delete_vta_debug_insns): Likewise.
417 (vt_debug_insns_local): Reemit or delete markers.
418 (variable_tracking_main_1): Likewise.
419 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
420 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
421 (gimple_debug_nonbind_marker_p): New.
422 (gimple_build_debug_bind): Adjust.
423 (gimple_build_debug_begin_stmt): New.
424 * doc/invoke.texi (max-debug-marker-count): New param.
425 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
426 (debug_parameter_ref, debug_marker): New.
427 (NOTE_INSN_BEGIN_STMT): New.
428 (DEBUG_INSN): Describe begin stmt markers.
430 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
431 without debug insns, we wouldn't, but clean up debug insns
432 after a control flow insn nevertheless.
434 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
435 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
436 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
437 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
438 (rtl_tidy_fallthru_edge): Likewise.
439 (rtl_verify_fallthru): Likewise.
440 (rtl_verify_bb_layout): Likewise.
441 (skip_insns_after_block): Likewise.
442 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
443 * dwarf2out.c: Include print-rtl.h.
444 (dwarf2out_next_real_insn): New.
445 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
446 Dump debug binds in asm comments.
447 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
448 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
449 callers to use gsi_start_nondebug_bb instead.
450 (gsi_after_labels): Skip gimple debug stmts.
451 (gsi_start_nondebug): New.
452 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
453 (proper_loop_form_for_interchange): Adjust.
454 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
455 * gimple.h (gimple_seq_last_nondebug_stmt): New.
456 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
457 (collect_fallthrough_labels): Likewise.
458 (should_warn_for_implicit_fallthrough): Likewise.
459 (warn_implicit_fallthrough_r): Likewise.
460 (expand_FALLTHROUGH_r): Likewise.
461 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
462 (graphite_copy_stmts_from_block): Skip nonbind markers.
463 * haifa-sched.c (sched_extend_bb): Skip debug insns.
464 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
465 * jump.c (clean_barriers): Skip debug insns.
466 * omp-expand.c (expand_parallel_call): Skip debug insns.
467 (expand_task_call): Likewise.
468 (remove_exit_barrier): Likewise.
469 (expand_omp_taskreg): Likewise.
470 (expand_omp_for_init_counts): Likewise.
471 (expand_omp_for_generic): Likewise.
472 (expand_omp_for_static_nochunk): Likewise.
473 (expand_omp_for_static_chunk): Likewise.
474 (expand_omp_simd): Likewise.
475 (expand_omp_taskloop_for_outer): Likewise.
476 (expand_omp_taskloop_for_inner): Likewise.
477 (expand_oacc_for): Likewise.
478 (expand_omp_sections): Likewise.
479 (expand_omp_single): Likewise.
480 (expand_omp_synch): Likewise.
481 (expand_omp_atomic_load): Likewise.
482 (expand_omp_atomic_store): Likewise.
483 (expand_omp_atomic_fetch_op): Likewise.
484 (expand_omp_atomic_pipeline): Likewise.
485 (expand_omp_atomic_mutex): Likewise.
486 (expand_omp_target): Likewise.
487 (grid_expand_omp_for_loop): Likewise.
488 (grid_expand_target_grid_body): Likewise.
489 (build_omp_regions_1): Likewise.
490 * omp-low.c (check_combined_parallel): Skip debug stmts.
491 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
492 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
493 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
495 * tree-cfg.c (make_blobs_1): Skip debug stmts.
496 (make_edges): Likewise.
497 (cleanup_dead_labels): Likewise.
498 (gimple_can_merge_blocks_p): Likewise.
499 (stmt_starts_bb_p): Likewise.
500 (gimple_block_label): Likewise.
501 (gimple_redirect_edge_and_branch): Likewise.
502 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
504 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
506 * tree-pretty-print (print_declaration): Omit initializer in slim
508 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
510 (eliminate_unnecessary_stmts): Stabilize block removal order.
511 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
512 * var-tracking.c (get_first_insn): New.
513 (vt_emit_notes): Call it.
514 (vt_initialize): Walk any insns before the first BB.
515 (delete_debug_insns): Likewise.
517 * gimple.h (enum gimple_debug_subcode): Add
518 GIMPLE_DEBUG_BEGIN_STMT.
519 (gimple_debug_begin_stmt_p): New.
520 (gimple_debug_nonbind_marker_p): New.
521 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
522 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
523 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
524 * insn-notes.def (BEGIN_STMT): New.
525 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
526 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
527 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
528 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
529 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
530 (INSN_DEBUG_MARKER_KIND): New.
531 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
532 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
533 (INSN_VAR_LOCATION_PTR): New.
534 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
536 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
537 deps only if debug bind insns are enabled.
538 (pass_expand::execute): Avoid deep TER and expand
539 debug locations for debug bind insns only.
540 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
541 debug stmts special handling down to debug bind stmts.
542 * combine.c (try_combine): Narrow debug insns special handling
543 down to debug bind insns.
544 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
545 Narrow debug insns preexisting special handling down to debug
547 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
548 handling down to debug bind insns.
549 * function.c (instantiate_virtual_regs): Skip debug markers,
550 adjust handling of debug binds.
551 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
552 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
553 * haifa-sched.c (schedule_insn): Narrow special handling of debug
554 insns to debug bind insns.
555 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
556 special handling of debug stmts to debug bind stmts.
557 * ipa-split.c (split_function): Likewise.
558 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
559 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
561 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
562 * regrename.c (build_def_use): Likewise.
563 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
564 (pass_cprop_hardreg): Narrow special casing of debug insns to
566 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
567 * reload1.c (reload): Likewise.
568 * sese.c (sese_insert_phis_for_liveouts): Narrow special
569 casing of debug stmts to debug bind stmts.
570 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
571 * ssa-iterators.h (num_imm_uses): Likewise.
572 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
573 debug stmts to debug bind stmts.
574 * tree-inline.c (tree_function_versioning): Narrow special casing
575 of debug stmts to debug bind stmts.
576 * tree-loop-distribution.c (generate_loops_for_partition):
577 Narrow special casing of debug stmts to debug bind stmts.
578 * tree-sra.c (analyze_access_subtree): Narrow special casing
579 of debug stmts to debug bind stmts.
580 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
581 stmts to debug bind stmts.
582 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
583 casing of debug stmts to debug bind stmts.
584 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
585 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
586 casing of debug stmts to debug bind stmts.
587 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
589 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
590 debug stmts to debug bind stmts.
591 (gimple_replace_ssa_lhs): Likewise.
592 (insert_debug_temp_for_var_def): Likewise.
593 (insert_debug_temps_for_defs): Likewise.
594 (reset_debug_uses): Likewise.
595 * tree-ssanames.c (release_ssa_name_fn): Likewise.
596 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
597 (adjust_debug_stmts): Likewise.
598 (adjust_phi_and_debug_stmts): Likewise.
599 (vect_do_peeling): Likewise.
600 * tree-vect-loop.c (vect_transform_loop): Likewise.
601 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
602 * var-tracking.c (adjust_mems): Narrow special casing of debug
603 insns to debug bind insns.
604 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
605 (compute_bb_dataflow, vt_find_locations): Likewise.
606 (vt_expand_loc, emit_notes_for_changes): Likewise.
607 (vt_init_cfa_base): Likewise.
608 (vt_emit_notes): Likewise.
609 (vt_initialize): Likewise.
610 (vt_finalize): Likewise.
612 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
613 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
614 (next_nonnote_nondebug_insn_bb): New.
615 (prev_nonnote_nondebug_insn_bb): New.
616 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
617 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
618 (prev_nonnote_nondebug_insn_bb): Declare.
619 (next_nonnote_nondebug_insn_bb): Declare.
620 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
621 * cfgrtl.c (get_last_bb_insn): Likewise.
622 * lra.c (push_insns): Likewise.
624 2017-12-11 David Malcolm <dmalcolm@redhat.com>
627 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
629 * selftest.c (selftest::test_runner::test_runner): New ctor.
630 (selftest::test_runner::~test_runner): New dtor.
631 * selftest.h (class selftest::test_runner): New class.
633 2017-12-11 Carl Love <cel@us.ibm.com>
635 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
636 vec_extract_fp32_from_shortl]): Add #defines.
637 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
638 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
639 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
640 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
641 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
642 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
643 * doc/extend.texi: Add documentation for the added builtins.
645 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
647 PR rtl-optimization/80693
648 PR rtl-optimization/81019
649 PR rtl-optimization/81020
650 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
651 are not mentioned in i3. Place the REG_UNUSED note on i2,
652 possibly modified to REG_DEAD, if it did not originate in i3.
654 2017-12-11 Jakub Jelinek <jakub@redhat.com>
656 * recog.c (store_data_bypass_p_1): New function.
657 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
658 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
660 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
662 PR rtl-optimization/83361
663 * ifcvt.c (if_convert): Call fixup_partitions.
665 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
667 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
668 early folding of splat_u{8,16,32}.
670 2017-12-11 Jakub Jelinek <jakub@redhat.com>
672 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
673 output_operand_lossage first argument with capital letter.
674 (aarch64_override_options): Don't start error and sorry first argument
677 2017-12-11 Andi Kleen <ak@linux.intel.com>
679 PR gcov-profile/83355
680 * auto-profile.c (string_table::get_index_by_decl): Don't
681 recurse when abstract origin points to itself.
683 2017-12-11 Bin Cheng <bin.cheng@arm.com>
685 PR tree-optimization/83320
686 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
687 (prune_datarefs_not_in_loop): Ditto.
689 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
691 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
693 2017-12-10 Jakub Jelinek <jakub@redhat.com>
695 PR tree-optimization/83337
696 * gimple-loop-interchange.cc (compute_access_stride): Handle
697 bitfield DRs properly.
699 2017-12-09 Jakub Jelinek <jakub@redhat.com>
701 PR tree-optimization/83338
702 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
703 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
704 vector of unsigned integers to vector of signed integers.
706 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
708 PR rtl-optimization/83317
709 * lra-constraints.c (process_address_1): Add insn code check.
711 2017-12-08 Michael Matz <matz@suse.de>
713 Fix PR tree-optimization/83323
714 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
715 head-controlled loops and loop BBs.
716 * common.opt (funroll-and-jam): Remove, instead ...
717 (floop-unroll-and-jam): ... reuse this option.
718 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
719 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
720 (-floop-unroll-and-jam): ... this option.
722 2017-12-08 Jakub Jelinek <jakub@redhat.com>
724 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
725 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
726 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
728 2017-12-08 Richard Biener <rguenther@suse.de>
731 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
732 handle accesses outside of zero-sized vars.
734 2017-12-08 Martin Jambor <mjambor@suse.cz>
736 PR tree-optimization/83141
737 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
738 test for MEM_REFs implicitely changing types with padding. Remove
740 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
742 2017-12-08 Jakub Jelinek <jakub@redhat.com>
744 * config/arc/arc.c (arc_attribute_table): Add exclusions to
746 * config/avr/avr.c (avr_attribute_table): Likewise.
747 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
748 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
749 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
750 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
751 of struct attribute_spec.
752 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
754 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
757 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
759 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
762 * profile-count.c (profile_count::from_gcov_type): Move from
763 profile-count.h; handle overflow.
764 * profile-count.h (profile_count::from_gcov_type): Move offline.
766 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
768 PR rtl-optimization/83304
769 * combine.c (move_deaths): If we do not know where a register died,
772 2017-12-08 Richard Biener <rguenther@suse.de>
774 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
775 Provide -fopt-info-loop feedback when we interchange in a nest.
777 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
779 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
780 for armv6 ARM CPU IDs.
782 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
784 * common/config/arm/arm-common.c: Include <algorithm>.
785 (INCLUDE_VECTOR): Define.
786 (compare_opt_names): New function.
787 (arm_rewrite_selected_arch): Only strip out extensions that can be
788 expressed through -mfpu. Sort the remaining extensions
791 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
793 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
794 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
795 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
796 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
798 2017-06-08 Tristan Gingold <gindold@adacore.com>
801 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
802 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
804 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
806 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
807 of struct attribute_spec.
809 2017-12-08 Julia Koval <julia.koval@intel.com>
811 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
812 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
813 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
814 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
815 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
816 _mm_maskz_dpwssds_epi32): Ditto.
818 2017-12-08 Richard Biener <rguenther@suse.de>
820 PR tree-optimization/81303
821 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
822 conditions try to create a comparison vector type matching
823 the data vector type.
824 (vectorizable_condition): Adjust.
825 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
826 Leave invariant conditions alone in case we can vectorize those.
828 2017-12-08 Julia Koval <julia.koval@intel.com>
830 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
831 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
832 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
833 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
834 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
836 2017-12-08 Julia Koval <julia.koval@intel.com>
838 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
839 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
840 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
841 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
842 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
843 _mm_maskz_dpbusds_epi32): New intrinsics.
845 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
847 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
849 (ld<bh>io_signed): Likewise.
850 (st<bhw>io): Likewise.
851 * config/nios2/predicates.md (ldstio_memory_operand): Allow
852 SMALL_INT12 constant integer operand.
854 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
856 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
858 (nios2_large_constant_allowed): ...this. Adjust uses.
859 (nios2_plus_symbolic_constant_p): Rename to...
860 (nios2_plus_large_constant_p): ...this. Adjust uses.
861 (nios2_legitimate_address_p): Correct CONST_INT handling.
862 (nios2_symbolic_memory_operand_p): Rename to...
863 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
864 (nios2_large_constant_p): Check for large constant integers too.
865 (nios2_split_large_constant): Handle constant integers.
866 (nios2_split_symbolic_memory_operand): Rename to...
867 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
868 (nios2_legitimize_constant_address): Handle constant integers.
869 (r0rel_constant_p): Handle small constant integers.
870 (nios2_print_operand_address): Handle r0-relative integer addresses.
871 * config/nios2/nios2-protos.h: Adjust for renamed functions.
872 * config/nios2/nios2.md: Adjust for renamed functions.
874 2017-12-07 Andrew Waterman <andrew@sifive.com>
876 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
877 (TARGET_HAVE_SRODATA_SECTION): New define.
878 (riscv_select_section): New function.
880 2017-12-08 Joseph Myers <joseph@codesourcery.com>
881 Alexander Monakov <amonakov@ispras.ru>
882 Jakub Jelinek <jakub@redhat.com>
885 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
887 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
889 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
890 a bool success value. Don't call output_operand_lossage here.
891 (aarch64_print_ldpstp_address): Return a bool success value.
892 (aarch64_print_operand_address): Call output_addr_const if
893 aarch64_print_address_internal fails.
894 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
895 'y'; call output_operand_lossage instead. Call output_operand_lossage
896 if aarch64_print_ldpstp_address fails.
898 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
900 * tree-vector-builder.h
901 (tree_vector_builder::binary_encoded_nelts): Declare.
902 * tree-vector-builder.c
903 (tree_vector_builder::binary_encoded_nelts): New function.
904 * fold-const.c (negate_expr_p): Likewise.
905 (operand_equal_p, fold_checksum_tree): Likewise.
906 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
907 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
908 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
909 (uniform_vector_p): Likewise.
910 * varasm.c (const_hash_1, compare_constant): Likewise.
911 * tree-ssa-ccp.c: Include tree-vector-builder.h.
912 (valid_lattice_transition): Operate directly on the VECTOR_CST
914 * ipa-icf.c: Include tree-vector-builder.h.
915 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
916 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
918 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
920 * tree.c (build_vector): Delete.
921 * tree.h (build_vector): Make static and move into the self-testing
924 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
926 * vector-builder.h (vector_builder::derived): New const overload.
927 (vector_builder::elt): New function.
928 * tree-vector-builder.h (tree_vector_builder::type): New function.
929 (tree_vector_builder::apply_step): Declare.
930 * tree-vector-builder.c (tree_vector_builder::apply_step): New
932 * gimple-fold.h (tree_vector_builder): Declare.
933 (gimple_build_vector): Take a tree_vector_builder instead of a
934 type and vector of elements.
935 * gimple-fold.c (gimple_build_vector): Likewise.
936 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
938 (get_initial_defs_for_reduction): Likewise.
939 (vectorizable_induction): Likewise.
941 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
943 * tree-vector-builder.h
944 (tree_vector_builder::new_binary_operation): Declare.
945 * tree-vector-builder.c
946 (tree_vector_builder::new_binary_operation): New function.
947 * fold-const.c (fold_relational_const): Use it.
948 (const_binop): Likewise. Check that both input vectors have
949 the same number of elements, thus excluding things like WIDEN_SUM.
950 Check whether it is possible to operate directly on the encodings
953 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
955 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
956 new_unary_operation, operating only on the encoded elements.
957 (const_unop): Likewise.
958 (exact_inverse): Likewise.
959 (distributes_over_addition_p): New function.
960 (const_binop): Use tree_vector_builder and new_unary_operation
961 for combinations of VECTOR_CST and INTEGER_CST. Operate only
962 on the encoded elements unless the encoding is strided and the
963 operation does not distribute over addition.
964 (fold_convert_const): Use tree_vector_builder and
965 new_unary_operation. Operate only on the encoded elements
966 for truncating integer conversions, or for non-stepped encodings.
968 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
970 * config/sparc/sparc.c: Include tree-vector-builder.h.
971 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
972 * expmed.c: Include tree-vector-builder.h.
973 (make_tree): Use tree_vector_builder instead of build_vector.
974 * fold-const.c: Include tree-vector-builder.h.
975 (const_binop): Use tree_vector_builder instead of build_vector.
976 (const_unop): Likewise.
977 (native_interpret_vector): Likewise.
978 (fold_vec_perm): Likewise.
979 (fold_ternary_loc): Likewise.
980 * gimple-fold.c: Include tree-vector-builder.h.
981 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
983 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
984 (simplify_vector_constructor): Use tree_vector_builder instead
986 * tree-vect-generic.c: Include tree-vector-builder.h.
987 (add_rshift): Use tree_vector_builder instead of build_vector.
988 (expand_vector_divmod): Likewise.
989 (optimize_vector_constructor): Likewise.
990 * tree-vect-loop.c: Include tree-vector-builder.h.
991 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
992 of build_vector. Explicitly use a stepped encoding for
994 * tree-vect-slp.c: Include tree-vector-builder.h.
995 (vect_get_constant_vectors): Use tree_vector_builder instead
997 (vect_transform_slp_perm_load): Likewise.
998 (vect_schedule_slp_instance): Likewise.
999 * tree-vect-stmts.c: Include tree-vector-builder.h.
1000 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1001 (vect_gen_perm_mask_any): Likewise.
1002 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1003 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1005 (build_vector_from_val): Likewise. Explicitly use a duplicate
1008 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1010 * doc/generic.texi (VECTOR_CST): Describe new representation of
1012 * vector-builder.h: New file.
1013 * tree-vector-builder.h: Likewise.
1014 * tree-vector-builder.c: Likewise.
1015 * Makefile.in (OBJS): Add tree-vector-builder.o.
1016 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1017 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1018 (tree_vector): Change the number of elements to
1019 vector_cst_encoded_nelts.
1020 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1021 (VECTOR_CST_ELTS): Delete.
1022 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1023 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1024 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1025 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1026 (VECTOR_CST_ENCODED_ELT): Likewise.
1027 (vector_cst_encoded_nelts): New function.
1028 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1029 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1030 (vector_cst_int_elt, vector_cst_elt): Declare.
1031 * tree.c: Include tree-vector-builder.h.
1032 (tree_code_size): Abort if passed VECTOR_CST.
1033 (tree_size): Update for new VECTOR_CST layout.
1034 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1035 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1036 (build_vector): Use tree_vector_builder.
1037 (vector_cst_int_elt, vector_cst_elt): New functions.
1038 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1039 encoded elements and then create the vector in the canonical form.
1040 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1041 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1042 (tree_c_tests): Call test_vector_cst_patterns.
1043 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1045 (hash_tree): Likewise.
1046 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1047 (streamer_write_tree_header): Likewise.
1048 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1049 (streamer_alloc_tree): Likewise. Update call to make_vector.
1050 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1052 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1054 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1055 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1056 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1058 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1059 Richard Biener <rguenther@suse.de>
1061 PR tree-optimization/81303
1062 * Makefile.in (gimple-loop-interchange.o): New object file.
1063 * common.opt (floop-interchange): Reuse the option from graphite.
1064 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1065 -floop-interchange and mention it for -O3.
1066 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1067 * gimple-loop-interchange.cc: New file.
1068 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1069 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1070 * passes.def (pass_linterchange): New pass.
1071 * timevar.def (TV_LINTERCHANGE): New time var.
1072 * tree-pass.h (make_pass_linterchange): New declaration.
1073 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1074 interchange. Record IV before/after increment in new parameters.
1075 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1076 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1078 (check_reduction_path): ...New function here.
1079 * tree-vectorizer.h (check_reduction_path): New declaration.
1081 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1084 PR rtl-optimization/80818
1085 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1086 always early clobbered.
1087 * lra-lives.c (process_bb_lives): Check input hard regs for early
1088 clobbered non-operand hard reg.
1090 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1093 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1094 types_compatible_p, just that TYPE_MODE is the same.
1096 2017-12-07 Martin Sebor <msebor@redhat.com>
1099 * attribs.c (empty_attribute_table): Initialize new member of
1100 struct attribute_spec.
1101 (decl_attributes): Add argument. Handle mutually exclusive
1102 combinations of attributes.
1103 (selftests::test_attribute_exclusions): New function.
1104 (selftests::attribute_c_tests): Ditto.
1105 * attribs.h (decl_attributes): Add default argument.
1106 * selftest.h (attribute_c_tests): Declare.
1107 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1108 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1110 * doc/extend.texi (Common Function Attributes): Update const and pure.
1111 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1112 of struct attribute_spec.
1113 * config/arc/arc.c (arc_attribute_table): Same.
1114 * config/arm/arm.c (arm_attribute_table): Same.
1115 * config/avr/avr.c ( avr_attribute_table): Same.
1116 * config/bfin/bfin.c (bfin_attribute_table): Same.
1117 * config/cr16/cr16.c (cr16_attribute_table): Same.
1118 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1119 * config/h8300/h8300.c (h8300_attribute_table): Same.
1120 * config/i386/i386.c (ix86_attribute_table): Same.
1121 * config/ia64/ia64.c (ia64_attribute_table): Same.
1122 * config/m32c/m32c.c (m32c_attribute_table): Same.
1123 * config/m32r/m32r.c (m32r_attribute_table): Same.
1124 * config/m68k/m68k.c (m68k_attribute_table): Same.
1125 * config/mcore/mcore.c (mcore_attribute_table): Same.
1126 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1127 * config/mips/mips.c (mips_attribute_table): Same.
1128 * config/msp430/msp430.c (msp430_attribute_table): Same.
1129 * config/nds32/nds32.c (nds32_attribute_table): Same.
1130 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1131 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1132 * config/rl78/rl78.c (rl78__attribute_table): Same.
1133 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1134 * onfig/rx/rx.c (rx_attribute_table): Same.
1135 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1136 * config/sh/sh.c (sh_attribute_table): Same.
1137 * config/sparc/sparc.c (sparc_attribute_table): Same.
1138 * config/spu/spu.c (spu_attribute_table): Same.
1139 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1140 * config/v850/v850.c (v850_attribute_table): Same.
1141 * config/visium/visium.c (visium_attribute_table): Same.
1143 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1146 * config/arm/arm.c (INCLUDE_STRING): Define.
1147 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1148 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1151 2017-12-07 Michael Matz <matz@suse.de>
1153 Add unroll and jam pass
1155 * gimple-loop-jam.c: New file.
1156 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1157 * common.opt (funroll-and-jam): New option.
1158 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1159 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1160 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1161 * passes.def: Add pass_loop_jam.
1162 * timevar.def (TV_LOOP_JAM): Add.
1163 * tree-pass.h (make_pass_loop_jam): Declare.
1164 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1165 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1166 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1167 to flow_loop_tree_node_add.
1168 (duplicate_subloops, copy_loops_to): Append to sibling list.
1169 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1170 * doc/invoke.texi (-funroll-and-jam): Document new option.
1171 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1173 2017-12-07 Richard Biener <rguenther@suse.de>
1175 PR tree-optimization/83296
1176 PR tree-optimization/67769
1177 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1178 flow sensitive info in an unrelated BB.
1179 (value_replacement): Use reset_flow_sensitive_info.
1180 (minmax_replacement): Reset flow sensitive info on the def
1181 we move. Do not reset flow sensitive info in the whole BB
1182 we move the stmt to.
1183 (abs_replacement): Likewise.
1185 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1188 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1189 rs6000_cpu to the given -mcpu=, or to the default processor.
1191 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1193 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1194 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1195 instead of rs6000_cpu_attr.
1196 (rs6000_adjust_cost): Ditto.
1197 (is_microcoded_insn): Ditto.
1198 (rs6000_adjust_priority): Ditto.
1199 (rs6000_issue_rate): Ditto.
1200 (rs6000_use_sched_lookahead): Ditto.
1201 (rs6000_use_sched_lookahead_guard): Ditto.
1202 (rs6000_sched_reorder): Ditto.
1203 (force_new_group): Ditto.
1204 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1205 (group_ending_nop): Ditto.
1207 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1209 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1210 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1211 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1212 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1214 (rs6000_reassociation_width): Ditto.
1215 (rs6000_emit_epilogue): Ditto.
1216 (rs6000_adjust_cost): Ditto.
1217 (is_microcoded_insn): Ditto.
1218 (is_cracked_insn): Ditto.
1219 (rs6000_adjust_priority): Ditto.
1220 (rs6000_sched_reorder): Ditto.
1221 (rs6000_sched_reorder2): Ditto.
1222 (insn_must_be_first_in_group): Ditto.
1223 (insn_must_be_last_in_group): Ditto.
1224 (rs6000_register_move_cost): Ditto.
1225 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1228 2017-12-07 Julia Koval <julia.koval@intel.com>
1230 * config.gcc: Add vaesintrin.h.
1231 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1233 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1234 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1236 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1237 * config/i386/immintrin.h: Include vaesintrin.h.
1238 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1239 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1240 _mm_aesdec_epi128): New intrinsics.
1242 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1244 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1245 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1246 c-family/c-spellcheck.cc.
1247 (best_macro_match::best_macro_match): Likewise.
1249 (struct edit_distance_traits<cpp_hashnode *>): Move to
1250 c-family/c-spellcheck.h.
1251 (class best_macro_match): Likewise.
1253 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1255 PR tree-optimization/83293
1256 * gimple-ssa-strength-reduction.c (insert_initializers): Use
1257 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
1258 might insert into empty bb.
1261 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
1262 simplify for plus with :c added, and pointer_plus without that.
1263 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
1264 with undefined overflow and the conversion is not widening,
1265 perform negation in utype and only convert to type afterwards.
1266 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
1267 simplify for plus with :c added, and pointer_plus without that.
1268 If type is integral with undefined overflow and the conversion is
1269 not widening, perform minus in utype and only convert to type
1270 afterwards. Move the last pointer_diff_expr simplify into the
1273 2017-12-06 Martin Sebor <msebor@redhat.com>
1275 PR tree-optimization/82646
1276 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
1277 strncpy, not maxlen.
1279 2017-12-06 Martin Sebor <msebor@redhat.com>
1281 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
1284 PR tree-optimization/83075
1285 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
1286 strncat/strncpy don't change length of source string.
1288 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
1291 2017-11-29 Martin Aberg <maberg@gaisler.com>
1293 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1294 to prevent b2bst errata sequence.
1295 (sqrtdf2_fix): Likewise.
1297 2017-12-06 Jakub Jelinek <jakub@redhat.com>
1299 PR tree-optimization/81945
1300 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
1301 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
1302 to dest_cfun has orig_loop_num set, either remap it to the new
1303 loop number if the loop got moved too, or clear it.
1305 2017-12-05 Steve Ellcey <sellcey@cavium.com>
1307 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
1309 (thunderx2t99_nothing): New insn reservation.
1310 (thunderx2t99_mrs): New insn reservation.
1311 (thunderx2t99_multiple): New insn reservation.
1312 (thunderx2t99_alu_basi): Add bfx to reservation.
1313 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
1315 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1318 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
1320 2017-12-05 Bin Cheng <bin.cheng@arm.com>
1322 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
1323 tree-ssa-pre.c::remove_dead_inserted_code.
1324 * tree-ssa-dce.h: New file.
1325 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
1326 (remove_dead_inserted_code): Move and rename to function
1327 tree-ssa-dce.c::simple_dce_from_worklist.
1328 (pass_pre::execute): Update use.
1330 2017-12-05 Richard Biener <rguenther@suse.de>
1332 PR tree-optimization/83277
1333 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
1334 to code-gen liveout vars.
1336 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
1338 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
1339 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
1340 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
1343 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
1345 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
1347 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
1348 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
1351 2017-12-05 Richard Biener <rguenther@suse.de>
1353 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
1354 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
1355 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
1356 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
1358 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1360 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
1362 2017-12-05 Martin Liska <mliska@suse.cz>
1363 Jakub Jelinek <jakub@redhat.com>
1365 * doc/invoke.texi: Document the options.
1366 * flag-types.h (enum sanitize_code): Add
1367 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1368 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
1369 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
1370 * opts.c: Define new sanitizer options.
1371 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
1372 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
1374 2017-12-05 Julia Koval <julia.koval@intel.com>
1376 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
1377 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
1378 (ix86_handle_option): Handle -mavx512vnni.
1379 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
1380 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
1381 * config/i386/i386-c (__AVX512VNNI__): New.
1382 * config/i386/i386.c (ix86_target_string): Handle new option.
1383 (ix86_valid_target_attribute_inner_p): Handle new option.
1384 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
1385 * config/i386/i386.opt (mavx512vnni): New option.
1387 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1390 * athlon.md: Disable for generic.
1391 * haswell.md: Enable for generic.
1392 * i386.c (ix86_sched_init_global): Add core hooks for generic.
1393 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
1395 (ix86_adjust_cost): Move generic to haswell path.
1397 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
1399 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
1400 instead of MEM_P in a couple more places. Fix formatting issues.
1402 2017-12-04 Jim Wilson <jimw@sifive.com>
1404 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
1405 instead of GP_REG_LAST-1.
1406 (riscv_adjust_libcall_cfi_prologue): Likewise.
1407 (riscv_adjust_libcall_cri_epilogue): Likewise.
1408 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
1411 2017-12-04 Luis Machado <luis.machado@linaro.org>
1413 * ipa-pure-const.c (check_decl): Add missing newline.
1414 (state_from_flags): Likewise.
1416 2017-12-04 Jeff Law <law@redhat.com>
1418 PR tree-optimizatin/78496
1419 * gimple-ssa-evrp-analyze.h
1420 (evrp_range_analyzer::get_vr_values): Simplify.
1421 * gimple-ssa-evrp-analyze.c: Corresponding changes.
1422 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
1423 and gimple-ssa-evrp-analyze.h.
1424 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
1425 (simplify_stmt_for_jump_threading): Copy a blob of code from
1426 tree-vrp.c to use ranges to simplify statements.
1427 (dom_opt_dom_walker::before_dom_children): Call
1428 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
1429 (dom_opt_dom_walker::after_dom_children): Similarly for
1430 evrp_range_analyzer::leave.
1431 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
1434 * gimple-ssa-evrp-analyze.c
1435 (evrp_range_analyzer::extract_range_from_stmt): Always use
1436 vr_values::update_value_range so preexisting range info is
1437 medged with new range info, even if the new range is VR_VARYING.
1439 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1441 * combine.c: Adjust comment.
1442 (use_crosses_set_p): Delete.
1443 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
1444 (try_combine): Ditto.
1446 2017-12-04 Richard Biener <rguenther@suse.de>
1448 PR tree-optimization/83255
1449 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
1450 Re-add zero-iteration check.
1452 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1454 PR rtl-optimization/83245
1455 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
1456 hard registers as earlyclobber, also if not in an asm.
1458 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
1462 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1465 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1466 rs6000_cpu based on cpu_index, not tune_index.
1468 2017-12-04 Richard Biener <rguenther@suse.de>
1470 PR tree-optimization/83238
1471 * graphite-scop-detection.c (scop_detection::merge_sese): Make
1472 code match comment, rejecting invalid SESE regions.
1474 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
1476 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
1477 require base operand is a REG_POINTER prior to reload on targets
1478 with non-equivalent space registers.
1480 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1482 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
1483 (update_specialized_profile): Fix updating of counts.
1484 (perhaps_add_new_callers): Likewise.
1486 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
1489 * x86-tune.def: Remove obsolette FIXMEs.
1490 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
1491 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
1492 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
1494 (X86_TUNE_PAD_RETURNS): Disable for generic.
1496 2017-12-02 Jakub Jelinek <jakub@redhat.com>
1498 PR tree-optimization/83170
1499 PR tree-optimization/83241
1500 * gimple-ssa-store-merging.c
1501 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
1502 gimple_vuse (ins_stmt) in case it has changed.
1503 (imm_store_chain_info::output_merged_store): Likewise.
1505 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
1509 * tree-cfg.c (pass_warn_function_return::execute): Handle
1510 __builtin_ubsan_handle_missing_return like __builtin_unreachable
1511 with BUILTINS_LOCATION.
1515 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
1516 is BLKmode for vector field with vector raw mode, use TYPE_MODE
1517 instead of DECL_MODE.
1519 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
1520 last argument to rtx pointer.
1521 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
1522 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
1523 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
1524 depending on the chosen ISAs.
1525 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
1526 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
1527 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
1529 * config/i386/sse.md (mov<mode>_internal): Likewise.
1530 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
1532 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1534 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
1536 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1538 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
1539 parameters from prototype.
1540 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
1541 parameters. Don't print a comment.
1542 (emit_fusion_gpr_load): Adjust.
1543 (emit_fusion_load_store): Adjust.
1544 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
1545 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
1546 comment on the second line.
1548 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1551 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1552 rs6000_cpu based on cpu_index, not tune_index.
1554 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
1556 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
1557 which_alternative instead of which_alternative + 1.
1558 (output_asm_insn): Print an extra tab if the template is short.
1560 2017-12-01 Jim Wilson <jimw@sifive.com>
1562 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
1564 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
1565 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
1567 * doc/tm.texi: Regenerate.
1569 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1572 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
1573 whether we can allocate pseudos before trying to fix an address.
1574 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
1575 memory address is indexed or indirect.
1576 (floatuns_<mode>si2_hw2): Likewise.
1578 2017-12-01 Jason Merrill <jason@redhat.com>
1580 * Makefile.in (TAGS): Add c-family/*.cc.
1582 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
1584 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
1585 (call_value_insn): Likewise.
1586 (sibcall_insn): Likewise.
1587 (sibcall_value_insn): Likewise.
1588 (movsi_aarch64): Likewise.
1589 (movdi_aarch64): Likewise.
1590 (add_losym_): Likewise.
1591 (ldr_got_small_): Likewise.
1592 (ldr_got_small_sidi): Likewise.
1593 (ldr_got_small_28k_): Likewise.
1594 (ldr_got_small_28k_sidi): Likewise.
1595 * config/aarch64/aarch64.c (aarch64_print_address_internal):
1596 Move output_addr_const to symbolic case. Add error check.
1598 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1600 * config/s390/predicates.md (plus16_Q_operand): New predicate.
1601 * config/s390/s390.md: Disable MVC merging peephole if it would
1602 disable operand forwarding.
1603 (new peephole2): Split MVCs if it would turn them into up to 2
1606 2017-12-01 Richard Biener <rguenther@suse.de>
1608 PR tree-optimization/83232
1609 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
1610 detection of same access. Instead of breaking the group here
1611 do not consider the duplicate. Add comment explaining real fix.
1613 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
1615 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
1617 2017-12-01 Sudakshina Das <sudi.das@arm.com>
1619 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
1621 2017-12-01 Jakub Jelinek <jakub@redhat.com>
1623 * function.h (struct function): Remove cilk_frame_decl,
1624 is_cilk_function and calls_cilk_spawn fields.
1625 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
1627 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
1628 cilk_elemental field.
1629 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
1630 * target.def: Adjust comment.
1631 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
1632 Don't test cilk_elemental.
1634 PR tree-optimization/83233
1635 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
1636 bswap_stat name for the struct.
1639 * tree.h (SWITCH_BREAK_LABEL_P): Define.
1640 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
1641 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
1642 SWITCH_BREAK_LABEL_P set on the label.
1643 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
1644 added for default case if it was missing and not all cases covered.
1645 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
1646 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
1648 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
1649 to avoid -Wimplicit-fallthrough warning.
1650 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
1651 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
1653 PR tree-optimization/83221
1654 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
1656 (init_reassoc): Formatting fix.
1659 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
1660 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
1664 * tree-cfg.c: Include asan.h.
1665 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
1667 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1669 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
1671 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1673 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
1674 vpcmpeqd instruction.
1676 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
1678 * config/i386/i386.c (standard_sse_constant_opcode): Fix
1679 registers type for 128bit mode.
1681 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1683 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
1685 * gengtype-state.c (read_state_pair): Likewise.
1686 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
1687 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
1688 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
1689 * ifcvt.c (noce_try_store_flag_constants): Likewise.
1690 * tree-ssa-ccp.c (ccp_finalize): Likewise.
1691 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
1692 * builtins.c (fold_builtin_3): Likewise.
1693 * graphite-scop-detection.c
1694 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
1695 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
1698 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
1701 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
1702 parameter. Return true if ignore_decl_rtl_set_p is true and other
1703 conditions are satisfied.
1704 * asan.h (asan_protect_global): Add new parameter.
1705 * varasm.c (categorize_decl_for_section): Pass true as second parameter
1706 to asan_protect_global calls.
1708 2017-11-30 Jim Wilson <jimw@sifive.com>
1710 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
1711 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
1712 -mstrict-align, add info on default value. Delete redundant lines for
1713 -mabi. Add missing -mexplicit-relocs docs.
1715 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1717 * config/arc/arc.md (trap): New pattern.
1719 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1721 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
1722 instruction to end into a delay slot.
1723 * config/arc/arc.md (cond_delay_insn): Check if the instruction
1724 can be placed into a delay slot against reg_note.
1726 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1728 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
1729 labels number of usages.
1731 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
1733 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
1735 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
1737 2017-11-30 Jakub Jelinek <jakub@redhat.com>
1740 * internal-fn.c (expand_mul_overflow): Optimize unsigned
1741 multiplication by power of 2 constant into two shifts + comparison.
1743 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
1746 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
1748 2017-11-30 Richard Biener <rguenther@suse.de>
1750 PR tree-optimization/83202
1751 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1752 allow_peel argument and guard peeling.
1753 (canonicalize_loop_induction_variables): Likewise.
1754 (canonicalize_induction_variables): Pass false.
1755 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
1756 peeling from cunrolli.
1758 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1760 * combine.c (try_combine): Print a message to dump file whenever
1761 I0, I1, or I2 cannot be combined into I3.
1763 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1765 PR rtl-optimization/83156
1766 PR rtl-optimization/82621
1767 * combine.c (try_combine): Don't split an I2 if one of the dests is
1768 set again before I3. Allow unused dests.
1770 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
1772 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
1774 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
1776 PR rtl-optimization/80818
1777 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
1778 recursively. Use insn code for clobber.
1779 (lra_set_insn_recog_data): Pass the new arg to
1780 collect_non_operand_hard_regs.
1781 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
1783 (lra_update_insn_regno_info): Pass insn to
1784 add_regs_to_insn_regno_info.
1786 2017-11-29 Jim Wilson <jimw@sifive.com>
1787 Andrew Waterman <andrew@sifive.com>
1789 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
1790 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
1791 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
1792 use SINGLE_SHIFT_COST.
1793 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
1794 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
1796 2017-11-29 Julia Koval <julia.koval@intel.com>
1798 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
1799 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
1800 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
1801 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
1802 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
1803 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
1804 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
1805 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
1806 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
1807 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
1808 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
1809 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
1810 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
1811 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1812 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1813 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1814 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
1815 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
1816 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
1817 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
1818 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
1819 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
1820 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1821 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1822 __builtin_ia32_vpshldv_v2di_mask,
1823 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
1824 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
1825 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
1827 2017-11-29 Julia Koval <julia.koval@intel.com>
1829 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
1830 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
1831 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1832 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
1833 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
1834 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
1835 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
1836 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
1837 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
1838 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
1839 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
1840 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
1841 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
1842 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
1843 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
1844 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
1845 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
1846 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
1847 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1848 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
1849 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
1851 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1853 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
1854 movsi_pic_gotdata_op instruction as a load for the UT699 errata
1857 2017-11-29 Martin Aberg <maberg@gaisler.com>
1859 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
1860 to prevent b2bst errata sequence.
1861 (sqrtdf2_fix): Likewise.
1863 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1865 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
1866 (div_sqrt_insn_p): New function.
1867 (sparc_do_work_around_errata): Insert NOP instructions to
1868 prevent sequences that could trigger the TN-0013 errata for
1869 certain LEON3 processors.
1870 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
1871 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
1872 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
1873 (in_branch_delay): Prevent div and sqrt in delay slot if
1875 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
1877 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1879 * config/sparc/sparc.c (atomic_insn_p): New function.
1880 (sparc_do_work_around_errata): Insert NOP instructions to
1881 prevent sequences that could trigger the TN-0010 errata for
1883 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
1884 instruction referable in atomic_insns_p.
1886 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1888 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1889 (atomic_compare_and_swap_leon3_1): Likewise.
1892 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1894 * config/sparc/sparc.c (fpop_insn_p): New function.
1895 (sparc_do_work_around_errata): Insert NOP instructions to
1896 prevent sequences that could trigger the TN-0012 errata for
1898 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1899 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1900 (in_branch_annul_delay): Prevent floating-point instructions
1901 in delay slot of annulled integer branch.
1903 2017-11-29 Richard Biener <rguenther@suse.de>
1905 PR tree-optimization/83202
1906 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1908 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1909 nodes vectorized to the same stmts multiple times.
1910 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1911 (vect_analyze_slp_instance): Adjust.
1912 (scalar_stmts_to_slp_tree_map_t): New typedef.
1913 (vect_schedule_slp_instance): Add a map recording the SLP node
1914 representing the vectorized stmts for a set of scalar stmts.
1915 Avoid code-generating redundancies.
1916 (vect_schedule_slp): Allocate map and pass it down.
1918 2017-11-29 Nathan Sidwell <nathan@acm.org>
1921 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1924 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1926 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1927 cases for printing LDP/STP memory addresses.
1928 (aarch64_print_address_internal): Renamed from
1929 aarch64_print_operand_address, added parameter, add Pmode check.
1930 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1931 (aarch64_print_operand_address): Indirect to
1932 aarch64_print_address_internal.
1933 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1934 'y' operand output specifier.
1936 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1939 * tree.c (build_simple_mem_ref_loc): Handle
1940 get_addr_base_and_unit_offset returning a MEM_REF.
1943 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1946 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1949 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1951 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1953 (layout::move_to_column): Likewise.
1955 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1958 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1959 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1961 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1962 Martin Jambor <mjambor@suse.cz>
1965 * tree.h (expr_type_first_operand_type_p): Declare
1966 * tree.c (expr_type_first_operand_type_p): New function.
1967 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1968 (ipa_value_from_jfunc): Adjust declaration.
1969 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1970 Use it as result type for arithmetics, unless it is NULL in which case
1971 be more conservative.
1972 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1973 ipa_get_jf_pass_through_result.
1974 (propagate_vals_across_pass_through): Likewise.
1975 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1976 is to propagate_vals_across_pass_through.
1977 (propagate_constants_across_call): Pass PARM_TYPE to
1978 propagate_scalar_across_jump_function.
1979 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1980 ipa_value_from_jfunc.
1981 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1982 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1983 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1984 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1985 target_type, pass it to ipa_value_from_jfunc.
1986 (update_indirect_edges_after_inlining): Pass parameter type to
1987 try_make_edge_direct_simple_call.
1989 2017-11-28 Jeff Law <law@redhat.com>
1991 * gimple-ssa-evrp-analyze.c
1992 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1993 refine ranges if scev_initialized_p returns true.
1994 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1996 2017-11-28 Julia Koval <julia.koval@intel.com>
1998 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1999 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2000 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2001 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2002 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2003 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2004 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2005 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2006 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2007 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2008 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2009 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2010 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2011 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2012 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2013 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2014 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2015 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2016 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2017 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2018 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2019 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2021 2017-11-28 Julia Koval <julia.koval@intel.com>
2023 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2024 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2025 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2026 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2027 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2028 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2029 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2030 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2031 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2032 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2033 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2034 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2035 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2036 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2037 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2038 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2039 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2040 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2041 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2042 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2043 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2044 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2045 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2046 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2047 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2048 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2049 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2050 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2051 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2052 __builtin_ia32_vpshld_v2di_mask): New builtins.
2053 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2054 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2056 2017-11-28 Richard Biener <rguenther@suse.de>
2058 PR tree-optimization/80776
2059 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2061 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2063 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2064 If the incoming edge is an effective fallthru because the other
2065 edge only reaches a __builtin_unreachable () then record ranges
2066 derived from the controlling condition in SSA info.
2067 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2068 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2070 2017-11-28 Olivier Hainque <hainque@adacore.com>
2072 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2073 on mingw build hosts.
2075 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2078 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2080 2017-11-28 Richard Biener <rguenther@suse.de>
2083 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2084 copies generated from memcpy use a character array as reference
2087 2017-11-28 Julia Koval <julia.koval@intel.com>
2088 Sebastian Peryt <sebastian.peryt@intel.com>
2090 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2091 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2092 cilk-common.o, cilk.h, cilk-common.c): Remove.
2094 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2095 * builtins.c (is_builtin_name): Remove cilkplus condition.
2096 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2097 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2098 cilk-builtins.def, cilkplus.def): Remove.
2099 * cif-code.def (CILK_SPAWN): Remove.
2100 * cilk-builtins.def: Delete.
2101 * cilk-common.c: Ditto.
2103 * cilkplus.def: Ditto.
2104 * config/darwin.h (fcilkplus): Delete.
2105 * cppbuiltin.c: Ditto.
2106 * doc/extend.texi: Remove cilkplus doc.
2107 * doc/generic.texi: Ditto.
2108 * doc/invoke.texi: Ditto.
2109 * doc/passes.texi: Ditto.
2110 * gcc.c (fcilkplus): Remove.
2111 * gengtype.c (cilk.h): Remove.
2112 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2114 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2116 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2118 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2119 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2120 cilkplus conditions.
2121 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2122 inline_read_section): Ditto.
2123 * ipa-inline-analysis.c (cilk.h): Remove.
2124 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2125 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2126 append_linker_options): Remove condition for fcilkplus.
2127 * lto/lto-lang.c (cilk.h): Remove.
2128 (lto_init): Remove condition for fcilkplus.
2129 * omp-expand.c (expand_cilk_for_call): Delete.
2130 (expand_omp_taskreg, expand_omp_for_static_chunk,
2131 expand_omp_for): Remove cilkplus
2133 (expand_cilk_for): Delete.
2134 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2135 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2136 execute_lower_omp, diagnose_sb_0): Ditto.
2137 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2138 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2139 * tree-nested.c: Ditto.
2140 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2141 (dump_generic_node): Ditto.
2142 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2143 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2145 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2147 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2149 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2150 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2151 for vector mode and !TARGET_SIMD.
2153 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2155 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2157 * tree.h (SWITCH_LABELS): Remove.
2158 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2159 assert SWITCH_BODY is non-NULL.
2160 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2162 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2164 PR tree-optimization/80788
2165 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2166 has TREE_OVERFLOW set, call drop_tree_overflow.
2168 2017-11-28 Richard Biener <rguenther@suse.de>
2170 PR tree-optimization/83158
2171 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2173 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2176 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2177 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2179 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2181 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2182 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2184 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2187 * varasm.c (compare_constant): Take the mode of the constants into
2188 account when comparing floating point constants.
2190 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2192 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2193 from explicit instantiation of debug_helper.
2194 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2196 2017-11-27 Richard Biener <rguenther@suse.de>
2198 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2201 2017-11-27 Richard Biener <rguenther@suse.de>
2203 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2204 (type_hash_canon): Also clear int_cst_hash_table entry for
2206 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2208 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2210 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2211 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2212 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2214 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2216 * hash-map.h (gt_cleare_cache): Avoid UB.
2218 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2220 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2221 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2222 (decide_unroll_runtime_iterations): Likewise.
2223 (decide_unroll_stupid): Likewise.
2225 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2228 * config/i386/i386.md: Add a loop with incssp.
2230 2017-11-27 Martin Jambor <mjambor@suse.cz>
2232 PR tree-optimization/81248
2233 * tree-sra.c (splice_param_accesses): Remove size check.
2234 (decide_one_param_reduction): Fix size check.
2235 * gimple-pretty-print.c (dump_profile): Silence warning.
2236 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2238 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2241 * dbxout.c (lastlineno): New variable.
2242 (dbx_debug_hooks): Use dbxout_switch_text_section as
2243 switch_text_section debug hook.
2244 (dbxout_function_end): Switch to current_function_section
2245 rather than function_section. If crtl->has_bb_partition,
2246 output just one N_FUN, depending on in_cold_section_p.
2247 (dbxout_source_line): Remember last lineno in lastlineno.
2248 (dbxout_switch_text_section): New function.
2249 (dbxout_function_decl): Adjust dbxout_block caller.
2250 (dbx_block_with_cold_children): New function.
2251 (dbxout_block): Return true if any LBRAC/RBRAC have been
2252 emitted. Use dbx_block_with_cold_children at depth == 0
2253 in second partition. Add PARENT_BLOCKNUM argument, pass
2254 it optionally adjusted to children. Output LBRAC/RBRAC
2255 around recursive call only if the block is in the current
2256 partition, if not and anything was output, emit empty
2258 * final.c (final_scan_insn): Compute cold_function_name
2259 before calling switch_text_section debug hook. Call
2260 that hook even if dwarf2out_do_frame if not emitting
2264 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
2265 TREE_READONLY decls.
2267 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2269 PR rtl-optimization/82488
2270 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
2272 2017-11-26 Julia Koval <julia.koval@intel.com>
2274 * config/i386/i386.c (processor_target_table): Add skylake_cost for
2276 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
2279 2017-11-26 Julia Koval <julia.koval@intel.com>
2281 * config/i386/driver-i386.c (host_detect_local_cpu):
2282 Detect skylake-avx512.
2284 2017-11-26 Julia Koval <julia.koval@intel.com>
2286 * config.gcc: Add -march=cannonlake.
2287 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
2288 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
2289 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
2290 (PTA_CANNONLAKE): New.
2291 (processor_target_table): Add cannonlake.
2292 (ix86_option_override_internal): Ditto.
2293 (fold_builtin_cpu): Ditto.
2294 (get_builtin_code_for_version): Handle cannonlake.
2295 (M_INTEL_COREI7_CANNONLAKE): New.
2296 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
2297 * doc/invoke.texi: Add -march=cannonlake.
2299 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
2301 * plugin.c (add_new_plugin): Use platform-specific library extensions.
2302 (try_init_one_plugin): Alternative implementation for MinGW.
2303 * Makefile.in (plugin_implib): New.
2304 (gengtype-lex.c): Fix broken AIX workaround.
2305 * configure: Regenerate.
2306 * doc/plugins.texi: Document support for MinGW.
2308 2017-11-25 Jakub Jelinek <jakub@redhat.com>
2310 PR rtl-optimization/81553
2311 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
2312 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
2313 is a shift where C1 has different mode than the whole shift, use C1's
2314 mode for MULT rather than the shift's mode.
2317 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
2318 builtins not enabled in the currently selected ISA.
2320 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
2322 PR tree-optimization/71026
2323 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
2324 (insert_reciprocals): Change to insert reciprocals before a division
2325 by a square and to insert the square of a reciprocal.
2326 (execute_cse_reciprocals_1): Change to consider division by a square.
2327 (register_division_in): Add importance parameter.
2329 2017-11-24 Richard Biener <rguenther@suse.de>
2331 PR tree-optimization/82402
2332 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
2333 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
2335 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
2337 * match.pd (0-ptr): New transformation.
2339 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
2342 * ipa-inline.c (inline_small_functions): Set current badnes correctly
2343 when skipping checking.
2345 2017-11-24 Richard Biener <rguenther@suse.de>
2347 PR tree-optimization/83128
2348 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
2349 (vn_reference_lookup_3): Likewise.
2351 2017-11-24 Jakub Jelinek <jakub@redhat.com>
2354 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
2355 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
2358 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
2359 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
2360 check for the latter with gimple_call_builtin_p. Do not handle
2361 BUILT_IN_STPNCPY_CHK which is not a pass through call.
2363 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
2365 * config/arm/arm_neon.h: Fix pragma GCC push_options before
2368 2017-11-23 Julia Koval <julia.koval@intel.com>
2370 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
2371 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
2372 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
2373 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
2374 _mm512_maskz_expandloadu_epi16): New intrinsics.
2375 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
2376 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
2377 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
2378 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
2379 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
2380 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
2381 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
2382 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
2383 _mm256_maskz_expandloadu_epi8): New intrinsics.
2384 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
2385 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
2386 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
2387 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
2388 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
2389 * config/i386/sse.md (VI248_VLBW): New iterator.
2390 (expand<mode>_mask, expand<mode>_maskz): New patterns.
2392 2017-11-23 Julia Koval <julia.koval@intel.com>
2394 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
2395 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
2396 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
2397 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
2398 _mm512_mask_compressstoreu_epi16): New.
2399 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
2400 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
2401 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
2402 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
2403 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
2404 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
2405 _mm256_mask_compressstoreu_epi8): New.
2406 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
2407 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
2408 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
2409 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
2410 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
2411 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
2412 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
2413 __builtin_ia32_compresshi128_mask,
2414 __builtin_ia32_compressstoreuqi512_mask,
2415 __builtin_ia32_compressstoreuhi512_mask,
2416 __builtin_ia32_compressstoreuqi256_mask,
2417 __builtin_ia32_compressstoreuqi128_mask,
2418 __builtin_ia32_compressstoreuhi256_mask,
2419 __builtin_ia32_compressstoreuhi128_mask): New builtins.
2420 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
2422 (ix86_expand_special_args_builtin): Handle new types.
2423 (s4fma_expand): Handle new builtin array.
2424 * config/i386/immintrin.h: Include new headers.
2425 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
2426 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
2428 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2430 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
2431 fixes. Declare temp and g variables at the top in order to avoid
2432 {} in most of the cases.
2434 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
2436 * match.pd (ptr-0): New transformation.
2438 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2440 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
2441 (struct addr_mode_cost_table): New.
2442 (struct tune_params): Add field addr_mode_costs.
2443 * config/arm/arm.c (generic_addr_mode_costs): New.
2444 (arm_slowmul_tune): Initialise addr_mode_costs field.
2445 (arm_fastmul_tune): Likewise.
2446 (arm_strongarm_tune): Likewise.
2447 (arm_xscale_tune): Likewise.
2448 (arm_9e_tune): Likewise.
2449 (arm_marvell_pj4_tune): Likewise.
2450 (arm_v6t2_tune): Likewise.
2451 (arm_cortex_tune): Likewise.
2452 (arm_cortex_a8_tune): Likewise.
2453 (arm_cortex_a7_tune): Likewise.
2454 (arm_cortex_a15_tune): Likewise.
2455 (arm_cortex_a35_tune): Likewise.
2456 (arm_cortex_a53_tune): Likewise.
2457 (arm_cortex_a57_tune): Likewise.
2458 (arm_exynosm1_tune): Likewise.
2459 (arm_xgene1_tune): Likewise.
2460 (arm_cortex_a5_tune): Likewise.
2461 (arm_cortex_a9_tune): Likewise.
2462 (arm_cortex_a12_tune): Likewise.
2463 (arm_cortex_a73_tune): Likewise.
2464 (arm_v7m_tune): Likewise.
2465 (arm_cortex_m7_tune): Likewise.
2466 (arm_v6m_tune): Likewise.
2467 (arm_fa726te_tune): Likewise.
2468 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
2470 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
2472 * config/arm/arm.c (arm_mem_costs): New function.
2473 (arm_rtx_costs_internal): Use arm_mem_costs.
2475 2017-11-23 Mark Wielaard <mark@klomp.org>
2477 * dwarf2out.c (init_sections_and_labels): Use generation to create
2478 unique ranges_section_label and ranges_base_label. Return generation.
2479 (output_rnglists): Add generation argument. Use generation to create
2480 unique ranges labels.
2481 (dwarf2out_finish): Get generation from init_sections_and_labels
2482 and pass generation to output_rnglists.
2484 2017-11-23 Mike Stump <mikestump@comcast.net>
2485 Eric Botcazou <ebotcazou@adacore.com>
2487 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
2488 * cfgloop.h (struct loop): Add unroll field.
2489 * function.h (struct function): Add has_unroll bitfield.
2490 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
2491 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
2492 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
2493 (pass_rtl_unroll_loops::gate): Likewise.
2494 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
2495 for which loop->unroll==1.
2496 (decide_unroll_constant_iterations): Use note for consistency and
2497 take loop->unroll into account. Return early if loop->unroll is set.
2498 Fix thinko in existing test.
2499 (decide_unroll_runtime_iterations): Use note for consistency and
2500 take loop->unroll into account.
2501 (decide_unroll_stupid): Likewise.
2502 * lto-streamer-in.c (input_cfg): Read loop->unroll.
2503 * lto-streamer-out.c (output_cfg): Write loop->unroll.
2504 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
2506 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
2507 (print_loop): Print loop->unroll if set.
2508 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
2509 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
2510 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
2512 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
2513 loop->unroll is set and smaller than the trip count. Otherwise bypass
2514 entirely the heuristics if loop->unroll is set. Remove dead note.
2515 Fix off-by-one bug in other note.
2516 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
2517 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
2519 (tree_unroll_loops_completely): Make static.
2520 (pass_complete_unroll::execute): Use correct type for variable.
2521 (pass_complete_unrolli::execute): Fix formatting.
2522 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
2524 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
2526 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
2527 enable when TARGET_PREFER_AVX128 is set.
2529 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2531 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
2533 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
2534 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
2536 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2538 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
2539 frequencies. Fix estimation of aggregate parameters.
2541 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
2543 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
2546 2017-11-23 Tom de Vries <tom@codesourcery.com>
2548 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
2550 2017-11-23 Tom de Vries <tom@codesourcery.com>
2552 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
2555 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
2558 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
2559 sibcall_value_pcrel_fdpic): Use local variable instead of
2561 (calli_tbr_rel): Add missing operand 2.
2562 (call_valuei_tbr_rel): Add missing operand 3.
2564 2017-11-23 Jakub Jelinek <jakub@redhat.com>
2567 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
2568 bitpos/bitsize covering the whole destination, use store_expr only if
2569 the complex mode is the same. Otherwise, use expand_normal and if
2570 it returns CONCAT, subreg each part separately instead of trying to
2571 subreg the whole result.
2573 2017-11-23 Richard Biener <rguenther@suse.de>
2575 PR tree-optimization/23094
2576 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
2577 come here from walking over backedges in the first iteration.
2578 (vn_reference_lookup_3): Skip clobbers that store the same value.
2580 2017-11-23 Richard Biener <rguenther@suse.de>
2582 PR tree-optimization/81403
2583 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
2584 a block we need a leader relative to.
2585 (phi_translate_1): For nary processing require a leader from
2586 get_representative_for given we run expression simplification
2587 using match-and-simplify. Remove previous fix.
2589 2017-11-22 Jeff Law <law@redhat.com>
2591 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2592 Use new method allocate_value_range rather than accessing the
2593 vrp_value_range_pool data member directly.
2594 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
2595 to use extract_range_from_stmt method to avoid need for
2596 extract_range_from_assignment method.
2597 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
2598 method rather than setting values_propgated data member directly.
2599 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
2600 and values propagated data members and extract_range_from_assignment
2601 method. Reorder private data members to conform to standards.
2602 Add new methods set_lattice_propagation_complete and
2603 allocate_value_range.
2605 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
2607 PR rtl-optimization/83030
2608 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
2609 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
2610 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
2611 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
2612 CROSSING_JUMP_P flag is already set before setting it.
2613 * cfgrtl.c (fixup_partition_crossing): Likewise.
2614 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
2617 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2619 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
2620 Handle the case where both arguments are using gen_const_vec_series.
2622 2017-11-22 David Malcolm <dmalcolm@redhat.com>
2625 * pretty-print.c (pp_format): Move quoting implementation to
2626 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
2627 to pass address of "quote" local.
2628 (pp_begin_quote): New function.
2629 (pp_end_quote): New function.
2630 * pretty-print.h (printer_fn): Convert penultimate param from bool
2632 (pp_begin_quote): New decl.
2633 (pp_end_quote): New decl.
2634 * tree-diagnostic.c (default_tree_printer): Convert penultimate
2635 param from bool to bool *.
2636 * tree-diagnostic.h (default_tree_printer): Likewise.
2638 2017-11-22 Jeff Law <law@redhat.com>
2640 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
2641 Initialize vr_values.
2642 (evrp_range_analyzer::try_find_new_range): Call methods attached to
2643 vr_values via vr_values class instance rather than delegators.
2644 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
2645 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2646 (evrp_range_analyzer::push_value_range): Likewise.
2647 (evrp_range_analyzer::pop_value_range): Likewise.
2648 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
2649 most delegators. Those remaining are exposed as public interfaces.
2650 Make vr_values a pointer and private.
2651 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
2653 (evrp_range_analyzer::get_vr_value): New method.
2654 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
2655 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
2656 (class evrp_dom_walker): Attach evrp_folder class, initialize
2657 it in the ctor. Remove temporary delegators.
2658 (evrp_dom_walker::before_dom_children): Call methods in attached
2659 evrp_range_analyzer class via class instance pointer. Use
2660 free value_range_constant_singleton to remove need for
2661 op_with_constant_singleton_value delegator method. Do not
2662 create a vrp_prop class instance for every call! Narrow
2663 scope of a couple variables.
2664 (evrp_dom_walker::cleanup): Call methods in attached
2665 evrp_range_analyzer class via class instance pointer.
2666 * vr-values.h (class vr_values): Privatize many methods and
2669 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2671 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
2672 * cfgexpand.c (expand_debug_expr): Remove handling for them.
2673 * expr.c (expand_expr_real_2): Likewise.
2674 * fold-const.c (const_unop): Likewise.
2675 * optabs-tree.c (optab_for_tree_code): Likewise.
2676 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2677 * tree-inline.c (estimate_operator_cost): Likewise.
2678 * tree-pretty-print.c (dump_generic_node): Likewise.
2679 (op_code_prio): Likewise.
2680 (op_symbol_code): Likewise.
2681 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
2682 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
2683 * internal-fn.c (direct_internal_fn_optab): New function.
2684 (direct_internal_fn_array, direct_internal_fn_supported_p
2685 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
2686 * fold-const-call.c (fold_const_reduction): New function.
2687 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
2689 * tree-vect-loop.c: Include internal-fn.h.
2690 (reduction_code_for_scalar_code): Rename to...
2691 (reduction_fn_for_scalar_code): ...this and return an internal
2693 (vect_model_reduction_cost): Take an internal_fn rather than
2695 (vect_create_epilog_for_reduction): Likewise. Build calls rather
2697 (vectorizable_reduction): Use internal functions rather than tree
2698 codes for the reduction operation. Update calls to the functions
2700 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
2701 Use calls to internal functions rather than REDUC tree codes.
2702 * config/aarch64/aarch64-simd.md: Update comment accordingly.
2704 2017-11-22 Olivier Hainque <hainque@adacore.com>
2706 * config/vxworks.c (vxworks_override_options): Pick default
2707 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
2708 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
2709 DWARF_GNAT_ENCODINGS_DEFAULT.
2710 * config/vxworksae.h: Likewise.
2712 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
2714 PR tree-optimization/83104
2715 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
2718 2017-11-22 Marek Polacek <polacek@redhat.com>
2719 H.J. Lu <hongjiu.lu@intel.com>
2720 Jason Merrill <jason@redhat.com>
2725 * calls.c (initialize_argument_information): Call
2726 warn_parameter_passing_abi target hook.
2727 (store_one_arg): Use 0 for empty record size. Don't push 0 size
2728 argument onto stack.
2729 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
2730 * common.opt: Update -fabi-version description.
2731 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
2732 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
2734 (ix86_is_empty_record): New function.
2735 (ix86_warn_parameter_passing_abi): New function.
2736 (TARGET_EMPTY_RECORD_P): Redefine.
2737 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
2738 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
2739 * doc/tm.texi: Regenerated.
2740 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
2741 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
2742 * dwarf2out.c (get_ultimate_context): Move to tree.c.
2743 * explow.c (hard_function_value): Call arg_int_size_in_bytes
2744 instead of int_size_in_bytes.
2745 * expr.c (copy_blkmode_to_reg): Likewise.
2746 * function.c (aggregate_value_p): Return 0 for empty types.
2747 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
2749 (locate_and_pad_parm): Call arg size_in_bytes instead
2751 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
2752 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
2753 * target.def (empty_record_p, warn_parameter_passing_abi): New target
2755 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
2756 (std_gimplify_va_arg_expr): Skip empty records. Call
2757 arg_size_in_bytes instead size_in_bytes.
2758 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
2759 * tree-core.h (tree_type_common): Add empty_flag.
2760 (tree_decl_common): Update comments.
2761 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
2763 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2764 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
2766 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
2767 * tree.c (default_is_empty_type): New function.
2768 (default_is_empty_record): New function.
2769 (arg_int_size_in_bytes): New function.
2770 (arg_size_in_bytes): New function.
2771 (get_ultimate_context): New function.
2772 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
2773 TRANSLATION_UNIT_WARN_EMPTY_P.
2774 (default_is_empty_record, arg_int_size_in_bytes,
2775 arg_size_in_bytes, get_ultimate_context): Declare.
2777 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2779 * config/arm/arm.c (cmse_clear_registers): New function.
2780 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
2781 code by call to cmse_clear_registers.
2782 (cmse_nonsecure_entry_clear_before_return): Likewise.
2784 2017-11-22 Tamar Christina <tamar.christina@arm.com>
2786 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
2787 (vdot_s32, vdotq_s32): New.
2788 (vdot_lane_u32, vdotq_lane_u32): New.
2789 (vdot_lane_s32, vdotq_lane_s32): New.
2792 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2795 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
2796 for unsigned values with fewer HWIs than the precision.
2797 (test_overflow): New function.
2798 (wide_int_cc_tests): Call it.
2800 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
2802 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
2803 has the same mode class as Pmode.
2805 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2807 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
2808 padding_bits_to_clear_ptr.
2809 (cmse_nonsecure_entry_clear_before_return): Likewise.
2811 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2813 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
2814 auto_sbitap instead of integer bitfield to control register needing
2817 2017-11-22 Jakub Jelinek <jakub@redhat.com>
2819 PR tree-optimization/83044
2820 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
2821 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
2822 that need the upper bound. Subtract offset from
2823 get_addr_base_and_unit_offset only if positive and subtract it
2824 before division by eltsize rather than after it.
2827 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
2828 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
2829 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
2833 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
2834 expand_binop with *mul_widen_optab, make sure at least one of the
2835 operands doesn't have VOIDmode.
2838 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
2840 PR rtl-optimization/82044
2841 PR tree-optimization/82042
2842 * dse.c (record_store): Check for overflow.
2843 (check_mem_read_rtx): Properly check for overflow if width == -1, call
2844 add_wild_read instead of clear_rhs_from_active_local_stores on
2845 overflow and log it into dump_file.
2847 2017-11-22 Richard Biener <rguenther@suse.de>
2849 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
2850 fake edges to exit when looking for a place to insert.
2851 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
2853 (insert_into_preds_of_block): Commit edge insertion immediately,
2854 assert that doesn't require new BBs.
2855 (fini_pre): Release expressions.
2856 (pass_pre::execute): Shuffle things around a bit, if the fn
2857 is too large do not compute AVAIL either as this is really the
2860 2017-11-22 Richard Biener <rguenther@suse.de>
2862 PR tree-optimization/83089
2863 * tree-if-conv.c (pass_if_conversion::execute): If anything
2864 changed reset SCEV and free the number of iteration estimates.
2866 2017-11-21 Martin Sebor <msebor@redhat.com>
2868 PR tree-optimization/82945
2869 * calls.h (warn_nonstring_bound): Remove unused function.
2871 2017-11-21 Martin Sebor <msebor@redhat.com>
2873 PR tree-optimization/82945
2874 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
2875 * calls.h (maybe_warn_nonstring_arg): Declare new function.
2876 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
2878 (initialize_argument_information): Call maybe_warn_nonstring_arg.
2879 * calls.h (get_attr_nonstring_decl): Declare new function.
2880 * doc/extend.texi (attribute nonstring): Update.
2881 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
2882 get_attr_nonstring_decl and handle it.
2883 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
2884 detection of nul-termination.
2885 (strlen_to_stridx): Change to a pointer.
2886 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2887 (pass_strlen::execute): Same.
2889 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2891 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2892 for the new option -mprefer-vector-width=[none|128|256|512].
2893 * config/i386/i386.c (ix86_target_string): remove old style options
2894 -mprefer-avx256 and make -mprefer-avx128 as alias.
2895 (ix86_option_override_internal): Apply defaults for the
2896 -mprefer-vector-width=[128|256] option.
2897 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2898 Implement macros to work with -mprefer-vector-width=.
2899 * config/i386/i386.opt: Implemented option
2900 -mprefer-vector-width=[none|128|256|512].
2901 * doc/invoke.texi: Documentation for
2902 -mprefer-vector-width=[none|128|256|512].
2904 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2906 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2908 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2910 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2912 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2914 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2916 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2917 * expr.c (expand_expr_real_2): Likewise.
2918 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2919 fold_binary_loc): Likewise.
2920 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2921 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2922 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2923 MINUS_EXPR transformations.
2924 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2925 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2926 * tree-inline.c (estimate_operator_cost): Likewise.
2927 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2928 op_symbol_code): Likewise.
2929 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2930 * vr-values.c (extract_range_from_binary_expr): Likewise.
2931 * varasm.c (initializer_constant_valid_p_1): Likewise.
2932 * tree.def: New tree code POINTER_DIFF_EXPR.
2934 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2936 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2937 integer suffix to movbe mnemonic.
2938 (*bswaphi2_movbe): Ditto.
2939 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2941 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2944 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2945 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2946 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2948 2017-11-21 Martin Liska <mliska@suse.cz>
2950 * tree-inline.c (expand_call_inline): Remove not needed
2953 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2954 Jakub Jelinek <jakub@redhat.com>
2957 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2958 Remove static keyword from f variable.
2960 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2962 PR tree-optimization/83086
2963 * gimple-ssa-store-merging.c
2964 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2965 rather than n.base_addr.
2967 2017-11-21 Martin Liska <mliska@suse.cz>
2969 PR rtl-optimization/82044
2970 PR tree-optimization/82042
2971 * dse.c (check_mem_read_rtx): Check for overflow.
2973 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2975 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2978 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2980 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2981 and its corresponding call arg location note.
2983 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2984 Andrew Burgess <andrew.burgess@embecosm.com>
2986 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2988 (arc_return_slot_offset): Likewise.
2989 (arc_eh_return_address_location): New declaration.
2990 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2991 (MUST_SAVE_REGISTER): Add exception handler case.
2992 (MUST_SAVE_RETURN_ADDR): Likewise.
2993 (arc_frame_pointer_required): Likewise.
2994 (arc_frame_pointer_needed): New function.
2995 (arc_compute_frame_size): Changed.
2996 (arc_expand_prologue): Likewise.
2997 (arc_expand_epilogue): Likewise.
2998 (arc_initial_elimination_offset): Likewise.
2999 (arc_return_slot_offset): Delete.
3000 (arc_eh_return_address_location): New function.
3001 (arc_builtin_setjmp_frame_value): Likewise.
3002 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3003 (EH_RETURN_STACKADJ_RTX): Define.
3004 (EH_RETURN_HANDLER_RTX): Likewise.
3005 * config/arc/arc.md (eh_return): Delete.
3007 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3009 * print-tree.h (debug_vec_tree): Remove prototype.
3010 * gdbinit.in (pvt): Remove macro.
3012 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3014 PR tree-optimization/83047
3015 * gimple-ssa-store-merging.c
3016 (imm_store_chain_info::output_merged_store): If the loads with the
3017 same vuse are in different basic blocks, for load_gsi pick a load
3018 location that is dominated by the other loads.
3021 * config/i386/i386.c (ix86_memmodel_check): Start
3022 -Winvalid-memory-model diagnostics with lowercase letter.
3025 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3026 set in_cold_section_p to the partition containing loc_list->first.
3027 When seeing loc_list->last_before_switch node, update secname and
3028 perform range_across_switch second partition handling only after that.
3031 * run-rtl-passes.c: Include debug.h.
3032 (run_rtl_passes): Call debug_hooks->assembly_start.
3033 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3037 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3038 OPTAB_DIRECT in calls to expand_simple_binop.
3040 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3043 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3045 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3048 * spellcheck.h (best_match::blithely_get_best_candidate): New
3051 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3053 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3054 with lower case letter.
3056 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3058 * config/i386/i386.md (bswaphi2): New expander.
3059 (*bswaphi2_movbe): New insn pattern.
3060 (bswaphi -> rorhi pepehole2): New peephole pattern.
3062 2017-11-20 Jeff Law <law@redhat.com>
3064 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3065 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3066 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3067 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3068 gimple-ssa-evrp-analyze.h.
3070 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3071 set BB_VISITED here.
3072 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3074 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3076 * tree-predcom.c: Add general comment on Store-Store chains.
3077 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3079 (get_chain_last_ref_at): Rename into...
3080 (get_chain_last_write_at): ...this.
3081 (get_chain_last_write_before_load): New function.
3082 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3083 CT_STORE_STORE when write reference is added.
3084 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3085 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3086 (initialize_root_vars_store_elim_1): Ditto.
3087 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3088 definition is created.
3089 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3090 chain by replacing it with dominant stored value.
3092 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3094 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3096 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3098 * vr-values.c (extract_range_from_binary_expr): Use a full range
3101 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3103 * config/arm/arm.md (R4_REGNUM): Define constant.
3104 (nonsecure_call_internal): Remove r4 clobber.
3105 (nonsecure_call_value_internal): Likewise.
3106 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3107 clobber and resequence match_operands.
3108 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3109 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3110 (nonsecure_call_value_reg_thumb2): Likewise.
3112 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3114 PR tree-optimization/78821
3115 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3116 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3117 to the address of the base rather than the base itself.
3118 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3119 (find_bswap_or_nop_finalize): New function.
3120 (find_bswap_or_nop): Use it.
3121 (bswap_replace): Return a tree rather than bool, change first
3122 argument from gimple * to gimple_stmt_iterator, allow inserting
3123 into an empty sequence, allow ins_stmt to be NULL - then emit
3124 all stmts into gsi. Fix up MEM_REF address gimplification.
3125 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3126 (struct store_immediate_info): Add N and INS_STMT non-static
3128 (store_immediate_info::store_immediate_info): Initialize them
3129 from newly added ctor args.
3130 (merged_store_group::apply_stores): Formatting fixes. Sort by
3132 (stmts_may_clobber_ref_p): For stores call also
3133 refs_anti_dependent_p.
3134 (gather_bswap_load_refs): New function.
3135 (imm_store_chain_info::try_coalesce_bswap): New method.
3136 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3137 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3138 (imm_store_chain_info::output_merged_store): Fail if number of
3139 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3140 and NOP_EXPR rhs_code.
3141 (pass_store_merging::process_store): Compute n and ins_stmt, if
3142 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3143 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3145 (pass_store_merging::execute): Calculate dominators.
3147 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3148 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3149 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3150 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3151 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3152 class pass_optimize_bswap, bswap_replace,
3153 pass_optimize_bswap::execute): Moved to ...
3154 * gimple-ssa-store-merging.c: ... this file.
3155 Include optabs-tree.h.
3156 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3157 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3158 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3159 anonymous namespace, remove static keywords.
3160 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3161 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3163 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3166 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3168 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3170 * vec.h (debug_helper): New function.
3171 (DEFINE_DEBUG_VEC): New macro.
3172 * hash-set.h (debug_helper): New function.
3173 (DEFINE_DEBUG_HASH_SET): New macro.
3174 * cfg.c (debug_slim (edge)): New function.
3175 Call DEFINE_DEBUG_VEC for edges.
3176 Call DEFINE_DEBUG_HASH_SET for edges.
3177 * cfghooks.c (debug_slim (basic_block)): New function.
3178 Call DEFINE_DEBUG_VEC for basic blocks.
3179 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3180 * print-tree.c (debug_slim): New function to handle trees.
3181 Call DEFINE_DEBUG_VEC for trees.
3182 Call DEFINE_DEBUG_HASH_SET for trees.
3183 (debug (vec<tree, va_gc>) &): Remove.
3184 (debug (<vec<tree, va_gc>) *): Remove.
3185 * print-rtl.c (debug_slim): New function to handle const_rtx.
3186 Call DEFINE_DEBUG_VEC for rtx_def.
3187 Call DEFINE_DEBUG_VEC for rtx_insn.
3188 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3189 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3190 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3191 (debug (vec<rtx_insn *> *ptr): Remove.
3192 (debug_insn_vector): Remove.
3193 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3195 2017-11-20 Tom de Vries <tom@codesourcery.com>
3197 PR rtl-optimization/82020
3198 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3199 IF_THEN_ELSE condition.
3201 2017-11-19 Jeff Law <law@redhat.com>
3203 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3204 of degenerates resulting from ignoring an edge.
3206 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3209 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3211 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3214 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3215 for uninitialied counts.
3217 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3220 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3222 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3225 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3227 2017-11-19 Tom de Vries <tom@codesourcery.com>
3229 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3232 2017-11-19 Tom de Vries <tom@codesourcery.com>
3234 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3237 2017-11-19 Tom de Vries <tom@codesourcery.com>
3239 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3242 2017-11-19 Tom de Vries <tom@codesourcery.com>
3244 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3246 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3248 2017-11-19 Tom de Vries <tom@codesourcery.com>
3250 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
3252 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
3253 semicolon after MCORE_EXPORT_NAME call.
3255 2017-11-19 Tom de Vries <tom@codesourcery.com>
3258 * vmsdbgout.c (vmsdbgout_early_finish): New function.
3259 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
3261 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
3263 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
3264 (cgraph_edge::clone): Cleanup updating of profile.
3265 * ipa-cp.c (update_profiling_info): Likewise.
3266 * ipa-inline-transform.c (inline_transform): Likewise.
3267 * ipa-inline.c (inline_small_functions): Add missing space to dump.
3268 * ipa-split.c (execute_split_functions): Do not split when function
3270 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
3271 * profile-count.c (profile_count::dump): Add global0.
3272 (profile_count::to_cgraph_frequency): Do not ICE when entry is
3274 (profile_count::to_sreal_scale): Likewise.
3275 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
3276 (profile_count::combine_with_ipa_count): New function.
3277 * profile-count.h (profile_guessed_global0adjusted): New.
3278 (profile_count::adjusted_zero): New.
3279 (profile_count::global0adjusted): New.
3280 (profile_count::combine_with_ipa_count): New.
3281 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
3282 correct profile of return block of split functions.
3283 (copy_cfg_body): Remove unused profile_count.
3284 (copy_body): Likewise.
3285 (expand_call_inline): Update.
3286 (tree_function_versioning): Update.
3288 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
3290 * hash-set.h (hash_set::empty): New.
3291 * tree-ssa-threadbackward.h: Delete.
3292 * tree-ssa-threadbackward.c (class thread_jumps): New.
3293 Move max_threaded_paths into class.
3294 (fsm_find_thread_path): Remove arguments that are now in class.
3295 (profitable_jump_thread_path): Rename to...
3296 (thread_jumps::profitable_jump_thread_path): ...this.
3297 (convert_and_register_jump_thread_path): Rename to...
3298 (thread_jumps::convert_and_register_current_path): ...this.
3299 (check_subpath_and_update_thread_path): Rename to...
3300 (thread_jumps::check_subpath_and_update_thread_path): ...this.
3301 (register_jump_thread_path_if_profitable): Rename to...
3302 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
3303 (handle_phi): Rename to...
3304 (thread_jumps::handle_phi): ...this.
3305 (handle_assignment): Rename to...
3306 (thread_jumps::handle_assignment): ...this.
3307 (fsm_find_control_statement_thread_paths): Rename to...
3308 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
3309 (find_jump_threads_backwards): Rename to...
3310 (thread_jumps::find_jump_threads_backwards): ...this.
3311 Initialize path local data.
3312 (pass_thread_jumps::execute): Call find_jump_threads_backwards
3313 from within thread_jumps class.
3314 (pass_early_thread_jumps::execute): Same.
3316 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3318 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
3320 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3322 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
3323 to canonical location.
3325 2017-11-17 Steve Ellcey <sellcey@cavium.com>
3328 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
3330 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
3332 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
3334 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
3335 rcrt1.o%s/grcrt1.o%s for -static-pie.
3337 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3339 * i386.c (ix86_multiplication_cost, ix86_division_cost,
3340 ix86_shift_rotate_cost): Break out from ...
3341 (ix86_rtx_costs): ... here.
3342 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
3345 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3347 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
3348 when entry block was promoted unlikely.
3349 (estimate_bb_frequencies): Increase frequency scale.
3350 * profile-count.h (profile_count): Export precision info.
3352 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3354 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
3355 disturbin profile of entry block.
3357 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3359 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
3362 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
3364 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
3365 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
3367 2017-11-17 Jeff Law <law@redhat.com>
3369 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
3370 from evrp_dom_walker class. Various methods moved into new class.
3371 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
3372 (evrp_range_analyzer::enter): New method.
3373 (evrp_range_analyzer::leave): New method.
3374 (evrp_dom_walker): Remove delegators no longer needed by this class.
3375 Replace vr_values data member with evrp_range_analyzer
3377 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
3378 method extracted from evrp_dom_walker::before_dom_children.
3379 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
3380 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
3382 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
3383 Add private copy constructor and move assignment operators.
3384 Privatize methods and class data where trivially possible.
3385 (evrp_dom_walker::cleanup): New function, extracted from
3386 execute_early_vrp. Simplify access to class data.
3388 * vr-values.h (get_output_for_vrp): Prototype.
3389 * vr-values.c (get_output_for_vrp): New function extracted from
3390 vrp_visit_assignment_or_call and extract_range_from_stmt.
3391 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
3393 2017-11-17 Luis Machado <luis.machado@linaro.org>
3395 * config/aarch64/aarch64.c
3396 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
3397 (qdf24xx_tunings) <autoprefetcher_model>: Set to
3398 tune_params::AUTOPREFETCHER_WEAK.
3400 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3403 * config/arm/arm.c (arm_valid_target_attribute_rec):
3404 Parse "arch=" and "+<ext>".
3405 (arm_valid_target_attribute_tree): Re-init global options.
3406 (arm_option_override): Make non-static.
3407 (arm_options_perform_arch_sanity_checks): Make errors fatal.
3408 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
3409 (__ARM_FEATURE_CRC32): Support undef.
3410 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
3411 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
3413 2017-11-17 David Malcolm <dmalcolm@redhat.com>
3415 * gdbinit.in (break-on-diagnostic): New command.
3417 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3419 * config/i386/i386.c (ix86_expand_epilogue): Change simple
3420 return to indirect jump for EH return if control-flow
3421 protection is enabled. Change explicit 'false' argument in
3422 pro_epilogue_adjust_stack with a value of
3424 * config/i386/i386.md (simple_return_indirect_internal):
3425 Remove SImode restriction to support 64-bit.
3427 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3429 * combine.c (added_notes_insn): New.
3430 (try_combine): Handle added_notes_insn like added_links_insn.
3431 Rewrite return value code.
3432 (distribute_notes): Set added_notes_insn to the earliest insn we added
3435 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
3437 PR rtl-optimization/82621
3438 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
3439 dest of one of those SETs is unused.
3441 2017-11-17 Richard Biener <rguenther@suse.de>
3444 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
3445 * tree-pretty-print.c (dump_generic_node): Handle
3446 annot_expr_parallel_kind.
3447 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
3448 * gimplify.c (gimple_boolify): Likewise.
3450 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3452 * config.gcc (extra_headers): Add cet.h for x86 targets.
3453 * config/i386/cet.h: New file.
3454 * doc/install.texi: Add --enable-cet/--disable-cet.
3456 2017-11-17 Richard Biener <rguenther@suse.de>
3458 PR tree-optimization/83017
3459 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
3460 (gen_parallel_loop): Properly count iterations.
3461 (parallelize_loops): Handle loop->can_be_parallel independent
3462 of flag_loop_parallelize_all. Make static profitability test match
3464 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
3465 * invoke.texi (parloops-min-per-thread): Document.
3467 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
3469 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
3470 upstreaming review comments.
3472 2017-11-17 Tamar Christina <tamar.christina@arm.com>
3474 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
3475 with fast unaligned access.
3476 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
3478 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
3480 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
3481 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
3483 2017-11-17 Richard Biener <rguenther@suse.de>
3485 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
3486 folding of references.
3488 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
3491 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
3492 of replacing call to strncmp with corresponding call to strcmp when
3495 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
3497 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
3498 option prefer-avx256 for skylake-avx512 configuration.
3499 * config/i386/i386.c (ix86_option_override_internal): Ditto.
3500 (get_builtin_code_for_version): Ditto.
3502 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3503 Monk Chiang <sh.chiang04@gmail.com>
3505 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
3506 (FIXED_REGISTERS): Reserve more register numbers.
3507 (CALL_USED_REGISTERS): Likewise.
3508 (REG_ALLOC_ORDER): Likewise.
3509 (REG_CLASS_CONTENTS): Likewise.
3510 (REGISTER_NAMES): Likewise.
3512 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
3513 Kito Cheng <kito.cheng@gmail.com>
3515 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
3517 * config/nds32/iterators.md: Add vector mode iterators and attributes.
3519 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
3521 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
3522 parameter list for vec_splats.
3524 2017-11-16 Joseph Myers <joseph@codesourcery.com>
3526 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
3528 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
3530 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
3533 * doc/install.texi: Document incompatibility of Perl >=5.6.26
3534 with the required version of automake 1.11.6.
3536 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
3538 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
3540 (DU_C2_3_power9): Correct reservation combinations.
3541 (FP_DIV_power9, VEC_DIV_power9): New.
3542 (power9-alu): Split out rotate/shift...
3543 (power9-rot): ...to here, correct dispatch resource.
3544 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
3546 (power9-fp): Correct latency.
3547 (power9-sdiv): Add div/sqrt resource.
3548 (power9-ddiv): Correct latency, add div/sqrt resource.
3549 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
3550 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
3552 (power9-qpdiv, power9-qpmul): Adjust resource usage.
3554 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3556 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
3557 switch statement mapping KF built-ins to TF built-ins if we don't
3558 have the proper ISA 3.0 assembler support.
3560 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3562 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
3563 (lower_emutls_function_body): Do not compute it.
3565 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3567 * ipa-split.c (split_bb_info): Turn time to sreal.
3568 (split_point): Likewise.
3569 (dump_split_point): Likewise.
3570 (fine_split_points): Likewise.
3571 (execute_split_functions): Only zero split_bbs; turn time to sreals.
3573 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3575 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
3578 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3580 * predict.c (combine_predictions_for_bb): Preserve zero predicted
3582 (expensive_function_p): Remove useless assert.
3583 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
3585 2017-11-16 Martin Sebor <msebor@redhat.com>
3587 PR tree-optimization/82588
3588 PR tree-optimization/82583
3589 * tree-vrp.c (check_array_ref): Handle flexible array members,
3590 string literals, and inner indices.
3591 (search_for_addr_array): Add detail to diagnostics.
3593 2017-11-16 Nathan Sidwell <nathan@acm.org>
3597 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
3598 (SET_DECL_ASSEMBLER_NAME): Forward to
3599 overwrite_decl_assembler_name.
3600 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
3601 (overwrite_decl_assembler_name): Declare.
3602 * tree.c (overwrite_decl_assembler_name): New.
3603 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
3604 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
3605 (LANG_HOOKS_INITIALIZER): Add it.
3606 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
3607 * langhooks.c (lhd_set_decl_assembler_name): Use
3608 SET_DECL_ASSEMBLER_NAME.
3609 (lhd_overwrite_decl_assembler_name): Default implementation.
3611 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3612 Jackson Woodruff <jackson.woodruff@arm.com>
3614 PR tree-optimization/71026
3615 * match.pd: Canonicalize constant multiplies in division.
3617 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
3619 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
3620 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
3621 Remove OPT_fomit_frame_pointer entry.
3622 * common/config/arc/arc-common.c: Likewise.
3623 * common/config/arm/arm-common.c: Likewise.
3624 * common/config/avr/avr-common.c: Likewise.
3625 * common/config/c6x/c6x-common.c: Likewise.
3626 * common/config/cr16/cr16-common.c: Likewise.
3627 * common/config/cris/cris-common.c: Likewise.
3628 * common/config/epiphany/epiphany-common.c: Likewise.
3629 * common/config/fr30/fr30-common.c: Likewise.
3630 * common/config/frv/frv-common.c: Likewise.
3631 * common/config/ia64/ia64-common.c: Likewise.
3632 * common/config/iq2000/iq2000-common.c: Likewise.
3633 * common/config/lm32/lm32-common.c: Likewise.
3634 * common/config/m32r/m32r-common.c: Likewise.
3635 * common/config/mcore/mcore-common.c: Likewise.
3636 * common/config/microblaze/microblaze-common.c: Likewise.
3637 * common/config/mips/mips-common.c: Likewise.
3638 * common/config/mmix/mmix-common.c: Likewise.
3639 * common/config/mn10300/mn10300-common.c: Likewise.
3640 * common/config/nios2/nios2-common.c: Likewise.
3641 * common/config/pa/pa-common.c: Likewise.
3642 * common/config/pdp11/pdp11-common.c: Likewise.
3643 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
3644 * common/config/riscv/riscv-common.c: Likewise.
3645 * common/config/rs6000/rs6000-common.c: Likewise.
3646 * common/config/rx/rx-common.c: Likewise.
3647 * common/config/s390/s390-common.c: Likewise.
3648 * common/config/sh/sh-common.c: Likewise.
3649 * common/config/sparc/sparc-common.c: Likewise.
3650 * common/config/tilegx/tilegx-common.c: Likewise.
3651 * common/config/tilepro/tilepro-common.c: Likewise.
3652 * common/config/v850/v850-common.c: Likewise.
3653 * common/config/visium/visium-common.c: Likewise.
3654 * common/config/xstormy16/xstormy16-common.c: Likewise.
3655 * common/config/xtensa/xtensa-common.c: Likewise.
3656 * invoke.texi (-fomit-frame-pointer): Update documentation.
3658 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
3660 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
3662 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3664 * cfg.c (scale_bbs_frequencies_int,
3665 cale_bbs_frequencies_gcov_type): Remove.
3666 * cfg.h (scale_bbs_frequencies_int,
3667 cale_bbs_frequencies_gcov_type): Remove.
3669 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3671 * tree-ssa-loop-manip.c
3672 (scale_dominated_blocks_in_loop): Update to profile counts.
3673 (tree_transform_and_unroll_loop): Likewise.
3675 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3677 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
3678 scale_bbs_frequencies_int.
3680 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3682 * final.c (compute_alignments): Use counts rather than frequencies.
3684 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3686 * cfgloopanal.c: Include sreal.h
3687 (average_num_loop_insns): Use counts and sreal for accounting.
3689 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3691 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
3694 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3695 Kito Cheng <kito.cheng@gmail.com>
3697 * config/nds32/constraints.md: Provide more constraints.
3698 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
3699 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
3700 support constraints usage.
3702 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3704 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
3706 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
3707 Kito Cheng <kito.cheng@gmail.com>
3709 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
3710 * config/nds32/nds32.opt: Refine the layout.
3711 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
3712 TARGET_EXT_STRING): Support new options.
3713 * config/nds32/nds32.h: Likewise.
3714 * config/nds32/nds32.md: Likewise.
3715 * config/nds32/nds32-predicates.c: Likewise.
3716 * config/nds32/constraints.md: Likewise.
3717 * common/config/nds32/nds32-common.c: Likewise.
3719 2017-11-16 Julia Koval <julia.koval@intel.com>
3722 * config/i386/gfniintrin.h: Add sse check.
3723 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
3725 2017-11-16 Julia Koval <julia.koval@intel.com>
3727 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
3728 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
3729 (ix86_handle_option): Handle -mavx512vbmi2.
3730 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
3731 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3732 * config/i386/i386-c.c (__AVX512VBMI2__): New.
3733 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
3734 (ix86_valid_target_attribute_inner_p): Ditto.
3735 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
3736 * config/i386/i386.opt (mavx512vbmi2): New option.
3737 * doc/invoke.texi: Add new option.
3739 2017-11-16 Julia Koval <julia.koval@intel.com>
3741 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
3742 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
3743 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
3744 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
3745 _mm512_gf2p8mul_epi8): New intrinsics.
3746 * config/i386/i386-builtin-types.def
3747 (V64QI_FTYPE_V64QI_V64QI): New type.
3748 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
3749 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
3750 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
3751 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
3752 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
3753 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3755 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
3757 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
3758 explicitly as a stream of bytes.
3760 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3762 * config/rs6000/altivec.h (vec_xst_be): New #define.
3763 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
3764 and externalize from *altivec_vperm_<mode>_internal.
3765 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
3767 (XL_BE_V8HI): Likewise.
3768 (XL_BE_V4SI): Likewise.
3769 (XL_BE_V4SI): Likewise.
3770 (XL_BE_V2DI): Likewise.
3771 (XL_BE_V4SF): Likewise.
3772 (XL_BE_V2DF): Likewise.
3773 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
3774 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
3775 all array entries with these keys: VSX_BUILTIN_VEC_XL,
3776 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
3777 VSX_BUILTIN_VEC_XST_BE.
3778 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
3779 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
3781 (altivec_init_builtins): Replace conditional calls to def_builtin
3782 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
3783 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
3784 with unconditional calls. Remove calls to def_builtin for
3785 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
3786 __builtin_vec_xst_be.
3787 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
3788 to define_expand, and add alternate RTL generation for P8.
3789 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
3790 vsx_ld_elemrev_v8hi.
3791 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
3792 add alternate RTL generation for P8.
3793 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
3794 vsx_ld_elemrev_v16qi.
3795 (vsx_st_elemrev_v8hi): Convert define_insn
3796 to define_expand, and add alternate RTL generation for P8.
3797 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
3798 vsx_st_elemrev_v8hi.
3799 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
3800 add alternate RTL generation for P8.
3801 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
3802 vsx_st_elemrev_v16qi.
3804 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
3807 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
3808 TARGET_AVX512ER check.
3809 (ix86_option_override_internal): Set MASK_VZEROUPPER if
3810 neither -mzeroupper nor -mno-zeroupper is used and
3811 TARGET_EMIT_VZEROUPPER is set.
3812 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
3813 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
3815 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3817 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3818 folding of vector compares.
3819 (fold_build_vec_cmp): New helper function.
3820 (fold_compare_helper): New helper function.
3821 (builtin_function_type): Add compare builtins to the list of functions
3822 having unsigned arguments. Cosmetic updates to comment indentation.
3823 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
3824 the not+eq combination.
3826 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3828 PR tree-optimization/82726
3829 PR tree-optimization/70754
3830 * tree-predcom.c (order_drefs_by_pos): New function.
3831 (combine_chains): Move code setting has_max_use_after to...
3832 (try_combine_chains): ...here. New parameter. Sort combined chains
3833 according to position information.
3834 (tree_predictive_commoning_loop): Update call to above function.
3835 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
3837 2017-11-15 Bin Cheng <bin.cheng@arm.com>
3839 PR tree-optimization/82726
3841 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3843 PR tree-optimization/70754
3844 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3845 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3846 combined stmt before it if not NULL.
3847 (combine_chains): Process refs reversely and compute dominance point
3851 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3853 PR tree-optimization/79663
3854 * tree-predcom.c (combine_chains): Process refs in reverse order
3855 only for ZERO length chains, and add explaining comment.
3857 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3859 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
3860 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
3861 * doc/invoke.texi (ARM Options): Add armv8.3-a.
3863 2017-11-15 Tamar Christina <tamar.christina@arm.com>
3865 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
3867 2017-11-15 Martin Liska <mliska@suse.cz>
3869 * tree-cfg.c (pass_warn_function_return::execute):
3870 Compare warn_return_type for greater than zero.
3872 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
3876 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
3877 to return true on Xeon and not on Xeon Phi.
3878 (ix86_check_avx256_register): Changed to ...
3879 (ix86_check_avx_upper_register): ... this. Add extra check for
3880 VALID_AVX512F_REG_OR_XI_MODE.
3881 (ix86_avx_u128_mode_needed): Changed
3882 ix86_check_avx256_register to ix86_check_avx_upper_register.
3883 (ix86_check_avx256_stores): Changed to ...
3884 (ix86_check_avx_upper_stores): ... this. Changed
3885 ix86_check_avx256_register to ix86_check_avx_upper_register.
3886 (ix86_avx_u128_mode_after): Changed
3887 avx_reg256_found to avx_upper_reg_found. Changed
3888 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3889 (ix86_avx_u128_mode_entry): Changed
3890 ix86_check_avx256_register to ix86_check_avx_upper_register.
3891 (ix86_avx_u128_mode_exit): Ditto.
3892 * config/i386/i386.h: (host_detect_local_cpu): New define.
3894 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3896 * config/arm/xgene1.md (xgene1): Split into automatons
3897 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3898 (xgene1_f_load): Adjust reservations and/or types.
3899 (xgene1_f_store): Likewise.
3900 (xgene1_load_pair): Likewise.
3901 (xgene1_store_pair): Likewise.
3902 (xgene1_fp_load1): Likewise.
3903 (xgene1_load1): Likewise.
3904 (xgene1_store1): Likewise.
3905 (xgene1_move): Likewise.
3906 (xgene1_alu): Likewise.
3907 (xgene1_simd): Likewise.
3908 (xgene1_bfm): Likewise.
3909 (xgene1_neon_load1): Likewise.
3910 (xgene1_neon_store1): Likewise.
3911 (xgene1_neon_logic): Likewise.
3912 (xgene1_neon_st1): Likewise.
3913 (xgene1_neon_ld1r): Likewise.
3914 (xgene1_alu_cond): Added.
3915 (xgene1_shift_reg): Likwise.
3916 (xgene1_bfx): Likewise.
3917 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3919 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3922 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3924 (can_widen_mult_without_libcall): New function.
3925 (expand_mul_overflow): If only checking unsigned mul overflow,
3926 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3927 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3928 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3929 (expand_DIVMOD): Formatting fix.
3930 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3931 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3932 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3934 PR tree-optimization/82977
3935 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3936 constructed temporary to strlen_to_stridx.put.
3938 2017-11-15 Martin Liska <mliska@suse.cz>
3940 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3941 * configure: Regenerate.
3943 2017-11-15 Martin Liska <mliska@suse.cz>
3946 * config/sh/sh-mem.cc: Use proper probability for
3949 2017-11-14 Jeff Law <law@redhat.com>
3951 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3952 the red zone for stack_clash_protection_final_dynamic_probe targets
3953 when the total dynamic stack size is zero bytes.
3955 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3956 blocks is post order.
3958 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3960 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3961 * final.c (rest_of_clean_state): Set it for the
3962 -fcompare-debug dump.
3963 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3964 class when TDF_COMPARE_DEBUG is set.
3966 * dwarf2out.c (gen_producer_string): Discard
3969 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3972 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3973 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3974 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3975 (__TGMATH_REAL_2_3): Remove macros.
3976 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3977 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3979 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3981 (remquo): Define using __TGMATH_REAL_3.
3983 2017-11-14 Jeff Law <law@redhat.com>
3985 * vr-values.c: New file with contents extracted from tree-vrp.c.
3986 * Makefile.in (OBJS): Add vr-values.o
3987 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3988 (set_value_range, set_and_canonicalize_value_range): Likewise.
3989 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3990 (value_range_constant_singleton, symbolic_range_p): Likewise.
3991 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3992 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3993 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3994 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3995 (find_case_label_range, find_case_label_index): Likewise.
3996 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3997 (range_int_cst_singleton_p, value_inside_range): Likewise.
3998 (get_single_symbol): Likewise.
3999 (switch_update): Move structure definition here.
4000 (to_remove_edges, to_update_switch_stmts): Provide externs.
4001 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4002 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4003 (vrp_val_is_min, set_value_range): Likewise.
4004 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4005 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4006 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4007 (range_is_nonnull, range_int_cst_p): Likewwise.
4008 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4009 (get_single_symbol, operand_less_p): Likewise
4010 (compare_values_warnv, compare_values): Likewise.
4011 (value_inside_range, value_range_constant_singleton): Likewise.
4012 (zero_nonzero_bitgs_from_vr): Likewise.
4013 (extract_range_from_binary_expr_1): Likewise.
4014 (overflow_comparison_p): Likewise.
4015 (to_remove_edges, to_update_switch_stmts): Likewise.
4016 (find_case_label-index, find_case_label_range): Likewise.
4017 (switch_update, set_value_range_to_nonnegative): Remove.
4018 (set_value_range_to_truthvalue): Likewise.
4019 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4020 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4021 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4022 (find_case_label_ranges, test_for_singularity): Likewise.
4023 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4024 (x_vr_values): Move to its remaining use site.
4026 2017-11-10 Jeff Law <law@redhat.com>
4028 * vr-values.h (VR_INITIALIZER): Move #define here.
4029 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4030 * Makefile.in (OBJS): Add tree-evrp.o
4031 * tree-vrp.h (assert_info): Move structure definition here.
4032 (set_value_range_to_varying): Prototype.
4033 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4034 (infer_value_range, register_edge_assert_for): Likewise.
4035 (stmt_interesting_for_vrp): Likewise.
4036 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4037 (set_value_range_to_varying): No longer static.
4038 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4039 (infer_value_range, register_edge_assert_for): Likewise.
4041 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4043 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4044 generate the XXBRD instruction.
4046 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4047 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4048 as being compatible if -mabi=ieeelongdouble.
4049 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4050 to setup float128 built-ins with hardware support.
4051 (BU_FLOAT128_HW_2): Likewise.
4052 (BU_FLOAT128_HW_3): Likewise.
4053 (BU_FLOAT128_HW_VSX_1): Likewise.
4054 (BU_FLOAT128_HW_VSX_2): Likewise.
4055 (scalar_extract_expq): Change float128 built-in functions to
4056 accommodate having both KFmode and TFmode functions. Use the
4057 KFmode variant as the default.
4058 (scalar_extract_sigq): Likewise.
4059 (scalar_test_neg_qp): Likewise.
4060 (scalar_insert_exp_q): Likewise.
4061 (scalar_insert_exp_qp): Likewise.
4062 (scalar_test_data_class_qp): Likewise.
4063 (sqrtf128_round_to_odd): Delete processing the round to odd
4064 built-in functions as special built-in functions, and define them
4065 as float128 built-ins. Use the KFmode variant as the default.
4066 (truncf128_round_to_odd): Likewise.
4067 (addf128_round_to_odd): Likewise.
4068 (subf128_round_to_odd): Likewise.
4069 (mulf128_round_to_odd): Likewise.
4070 (divf128_round_to_odd): Likewise.
4071 (fmaf128_round_to_odd): Likewise.
4072 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4073 support for KFmode and TFmode xststdcqp calls.
4074 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4075 point, switch the built-in handlers for the get/set float128
4076 exponent, get float128 mantissa, float128 test built-ins, and the
4077 float128 round to odd built-in functions. Eliminate creating the
4078 float128 round to odd built-in functions as special built-ins.
4079 (rs6000_init_builtins): Eliminate special creation of the float128
4080 round to odd built-in functions.
4081 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4082 function insns to support both TFmode and KFmode variants.
4083 (xsxsigqp_<mode>): Likewise.
4084 (xsiexpqpf_<mode>): Likewise.
4085 (xsiexpqp_<mode>): Likewise.
4086 (xststdcqp_<mode>): Likewise.
4087 (xststdcnegqp_<mode>): Likewise.
4088 (xststdcqp_<mode>): Likewise.
4090 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4092 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4093 unused path_in_freq_ptr parameter.
4094 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4096 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4098 * ipa-inline.c (edge_badness): Dump sreal frequency.
4099 (compute_inlined_call_time): Match natural implementaiton ...
4100 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4101 forgotten division by CGRAPH_FREQ_BASE.
4103 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4105 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4106 Solaris 11. Update comment.
4107 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4109 * config/sol2.h (STARTFILE_SPEC): Likewise.
4110 * configure: Regenerate.
4112 2017-11-14 Carl Love <cel@us.ibm.com>
4114 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4115 le_ and be_ prefixes to swap* variables. Remove
4116 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4119 2017-11-14 Jason Merrill <jason@redhat.com>
4121 Support GTY((cache)) on hash_map.
4122 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4123 (ggc_cache_remove): Override it instead of ggc_mx.
4124 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4125 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4126 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4127 (simple_cache_map_traits): Override maybe_mx.
4128 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4129 (hash_map): Friend gt_cleare_cache.
4130 (gt_cleare_cache): New.
4131 * tree.h (tree_cache_traits): New hash_map traits class.
4132 (tree_cache_map): New typedef.
4134 2017-11-14 Richard Biener <rguenther@suse.de>
4136 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4137 paramter and handling.
4138 (cleanup_control_flow_bb): Likewise.
4139 (cleanup_control_flow_pre): New helper performing a DFS walk
4140 to call cleanup_control_flow_bb in PRE order.
4141 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4142 via cleanup_control_flow_pre.
4144 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4146 * config/aarch64/aarch64-simd.md
4147 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4148 (*aarch64_simd_bsl<mode>_alt): Likewise.
4149 (aarch64_simd_bsldi_internal): New.
4150 (aarch64_simd_bsldi_alt): Likewise.
4152 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4154 * tracer.c (better_p): Do not compare frequencies.
4155 * reg-stack.c (better_edge): Likewise.
4156 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4159 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4161 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4162 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4163 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4164 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4165 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4166 * predict.c (maybe_hot_frequency_p): Inline to ...
4167 (maybe_hot_count_p): ... here; rewrite to counts.
4168 (counts_to_freqs): Rename to ...
4169 (update_max_bb_count): ... this one.
4170 (expensive_function_p): Use counts.
4171 (estimate_bb_frequencies): Update.
4172 (rebuild_frequencies): Update.
4173 * predict.h (counts_to_freqs): Rename to ...
4174 (update_max_bb_count): ... this one.
4175 * profile.c (compute_branch_probabilities): Add debug info
4176 * tree-inline.c (expand_call_inline): Update debug info.
4177 (optimize_inline_calls): Use update_max_bb_count..
4178 (tree_function_versioning): Use update_max_bb_count..
4179 * value-prof.c (gimple_value_profile_transformations):
4180 Do not use update_max_bb_count.
4182 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4184 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4185 always use frequencies.
4187 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4189 * bb-reorder.c: Remove frequencies from comments.
4190 (better_edge_p): Use profile counts.
4191 (find_traces): Dump profile counts.
4192 (rotate_loop): Use profile counts.
4193 (find_traces_1_round): Likewise.
4194 (connect_better_edge_p): Use counts instead of probabilities for
4196 (copy_bb_p): Drop early check for non-0 frequency.
4197 (sanitize_hot_paths): Update comments.
4199 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4201 * ipa-split.c (struct split_point): Add count.
4202 (consider_split): Do not compute incoming frequency; compute incoming
4203 count and store it to split_point.
4204 (split_function): Set count of the call to split part correctly.
4206 2017-11-13 Carl Love <cel@us.ibm.com>
4208 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4210 2017-11-13 Tom Tromey <tom@tromey.com>
4212 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4214 2017-11-13 Carl Love <cel@us.ibm.com>
4216 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4217 Add support for builtins:
4218 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4219 vector {un,}signed {char,int,short},
4220 vector {un,}signed {char,int,short}) arguments.
4221 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4222 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4223 Add BU_P9V_AV_2 expansions for the builtins.
4224 * config/rs6000/altivec.h (vec_first_match_index,
4225 vec_first_mismatch_index, vec_first_match_or_eos_index,
4226 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4227 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4228 new extern declaration.
4229 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4230 * config/rs6000/vsx.md (first_match_index_<mode>,
4231 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4232 first_mismatch_or_eos_index_<mode>): Add define expand.
4233 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4234 * doc/extend.texi: Update the built-in documenation file for the new
4237 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4239 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4240 into the min/max operations for _Float<N> and _Float<N>X types.
4242 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4245 * dwarf2out.c (do_eh_frame): New static variable.
4246 (dwarf2out_begin_prologue): Set it.
4247 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4249 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4251 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
4253 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
4255 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4258 * debug.h (dwarf2out_do_eh_frame): Declare.
4259 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
4260 (dwarf2out_do_frame): Use it.
4261 (dwarf2out_do_cfi_asm): Likewise.
4262 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
4263 (dwarf2out_assembly_start): Likewise.
4264 (dwarf2out_begin_prologue): Fix comment.
4265 * toplev.c (compile_file): Always call dwarf2out_frame_finish
4266 if the target needs either debug or unwind DWARF2 info.
4267 * lto-opts.c (lto_write_options): Do not save -fexceptions,
4268 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
4269 -ftrapping-math, -ftrapv and -fwrapv.
4271 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4273 * cgraph.c (cgraph_edge::sreal_frequency): New function.
4274 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
4275 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
4276 (estimate_edge_size_and_time): Likewise.
4277 (ipa_merge_fn_summary_after_inlining): Likewise.
4278 * ipa-inline.c (cgraph_freq_base_rec): Remove.
4279 (compute_uninlined_call_time): Use sreal_frequency.
4280 (compute_inlined_call_time): Likewise.
4281 (ipa_inline): Do not initialize cgraph_freq_base_rec.
4282 * profile-count.c: Include sreal.h.
4283 (profile_count::to_sreal_scale): New.
4284 * profile-count.h: Forward declare sreal.
4285 (profile_count::to_sreal_scale): Declare.
4287 2017-11-13 Nathan Sidwell <nathan@acm.org>
4289 * diagnostic.c (maybe_line_and_column): New.
4290 (diagnostic_get_location_text): Use it.
4291 (diagnostic_report_current_module): Likewise.
4292 (test_diagnostic_get_location_text): Add tests.
4294 2017-11-13 Luis Machado <luis.machado@linaro.org>
4296 * doc/md.texi (Specifying processor pipeline description): Fix
4297 incorrect latency for the div instruction example.
4299 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4301 PR tree-optimization/78821
4302 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
4303 that bit_not_p is the same.
4304 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
4305 (split_group): Count precisely bit_not_p bits in each statement.
4306 (invert_op): New function.
4307 (imm_store_chain_info::output_merged_store): Use invert_op to
4308 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
4309 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
4311 2017-11-13 Martin Liska <mliska@suse.cz>
4313 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
4314 (struct source_info): Likewise.
4315 (add_branch_counts): Likewise.
4316 (add_line_counts): Likewise.
4317 (function_summary): Likewise.
4318 (output_intermediate_line): Likewise.
4319 (generate_results): Likewise.
4321 2017-11-13 Martin Liska <mliska@suse.cz>
4323 * gcov.c (struct block_info): Remove typedef for block_t.
4324 (struct line_info): Likewise.
4325 (line_info::has_block): Likewise.
4326 (EXIT_BLOCK): Likewise.
4327 (unblock): Likewise.
4328 (circuit): Likewise.
4329 (get_cycles_count): Likewise.
4330 (process_file): Likewise.
4331 (read_graph_file): Likewise.
4332 (solve_flow_graph): Likewise.
4333 (find_exception_blocks): Likewise.
4334 (add_line_counts): Likewise.
4335 (accumulate_line_info): Likewise.
4336 (output_line_details): Likewise.
4338 2017-11-13 Martin Liska <mliska@suse.cz>
4340 * gcov.c (struct arc_info): Remove typedef for arc_t.
4341 (struct line_info): Likewise.
4342 (add_branch_counts): Likewise.
4343 (output_branch_count): Likewise.
4344 (function_info::~function_info): Likewise.
4345 (circuit): Likewise.
4346 (output_intermediate_line): Likewise.
4347 (read_graph_file): Likewise.
4348 (solve_flow_graph): Likewise.
4349 (find_exception_blocks): Likewise.
4350 (add_line_counts): Likewise.
4351 (accumulate_line_info): Likewise.
4352 (output_line_details): Likewise.
4353 (output_function_details): Likewise.
4355 2017-11-13 Martin Liska <mliska@suse.cz>
4357 * gcov.c (struct function_info): Remove typedef for function_t.
4358 (struct source_info): Likewise.
4359 (source_info::get_functions_at_location): Likewise.
4360 (solve_flow_graph): Likewise.
4361 (find_exception_blocks): Likewise.
4362 (add_line_counts): Likewise.
4363 (output_intermediate_file): Likewise.
4364 (process_file): Likewise.
4365 (generate_results): Likewise.
4366 (release_structures): Likewise.
4367 (read_graph_file): Likewise.
4368 (read_count_file): Likewise.
4369 (accumulate_line_counts): Likewise.
4370 (output_lines): Likewise.
4372 2017-11-13 Martin Liska <mliska@suse.cz>
4374 * gcov.c (function_info::function_info): Remove num_counts
4375 and add vector<gcov_type>.
4376 (function_info::~function_info): Use the vector.
4377 (process_file): Likewise.
4378 (read_graph_file): Likewise.
4379 (read_count_file): Likewise.
4380 (solve_flow_graph): Likewise.
4382 2017-11-13 Martin Liska <mliska@suse.cz>
4384 * gcov.c (function_info::is_artificial): New function.
4385 (process_file): Erase all artificial early.
4386 (generate_results): Skip as all artificial are already
4389 2017-11-13 Martin Liska <mliska@suse.cz>
4391 * gcov.c (read_graph_file): Store to global vector of functions.
4392 (read_count_file): Iterate the vector.
4393 (process_file): Likewise.
4394 (generate_results): Likewise.
4395 (release_structures): Likewise.
4397 2017-11-13 Jakub Jelinek <jakub@redhat.com>
4399 PR tree-optimization/82954
4400 * gimple-ssa-store-merging.c
4401 (imm_store_chain_info::coalesce_immediate_stores): If
4402 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
4404 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
4406 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4407 Upddate call to ENDIAN_LANE_N.
4408 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
4409 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
4410 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
4411 and use aarch64_endian_lane_rtx.
4412 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
4414 2017-11-12 Tom de Vries <tom@codesourcery.com>
4416 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
4418 2017-11-12 Tom de Vries <tom@codesourcery.com>
4420 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
4422 2017-11-12 Tom de Vries <tom@codesourcery.com>
4424 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
4426 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4427 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
4428 * defaults.h (ASM_OUTPUT_LABELREF): Same.
4430 2017-11-11 Martin Sebor <msebor@redhat.com>
4433 * doc/extend.texi (attribute nonstring): Remove spurious argument.
4436 * prefic.c (translate_name): Replace strncpy with memcpy to
4437 avoid -Wstringop-truncation.
4439 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4441 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
4443 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4445 * predict.c (maybe_hot_frequency_p): Do not use cfun.
4447 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4449 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
4452 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
4454 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
4455 enable generating XXBRH if the value is in a vector register.
4456 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
4457 value is in a vector register.
4458 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
4459 register to register bswap64's instead of doing the GPR sequence
4460 used on previous machines.
4461 (bswapdi2_xxbrd): New insn.
4462 (bswapdi2_reg): Disallow on ISA 3.0.
4463 (register to register bswap64 splitter): Do not split the insn on
4464 ISA 3.0 systems that use XXBRD.
4466 2017-11-10 Martin Sebor <msebor@redhat.com>
4469 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
4471 (find_subframework_file): Same.
4473 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
4475 * auto-profile.c (afdo_indirect_call): Drop frequency.
4476 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
4477 (cgraph_node::create_edge): Drop frequency argument.
4478 (cgraph_node::create_indirect_edge): Drop frequency argument.
4479 (cgraph_edge::make_speculative): Drop frequency arguments.
4480 (cgraph_edge::resolve_speculation): Do not update frequencies
4481 (cgraph_edge::dump_edge_flags): Do not dump frequency.
4482 (cgraph_node::dump): Check consistency in IPA mode.
4483 (cgraph_edge::maybe_hot_p): Use IPA counter.
4484 (cgraph_edge::verify_count_and_frequency): Rename to ...
4485 (cgraph_edge::verify_count): ... this one; drop frequency checking.
4486 (cgraph_node::verify_node): Update.
4487 * cgraph.h (struct cgraph_edge): Drop frequency.
4488 (cgraph_edge::frequency): New function.
4489 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
4491 (cgraph_edge::rebuild_edges): Likewise.
4492 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
4493 (duplicate_thunk_for_node): Do not pass frequency.
4494 (cgraph_node::create_clone): Scale only counts.
4495 (cgraph_node::create_virtual_clone): Do not pass frequency.
4496 (cgraph_node::create_edge_including_clones): Do not pass frequency.
4497 (cgraph_node::create_version_clone): Do not pass frequency.
4498 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
4499 (cgraph_node::expand_thunk): Do not pass frequency.
4500 (cgraph_node::create_wrapper): Do not pass frequency.
4501 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
4503 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
4504 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
4505 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
4506 (gather_caller_stats): Use frequency function.
4507 (ipcp_cloning_candidate_p): Use frequency function.
4508 (ipcp_propagate_stage): Use frequency function.
4509 (get_info_about_necessary_edges): Use frequency function.
4510 (update_profiling_info): Update only IPA profile.
4511 (update_specialized_profile): Use frequency functoin.
4512 (perhaps_add_new_callers): Update only IPA profile.
4513 * ipa-devirt.c (ipa_devirt): Use IPA profile.
4514 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
4515 (dump_ipa_call_summary): Use frequency function.
4516 (estimate_edge_size_and_time): Use frequency function.
4517 (ipa_merge_fn_summary_after_inlining): Use frequency function.
4518 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
4519 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
4520 (update_noncloned_counts): ... ths one; scale counts only.
4521 (clone_inlined_nodes): Do not scale frequency.
4522 (inline_call): Do not pass frequency.
4523 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
4524 (compute_inlined_call_time): Use IPA profile.
4525 (want_inline_small_function_p): Use IPA profile.
4526 (want_inline_self_recursive_call_p): Use IPA profile.
4527 (edge_badness): Use IPA profile.
4528 (lookup_recursive_calls): Use IPA profile.
4529 (recursive_inlining): Do not pass frequency.
4530 (resolve_noninline_speculation): Do not update frequency.
4531 (inline_small_functions): Collect max of IPA profile.
4532 (dump_overall_stats): Dump IPA porfile.
4533 (dump_inline_stats): Dump IPA porfile.
4534 (ipa_inline): Collect IPA stats.
4535 * ipa-inline.h (clone_inlined_nodes): Update prototype.
4536 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
4537 (ipa_propagate_frequency): Use frequency function.
4538 (ipa_profile): Cleanup.
4539 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
4540 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
4541 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
4542 (input_node): Do not stream frequency.
4543 (input_edge): Do not stream frequency.
4544 (merge_profile_summaries): Scale only IPA profiles.
4545 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
4546 * predict.c (drop_profile): Do not recompute frequency.
4547 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
4548 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
4549 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
4550 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
4551 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
4552 * tree-inline.c (copy_bb): Do not scale frequency.
4553 (expand_call_inline): Do not scale frequency.
4554 (tree_function_versioning): Do not scale frequency.
4555 * ubsan.c (ubsan_create_edge): Do not pass frequency.
4557 2017-11-10 Julia Koval <julia.koval@intel.com>
4559 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
4560 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
4561 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
4562 (_mm256_maskz_gf2p8affine_epi64_epi8)
4563 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
4564 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
4565 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
4566 (__builtin_ia32_vgf2p8affineqb_v32qi)
4567 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
4568 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
4570 2017-11-10 Tamar Christina <tamar.christina@arm.com>
4574 (arm_option_override): Refactor.
4575 (arm_option_reconfigure_globals): New.
4576 (arm_options_perform_arch_sanity_checks): New.
4577 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
4579 (arm_options_perform_arch_sanity_checks): Likewise
4581 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
4583 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
4584 (power9-qpmul): New.
4585 * rs6000/rs6000.md ("type" attr): Add qmul.
4586 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
4587 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
4588 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
4590 2017-11-10 Martin Sebor <msebor@redhat.com>
4593 * builtins.c (compute_objsize): Handle arrays that
4594 compute_builtin_object_size likes to fail for. Make extern.
4595 * builtins.h (compute_objsize): Declare.
4596 (check_strncpy_sizes): New function.
4597 (expand_builtin_strncpy): Call check_strncpy_sizes.
4598 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
4599 -Wstringop-truncation.
4600 (gimple_fold_builtin_strncat): Same.
4601 * gimple.c (gimple_build_call_from_tree): Set call location.
4602 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
4603 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
4604 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
4605 (handle_builtin_strlen): Use strlen_to_stridx.
4606 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
4608 Use strlen_to_stridx.
4609 (pass_strlen::execute): Release strlen_to_stridx.
4610 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
4611 (-Wstringop-truncation): Document new option.
4613 2017-11-10 Martin Liska <mliska@suse.cz>
4615 PR gcov-profile/82702
4616 * gcov.c (main): Handle intermediate files in a different
4618 (get_gcov_intermediate_filename): New function.
4619 (output_gcov_file): Remove support of intermediate files.
4620 (generate_results): Allocate intermediate file.
4621 (release_structures): Clean-up properly fn_end.
4622 (output_intermediate_file): Start iterating with line 1.
4624 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4626 PR tree-optimization/82929
4627 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4628 ops_swapped_p non-static data member.
4629 (store_immediate_info::store_immediate_info): Clear it.
4630 (imm_store_chain_info::coalesce_immediate_stores): If swapping
4631 ops set ops_swapped_p.
4632 (count_multiple_uses): Handle ops_swapped_p.
4634 2017-11-10 Martin Liska <mliska@suse.cz>
4636 * coverage.c (coverage_init): Stream information about
4637 support of has_unexecuted_blocks.
4638 * doc/gcov.texi: Document that.
4639 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
4640 * gcov.c (read_graph_file): Likewise.
4641 (output_line_beginning): Fix a small issue with
4644 2017-11-10 Bin Cheng <bin.cheng@arm.com>
4646 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
4647 reference of trivial component.
4649 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4652 * gimple-ssa-store-merging.c
4653 (pass_store_merging::terminate_all_aliasing_chains): For
4654 gimple_store_p stmts also call refs_output_dependent_p.
4656 PR rtl-optimization/82913
4657 * compare-elim.c (try_merge_compare): Punt if def_insn is not
4660 2017-11-09 Jeff Law <law@redhat.com>
4662 * vr-values.h: New file with vr_values class.
4663 * tree-vrp.c: Include vr-values.h
4664 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
4665 data objects into the vr_values class.
4666 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
4667 (get_value_range): Make it a member function within vr_values class.
4668 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
4669 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
4670 (op_with_constant_singleton_value_range): Likewise.
4671 (extract_range_for_var_from_comparison_expr): Likewise.
4672 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
4673 (extract_range_from_binary_expr): Likewise.
4674 (extract_range_from_unary_expr): Likewise.
4675 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
4676 (check_for_binary_op_overflow, extract_range_basic): Likewise.
4677 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
4678 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
4679 (compare_name_with_value, compare_names): Likewise.
4680 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
4681 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
4682 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
4683 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
4684 (extract_range_from_phi_node): Likewise.
4685 (simplify_truth_ops_using_ranges): Likewise.
4686 (simplify_div_or_mod_using_ranges): Likewise.
4687 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
4688 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
4689 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
4690 (simplify_float_conversion_using_ranges): Likewise.
4691 (simplify_internal_call_using_ranges): Likewise.
4692 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
4693 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
4694 poitner via x_vr_values for calls into gimple folder.
4695 (vrp_initialize_lattice): Make this the vr_values ctor.
4696 (vrp_free_lattice): Make this the vr_values dtor.
4697 (set_vr_value): New function.
4698 (class vrp_prop): Add vr_values data member. Add various member
4699 functions as well as member functions that delegate to vr_values.
4700 (check_array_ref): Make a member function within vrp_prop class.
4701 (search_for_addr_array, vrp_initialize): Likewise.
4702 (vrp_finalize): Likewise. Revamp to avoid direct access to
4703 vr_value, values_propagated, etc.
4704 (check_array_bounds): Extract vrp_prop class instance pointer from
4705 walk info structure. Use it to call member functions.
4706 (check_all_array_refs): Make a member function within vrp_prop class.
4707 Smuggle class instance pointer via walk info structure.
4708 (x_vr_values): New local static.
4709 (vrp_valueize): Use x_vr_values to get class instance.
4710 (vr_valueize_1): Likewise.
4711 (class vrp_folder): Add vr_values data member. Add various member
4712 functions as well as member functions that delegate to vr_values.
4713 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
4714 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
4715 class instance from vr_values. Use it to call member functions.
4716 (vrp_dom_walker): Add vr_values data member.
4717 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
4718 instance via x_vr_values.
4719 (identify_jump_threads): Accept vr_values as argument. Store
4720 it into the walker structure.
4721 (evrp_dom_walker): Add vr_values class data member. Add various
4723 (evrp_dom_walker::try_find_new_range): Use vr_values data
4724 member to access the memory allocator.
4725 (evrp_dom_walker::before_dom_children): Store vr_values class
4726 instance into the vrp_folder class.
4727 (evrp_dom_walker::push_value_range): Rework to avoid direct
4728 access to num_vr_values and vr_value.
4729 (evrp_dom_walker::pop_value_range): Likewise.
4730 (execute_early_vrp): Remove call to vrp_initialize_lattice.
4731 Use vr_values to get to dump_all_value_ranges member function.
4732 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
4733 and simplify_cond_using_ranges_2 via vrp_prop class instance.
4734 Pass vr_values class instance down to identify_jump_threads.
4735 Remove call to vrp_free_lattice.
4736 (debug_all_value_ranges): Remove.
4738 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
4739 (vrp_folder): Likewise.
4741 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4742 Get it from the existing bitmap instead.
4743 (vrp_intersect_ranges_1): Likewise.
4745 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4747 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
4749 (store_immediate_info::store_immediate_info): Add bitnotp argument,
4750 set bit_not_p to it.
4751 (imm_store_chain_info::coalesce_immediate_stores): Break group
4752 if bit_not_p is different.
4753 (count_multiple_uses, split_group,
4754 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
4755 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
4756 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
4757 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
4758 to store_immediate_info ctor.
4760 2017-11-09 Jim Wilson <jimw@sifive.com>
4762 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4763 (scan_prog_file): Likewise.
4765 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4767 * bb-reorder.c (max_entry_frequency): Remove.
4768 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
4769 connect_traces, push_to_next_round_p): Remove prototypes.
4770 (find_traces_1_round): Use counts only.
4771 (push_to_next_round_p): Likewise.
4772 (find_traces): Likewise.
4773 (rotate_loop): Likewise.
4774 (find_traces_1_round): Likewise.
4775 (connect_traces): Likewise.
4776 (edge_order): Likewise.
4778 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
4780 * config/arm/arm.c (output_return_instruction): Add comments to
4781 indicate requirement for cmse_nonsecure_entry return to account
4782 for the size of clearing instruction output here.
4783 (thumb_exit): Likewise.
4784 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
4785 return in hardfloat mode.
4787 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
4789 * config/rs6000/rs6000.c (machine_function): Add a bool,
4790 "toc_is_wrapped_separately".
4791 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
4792 if it wasn't explicitly set or unset, we are optimizing for speed, and
4793 doing separate shrink-wrapping.
4794 (rs6000_get_separate_components): Enable the TOC component if
4795 saving the TOC register in the prologue.
4796 (rs6000_components_for_bb): Handle the TOC component.
4797 (rs6000_emit_prologue_components): Store the TOC register where needed.
4798 (rs6000_set_handled_components): Mark TOC as handled, if handled.
4799 (rs6000_emit_prologue): Don't save the TOC if that is already done.
4801 2017-11-09 Martin Jambor <mjambor@suse.cz>
4803 * ipa-param-manipulation.c: New file.
4804 * ipa-param-manipulation.h: Likewise.
4805 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
4806 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
4807 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
4808 (ipa_parm_adjustment): Likewise.
4809 (ipa_parm_adjustment_vec): Likewise.
4810 (ipa_get_vector_of_formal_parms): Moved declaration to
4811 ipa-param-manipulation.h.
4812 (ipa_get_vector_of_formal_parm_types): Likewise.
4813 (ipa_modify_formal_parameters): Likewise.
4814 (ipa_modify_call_arguments): Likewise.
4815 (ipa_combine_adjustments): Likewise.
4816 (ipa_dump_param_adjustments): Likewise.
4817 (ipa_modify_expr): Likewise.
4818 (ipa_get_adjustment_candidate): Likewise.
4819 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
4820 ipa-param-manipulation.c.
4821 (ipa_get_vector_of_formal_parm_types): Likewise.
4822 (ipa_modify_formal_parameters): Likewise.
4823 (ipa_modify_call_arguments): Likewise.
4824 (ipa_modify_expr): Likewise.
4825 (get_ssa_base_param): Likewise.
4826 (ipa_get_adjustment_candidate): Likewise.
4827 (index_in_adjustments_multiple_times_p): Likewise.
4828 (ipa_combine_adjustments): Likewise.
4829 (ipa_dump_param_adjustments): Likewise.
4830 * tree-sra.c: Also include ipa-param-manipulation.h
4831 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
4834 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4835 Alan Hayward <alan.hayward@arm.com>
4836 David Sherwood <david.sherwood@arm.com>
4838 * doc/sourcebuild.texi (vect_masked_store): Document.
4840 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4841 Alan Hayward <alan.hayward@arm.com>
4842 David Sherwood <david.sherwood@arm.com>
4844 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
4846 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4847 Alan Hayward <alan.hayward@arm.com>
4848 David Sherwood <david.sherwood@arm.com>
4850 * doc/sourcebuild.texi (vect_variable_length): Document.
4852 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4853 Alan Hayward <alan.hayward@arm.com>
4854 David Sherwood <david.sherwood@arm.com>
4856 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
4858 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4859 Alan Hayward <alan.hayward@arm.com>
4860 David Sherwood <david.sherwood@arm.com>
4862 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
4864 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4865 Alan Hayward <alan.hayward@arm.com>
4866 David Sherwood <david.sherwood@arm.com>
4868 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
4869 previously undocumented selectors.
4870 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
4872 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4874 * doc/rtl.texi (const_vector): Say that elements can be
4875 const_wide_ints too.
4876 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
4877 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
4878 (gen_vec_duplicate): Use it instead of CONSTANT_P.
4879 * optabs.c (expand_vector_broadcast): Likewise.
4881 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4882 Alan Hayward <alan.hayward@arm.com>
4883 David Sherwood <david.sherwood@arm.com>
4885 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4886 scaled index even if the unscaled address was invalid.
4887 Don't increase the complexity of using a scale in that case.
4889 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4890 Alan Hayward <alan.hayward@arm.com>
4891 David Sherwood <david.sherwood@arm.com>
4893 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4894 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4895 UNITS_PER_WORD bytes.
4896 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4897 based on REGMODE_NATURAL_SIZE of the inner mode.
4898 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4899 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4900 * expmed.c (lowpart_bit_field_p): Divide the value up into
4901 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4902 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4903 whether something is likely to occupy more than one register.
4905 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4908 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4909 profile_count::adjust_for_ipa_scaling.
4910 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4911 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4913 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4915 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4917 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4918 (split_group): Add total_orig and total_new arguments, estimate the
4919 number of statements related to the store group without store merging
4920 and with store merging.
4921 (imm_store_chain_info::output_merged_store): Adjust split_group
4922 callers, punt if estimated number of statements with store merging
4923 is not smaller than estimated number of statements without it.
4925 (handled_load): Remove has_single_use checks.
4926 (pass_store_merging::process_store): Likewise.
4928 2017-11-09 Richard Biener <rguenther@suse.de>
4930 PR tree-optimization/82902
4931 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4933 2017-11-09 Martin Liska <mliska@suse.cz>
4936 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4939 2017-11-09 Martin Liska <mliska@suse.cz>
4941 PR tree-optimization/82669
4942 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4944 2017-11-09 Martin Liska <mliska@suse.cz>
4946 PR gcov-profile/48463
4947 * coverage.c (coverage_begin_function): Output also end locus
4948 of a function and information whether the function is
4950 * gcov-dump.c (tag_function): Parse and print the information.
4951 * gcov.c (INCLUDE_MAP): Add include.
4952 (INCLUDE_SET): Likewise.
4953 (struct line_info): Move earlier in the source file because
4954 of vector<line_info> in function_info structure.
4955 (line_info::line_info): Likewise.
4956 (line_info::has_block): Likewise.
4957 (struct source_info): Add new member index.
4958 (source_info::get_functions_at_location): New function.
4959 (function_info::group_line_p): New function.
4960 (output_intermediate_line): New function.
4961 (output_intermediate_file): Use the mentioned function.
4962 (struct function_start): New.
4963 (struct function_start_pair_hash): Likewise.
4964 (process_file): Add code that identifies group functions.
4965 Assign lines either to global or function scope.
4966 (generate_results): Skip artificial functions.
4967 (find_source): Assign index for each source file.
4968 (read_graph_file): Read new flag artificial and end_line.
4969 (add_line_counts): Assign it either to global of function scope.
4970 (accumulate_line_counts): Isolate core of the function to
4971 accumulate_line_info and call it for both function and global
4973 (accumulate_line_info): New function.
4974 (output_line_beginning): Fix GNU coding style.
4975 (print_source_line): New function.
4976 (output_line_details): Likewise.
4977 (output_function_details): Likewise.
4978 (output_lines): Iterate both source (global) scope and function
4980 (struct function_line_start_cmp): New class.
4981 * doc/gcov.texi: Reflect changes in documentation.
4983 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4986 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4987 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4988 and similarly for not instead of neg.
4990 2017-11-08 Andi Kleen <ak@linux.intel.com>
4992 * config/i386/i386.opt: Add -mforce-indirect-call.
4993 * config/i386/predicates.md: Check for flag_force_indirect_call.
4994 * doc/invoke.texi: Document -mforce-indirect-call
4996 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4998 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5000 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5001 config/riscv/riscv.c (predict.h): New include.
5002 (riscv_slow_unaligned_access_p): No longer static.
5003 (riscv_block_move_straight): Add require.
5004 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5006 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5009 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5010 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5011 nonimmediate_operand predicate for operand 1 instead of
5014 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5016 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5018 * config/aarch64/constraints.md (Uml): New constraint.
5019 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5022 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5024 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5025 of two vec_duplicates into a vec_concat.
5027 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5029 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5030 Simplify vec_merge of vec_duplicate and vec_concat.
5031 * config/aarch64/constraints.md (Utq): New constraint.
5032 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5035 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5037 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5038 Simplify vec_merge of vec_duplicate and const_vector.
5039 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5041 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5042 mode iterator. Update predicate on operand 1 to
5043 handle non-const_vec constants. Delete constraints.
5044 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5046 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5048 PR tree-optimization/78821
5049 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5051 (store_operand_info::store_operand_info): Initialize it to false.
5052 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5053 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5054 store in the group, and if chain_info is non-NULL, to ignore altogether
5056 (compatible_load_p): Fail if bit_not_p does not match.
5057 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5058 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5059 (pass_store_merging::process_store): Adjust
5060 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5061 call terminate_all_aliasing_chains newly when adding a store into
5062 a chain with non-NULL chain_info.
5064 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5066 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5068 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5070 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5072 (aarch64_layout_frame): Initialise emit_frame_chain.
5073 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5074 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5076 2017-11-08 Martin Liska <mliska@suse.cz>
5078 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5079 of gimple_call_internal_p.
5081 2017-11-07 Tom de Vries <tom@codesourcery.com>
5083 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5085 2017-11-07 Tom de Vries <tom@codesourcery.com>
5087 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5090 2017-11-08 Martin Liska <mliska@suse.cz>
5093 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5095 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5097 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5099 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5102 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5103 (*cmp<mode>_ccz_1): New insn with $k alternative.
5106 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5107 RTX_COMM_COMPARE as commutative as well.
5108 (ix86_binary_operator_ok): Formatting fix.
5109 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5110 *<code><mode>3<mask_name><round_saeonly_name>,
5111 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5112 *<s>mul<mode>3_highpart<mask_name>,
5113 *vec_widen_umult_even_v16si<mask_name>,
5114 *vec_widen_umult_even_v8si<mask_name>,
5115 *vec_widen_umult_even_v4si<mask_name>,
5116 *vec_widen_smult_even_v16si<mask_name>,
5117 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5118 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5119 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5120 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5121 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5122 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5123 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5124 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5125 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5126 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5127 ix86_binary_operator_ok. Formatting fixes.
5128 (*<plusminus_insn><mode>3<mask_name><round_name>,
5129 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5132 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5134 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5135 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5136 eq and ne if TARGET_ISEL.
5137 (cmp): New code_iterator.
5138 (UNS, UNSU_, UNSIK): New code_attrs.
5139 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5140 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5142 ("eq<mode>3"): ... this.
5143 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5145 ("ne<mode>3"): ... this.
5147 2017-11-07 Julia Koval <julia.koval@intel.com>
5150 * common/config/i386/i386-common.c
5151 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5152 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5153 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5154 * config/i386/i386.opt: Ditto.
5155 * config/i386/i386.c (ix86_target_string): Ditto.
5156 (ix86_option_override_internal): Ditto.
5157 (ix86_init_mpx_builtins): Move MPX to args2.
5158 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5159 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5160 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5161 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5162 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5163 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5164 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5166 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5169 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5170 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5171 (zero-extendsidi peephole2): Remove peephole.
5173 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5176 * stor-layout.c: Include attribs.h.
5177 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5178 explicit lookup of "aligned" attribute.
5180 2017-11-07 Andrew Waterman <andrew@sifive.com>
5182 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5183 (riscv_expand_block_move): Likewise.
5184 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5186 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5187 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5188 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5189 (riscv_adjust_block_mem): Likewise.
5190 (riscv_block_move_loop): Likewise.
5191 (riscv_expand_block_move): Likewise.
5192 * config/riscv/riscv.md (movmemsi): New pattern.
5194 2017-11-07 Michael Clark <michaeljclark@mac.com>
5196 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5197 (MUSL_DYNAMIC_LINKER): Likewise.
5199 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5201 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5202 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5205 2017-11-07 Richard Biener <rguenther@suse.de>
5207 * match.pd: Fix build.
5209 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5210 Jackson Woodruff <jackson.woodruff@arm.com>
5212 PR tree-optimization/71026
5213 * match.pd: Canonicalize negate in division.
5215 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5218 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5220 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5222 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5223 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5225 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5227 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5228 non-scalar integral types.
5229 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5230 (-(A-B), -(~A)): New transformations.
5232 2017-11-07 Tom de Vries <tom@codesourcery.com>
5234 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5235 semicolon after "do {} while (0)".
5236 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5237 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5238 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5239 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5240 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5242 2017-11-07 Tom de Vries <tom@codesourcery.com>
5244 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5245 after "do {} while (0)".
5246 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5247 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5248 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5249 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5250 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5252 2017-11-07 Tom de Vries <tom@codesourcery.com>
5255 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
5257 (arm_rtx_costs_internal): Add missing semicolon after
5258 HANDLE_NARROW_SHIFT_ARITH call.
5260 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5262 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
5263 disable isel if it was not set explicitly.
5265 2017-11-06 James Bowman <james.bowman@ftdichip.com>
5267 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
5268 (add_type_attribute) likewise.
5270 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
5272 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
5273 of struct ix86_frame.
5274 (ix86_initial_elimination_offset): Likewise.
5275 (ix86_expand_split_stack_prologue): Likewise.
5277 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
5279 * tree-vrp.h (enum value_range_type): Update stale comment.
5281 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5282 Alan Hayward <alan.hayward@arm.com>
5283 David Sherwood <david.sherwood@arm.com>
5285 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
5286 (aarch64_expand_vec_perm_const): Take the number of units too.
5287 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
5288 (aarch64_expand_vec_perm_const): Likewise.
5289 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
5290 (vec_perm<mode>): Update accordingly.
5292 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5293 Alan Hayward <alan.hayward@arm.com>
5294 David Sherwood <david.sherwood@arm.com>
5296 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
5297 Take the number of units too.
5298 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
5299 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
5300 but check for a vector mode before rather than after the call.
5301 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5302 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
5303 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
5304 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
5305 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
5306 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
5307 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
5308 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
5309 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
5310 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
5311 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
5312 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
5313 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
5314 (aarch64_sqdmull2_n<mode>): Update accordingly.
5316 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5317 Alan Hayward <alan.hayward@arm.com>
5318 David Sherwood <david.sherwood@arm.com>
5320 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
5321 the number of units too.
5322 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
5323 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
5324 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
5325 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
5326 (vec_store_lanesxi<mode>): Update accordingly.
5328 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5329 Alan Hayward <alan.hayward@arm.com>
5330 David Sherwood <david.sherwood@arm.com>
5332 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
5333 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
5334 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
5335 of units rather than the mode.
5336 * config/aarch64/iterators.md (nunits): New mode attribute.
5337 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
5338 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
5339 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
5340 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
5341 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5342 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
5343 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
5344 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
5345 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
5346 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
5347 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5348 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
5349 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
5350 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
5351 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
5352 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
5353 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
5354 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
5355 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
5356 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
5357 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
5358 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
5359 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
5360 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
5361 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
5362 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
5363 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
5364 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
5365 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
5366 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
5367 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
5368 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
5369 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
5370 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
5371 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
5372 (aarch64_simd_vec_setv2di): Likewise.
5374 2017-11-06 Carl Love <cel@us.ibm.com>
5376 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
5378 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
5379 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
5381 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
5383 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
5384 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
5385 Add power 8 macro expansions.
5386 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
5387 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
5388 power 8 instructions. (VSX_XXBR): Add iterator.
5390 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
5392 * config/arm/arm.md (predicable_short_it): Change default to "no",
5393 improve documentation, remove uses that are identical to the default.
5394 (enabled_for_depr_it): Rename to enabled_for_short_it.
5395 * gcc/config/arm/arm-fixed.md (predicable_short_it):
5396 Remove default uses.
5397 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
5398 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
5399 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
5400 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
5402 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5405 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
5406 float128 helper macros, which are no longer used after deleting
5407 the old 'q' built-in functions, and moving the round to odd
5408 built-in functions to being special built-in functions.
5409 (BU_FLOAT128_2): Likewise.
5410 (BU_FLOAT128_1_HW): Likewise.
5411 (BU_FLOAT128_2_HW): Likewise.
5412 (BU_FLOAT128_3_HW): Likewise.
5413 (FABSQ): Delete old 'q' built-in functions.
5414 (COPYSIGNQ): Likewise.
5415 (SQRTF128_ODD): Move round to odd built-in functions to be
5416 special built-in functions, so that we can handle
5417 -mabi=ieeelongdouble.
5418 (TRUNCF128_ODD): Likewise.
5419 (ADDF128_ODD): Likewise.
5420 (SUBF128_ODD): Likewise.
5421 (MULF128_ODD): Likewise.
5422 (DIVF128_ODD): Likewise.
5423 (FMAF128_ODD): Likewise.
5424 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
5425 built-in names to 'f128'.
5426 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
5427 old 'q' built-in functions, as the machine independent code for
5428 'f128' built-in functions handles this.
5429 (rs6000_expand_builtin): Add expansion for float128 round to odd
5430 functions, keying off on -mabi=ieeelongdouble of whether to use
5431 the KFmode or TFmode variant.
5432 (rs6000_init_builtins): Initialize the _Float128 round to odd
5434 * doc/extend.texi (PowerPC Built-in Functions): Document the old
5435 _Float128 'q' built-in functions are now mapped into the new
5436 'f128' built-in functions.
5438 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
5440 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
5441 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
5443 2017-11-06 David Malcolm <dmalcolm@redhat.com>
5446 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
5447 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
5448 * toplev.c: Include "ipa-fnsummary.h".
5449 (toplev::finalize): Call ipa_fnsummary_c_finalize.
5451 2017-11-06 Jakub Jelinek <jakub@redhat.com>
5453 PR tree-optimization/82838
5454 * gimple-ssa-store-merging.c
5455 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
5456 on a separate gimple_seq which is then appended to seq.
5458 2017-11-06 Jeff Law <law@redhat.com>
5461 * config/i386/i386.c (PROBE_INTERVAL): Remove.
5462 (get_probe_interval): New functions.
5463 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
5464 (ix86_adjust_stack_and_probe): Likewise.
5465 (output_adjust_stack_and_probe): Likewise.
5466 (ix86_emit_probe_stack_range): Likewise.
5467 (ix86_expand_prologue): Likewise.
5469 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5471 PR tree-optimization/82816
5472 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
5473 if the modes of the two types are the same.
5474 (convert_plusminus_to_widen): Likewise.
5476 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5478 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
5480 (usadv16qi): New define_expand.
5481 (usadv8hi): New define_expand.
5483 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
5486 * ipa-inline-transform.c (update_noncloned_frequencies): Always
5488 (inline_transform): Likewise.
5489 * predict.c (counts_to_freqs): Remove useless conditional.
5490 * profile-count.h (profile_count::apply_scale): Move sanity check.
5491 * tree-inline.c (copy_bb): Always scale.
5492 (copy_cfg_body): Likewise.
5494 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
5497 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
5499 (*cmp_ite0): Add enabled_for_depr_it attribute.
5500 (*cmp_ite1): Likewise.
5502 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
5504 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
5507 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
5509 * tree-vrp.c (vrp_int_const_binop): Return true on success and
5510 return the value by pointer.
5511 (extract_range_from_multiplicative_op_1): Update accordingly.
5512 Return as soon as an operation fails.
5514 2017-11-05 Tom de Vries <tom@codesourcery.com>
5517 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
5518 (DEF_SANITIZER_BUILTIN): ... here.
5519 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
5520 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
5522 2017-11-05 Tom de Vries <tom@codesourcery.com>
5525 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
5527 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
5528 ASM_OUTPUT_BEFORE_CASE_LABEL call.
5529 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
5531 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5532 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5533 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
5535 2017-11-05 Tom de Vries <tom@codesourcery.com>
5538 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
5541 2017-11-04 Michael Clark <michaeljclark@mac.com>
5543 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
5544 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
5546 (*addsi3_extended): Likewise.
5547 (*addsi3_extended2): Likewise.
5548 (<optab>si3): Likewise.
5549 (<optab>di3): Likewise.
5550 (<optab><mode>3): Likewise.
5551 (<*optabe>si3_internal): Likewise.
5552 (zero_extendqi<SUPERQI:mode>2): Likewise.
5553 (*add<mode>hi3): Likewise.
5554 (*xor<mode>hi3): Likewise.
5555 (<optab>di3): Likewise.
5556 (*<optab>si3_extend): Likewise.
5557 (*sge<u>_<X:mode><GPR:mode>): Likewise.
5558 (*slt<u>_<X:mode><GPR:mode>): Likewise.
5559 (*sle<u>_<X:mode><GPR:mode>): Likewise.
5561 2017-11-04 Andrew Waterman <andrew@sifive.com>
5563 * config/riscv/riscv.c (riscv_option_override): Conditionally set
5564 TARGET_STRICT_ALIGN based upon -mtune argument.
5566 2017-11-04 Andrew Waterman <andrew@sifive.com>
5568 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
5570 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
5572 * config/i386/i386.c (choose_basereg): Use optional scratch
5573 register and add assertion.
5574 (x86_emit_outlined_ms2sysv_save): Use scratch register when
5575 needed, and don't allocate stack.
5576 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
5577 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
5578 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
5580 2017-11-03 Jeff Law <law@redhat.com>
5582 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
5583 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
5584 to probe at the start of a noreturn function.
5586 2017-11-03 Jakub Jelinek <jakub@redhat.com>
5588 PR tree-optimization/78821
5589 * gimple-ssa-store-merging.c: Update the file comment.
5590 (MAX_STORE_ALIAS_CHECKS): Define.
5591 (struct store_operand_info): New type.
5592 (store_operand_info::store_operand_info): New constructor.
5593 (struct store_immediate_info): Add rhs_code and ops data members.
5594 (store_immediate_info::store_immediate_info): Add rhscode, op0r
5595 and op1r arguments to the ctor, initialize corresponding data members.
5596 (struct merged_store_group): Add load_align_base and load_align
5598 (merged_store_group::merged_store_group): Initialize them.
5599 (merged_store_group::do_merge): Update them.
5600 (merged_store_group::apply_stores): Pick the constant for
5601 encode_tree_to_bitpos from one of the two operands, or skip
5602 encode_tree_to_bitpos if neither operand is a constant.
5603 (class pass_store_merging): Add process_store method decl. Remove
5604 bool argument from terminate_all_aliasing_chains method decl.
5605 (pass_store_merging::terminate_all_aliasing_chains): Remove
5606 var_offset_p argument and corresponding handling.
5607 (stmts_may_clobber_ref_p): New function.
5608 (compatible_load_p): New function.
5609 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
5610 if there is overlap and rhs_code is not INTEGER_CST. For
5611 non-overlapping stores terminate group if rhs is not mergeable.
5612 (get_alias_type_for_stmts): Change first argument from
5613 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
5614 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
5615 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
5617 (get_location_for_stmts): Change first argument from
5618 auto_vec<gimple *> & to vec<gimple *> &.
5619 (struct split_store): Remove orig_stmts data member, add orig_stores.
5620 (split_store::split_store): Create orig_stores rather than orig_stmts.
5621 (find_constituent_stmts): Renamed to ...
5622 (find_constituent_stores): ... this. Change second argument from
5623 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
5624 to info structures rather than the statements.
5625 (split_group): Rename ALLOW_UNALIGNED argument to
5626 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
5627 it. Adjust find_constituent_stores caller.
5628 (imm_store_chain_info::output_merged_store): Handle rhs_code other
5629 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
5630 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
5631 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
5632 (mem_valid_for_store_merging): New function.
5633 (handled_load): New function.
5634 (pass_store_merging::process_store): New method.
5635 (pass_store_merging::execute): Use process_store method. Adjust
5636 terminate_all_aliasing_chains caller.
5638 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5640 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5641 Return true for more constants, symbols and label references.
5642 (aarch64_valid_floating_const): Remove unused function.
5644 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5647 * config/aarch64/aarch64.c (aarch64_layout_frame):
5648 Undo forcing of LR at bottom of frame.
5650 2017-11-03 Jeff Law <law@redhat.com>
5653 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
5654 for int_registers_saved.
5656 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
5657 extracted from tree-ssa-dom.c.
5658 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
5659 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
5660 (record_equivalences_from_incoming_edge): Add additional argument
5661 to single_pred_edge_ignoring_loop_edges call.
5662 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
5663 (uncprop_dom_walker::before_dom_children): Add additional argument
5664 to single_pred_edge_ignoring_loop_edges call.
5665 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
5666 single_pred_edge_ignoring_loop_edges rather than open coding.
5667 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
5669 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
5671 * match.pd (-(-A)): Rewrite.
5673 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5675 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
5676 (rs6000_emit_int_cmove): New declaration.
5677 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
5678 (rs6000_emit_sISEL): Delete.
5679 (rs6000_emit_int_cmove): Make non-static.
5680 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
5681 instead of rs6000_emit_sISEL.
5683 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
5685 * asan.c (create_cond_insert_point): Maintain profile.
5686 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
5688 * basic-block.h (struct basic_block_def): Remove frequency.
5689 (EDGE_FREQUENCY): Use to_frequency
5690 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
5692 (find_traces): Update to use to_frequency.
5693 (find_traces_1_round): Likewise; use only IPA counts.
5694 (bb_to_key): Likewise.
5695 (connect_traces): Use IPA counts only.
5696 (copy_bb_p): Update to use to_frequency.
5697 (fix_up_crossing_landing_pad): Likewise.
5698 (sanitize_hot_paths): Likewise.
5699 * bt-load.c (basic_block_freq): Likewise.
5700 * cfg.c (init_flow): Set count_max to uninitialized.
5701 (check_bb_profile): Remove frequencies; check counts.
5702 (dump_bb_info): Do not dump frequencies.
5703 (update_bb_profile_for_threading): Update counts only.
5704 (scale_bbs_frequencies_int): Likewise.
5705 (MAX_SAFE_MULTIPLIER): Remove.
5706 (scale_bbs_frequencies_gcov_type): Update counts only.
5707 (scale_bbs_frequencies_profile_count): Update counts only.
5708 (scale_bbs_frequencies): Update counts only.
5709 * cfg.h (struct control_flow_graph): Add count-max.
5710 (update_bb_profile_for_threading): Update prototype.
5711 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
5712 (find_many_sub_basic_blocks): Likewise.
5713 * cfgcleanup.c (try_forward_edges): Likewise.
5714 (try_crossjump_to_edge): Likewise.
5715 * cfgexpand.c (expand_gimple_cond): Likewise.
5716 (expand_gimple_tailcall): Likewise.
5717 (construct_init_block): Likewise.
5718 (construct_exit_block): Likewise.
5719 * cfghooks.c (verify_flow_info): Check consistency of counts.
5720 (dump_bb_for_graph): Do not dump frequencies.
5721 (split_block_1): Do not update frequencies.
5722 (split_edge): Do not update frequencies.
5723 (make_forwarder_block): Do not update frequencies.
5724 (duplicate_block): Do not update frequencies.
5725 (account_profile_record): Do not update frequencies.
5726 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
5727 for global heuristics.
5728 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
5729 (expected_loop_iterations_unbounded): Use counts only.
5730 * cfgloopmanip.c (scale_loop_profile): Simplify.
5731 (create_empty_loop_on_edge): Simplify
5733 (duplicate_loop_to_header_edge): Simplify
5734 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
5735 (update_br_prob_note): Take care of removing note when profile
5737 (relink_block_chain): Do not dump frequency.
5738 (rtl_account_profile_record): Use to_frequency.
5739 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
5740 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
5741 (cgraph_update_edges_for_call_stmt_node): Likewise.
5742 (cgraph_edge::verify_count_and_frequency): Update.
5743 (cgraph_node::verify_node): Temporarily disable frequency verification.
5744 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
5745 to_cgraph_frequency.
5746 (cgraph_edge::rebuild_edges): Convert to ipa counts.
5747 * cgraphunit.c (init_lowered_empty_function): Do not initialize
5749 (cgraph_node::expand_thunk): Update profile.
5750 * except.c (dw2_build_landing_pads): Do not update frequency.
5751 * final.c (compute_alignments): Use to_frequency.
5752 (dump_basic_block_info): Do not dump frequency.
5753 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
5754 (dump_gimple_bb_header): Do not dump frequency.
5755 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
5757 * gimple-streamer-in.c (input_bb): Do not stream frequency.
5758 * gimple-streamer-out.c (output_bb): Do not stream frequency.
5759 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
5760 (init_before_recovery): Do not update frequency.
5761 (sched_create_recovery_edges): Do not update frequency.
5762 * hsa-gen.c (convert_switch_statements): Do not update frequency.
5763 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
5764 (ipa_cp_c_finalize): Set max_count to uninitialized.
5765 * ipa-fnsummary.c (get_minimal_bb): Use counts.
5766 (param_change_prob): Use counts.
5767 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
5769 * ipa-split.c (consider_split): Use to_frequency.
5770 (split_function): Use to_frequency.
5771 * ira-build.c (loop_compare_func): Likewise.
5772 (mark_loops_for_removal): Likewise.
5773 (mark_all_loops_for_removal): Likewise.
5774 * loop-doloop.c (doloop_modify): Do not update frequency.
5775 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
5777 * lto-streamer-in.c (input_function): Update count_max.
5778 * omp-expand.c (expand_omp_taskreg): Update count_max.
5779 * omp-simd-clone.c (simd_clone_adjust): Update profile.
5780 * predict.c (maybe_hot_frequency_p): Use to_frequency.
5781 (maybe_hot_count_p): Use ipa counts only.
5782 (maybe_hot_bb_p): Simplify.
5783 (maybe_hot_edge_p): Simplify.
5784 (probably_never_executed): Do not take frequency argument.
5785 (probably_never_executed_bb_p): Do not pass frequency.
5786 (probably_never_executed_edge_p): Likewise.
5787 (combine_predictions_for_bb): Check that profile is nonzero.
5788 (propagate_freq): Do not set frequency.
5789 (drop_profile): Simplify.
5790 (counts_to_freqs): Simplify.
5791 (expensive_function_p): Use to_frequency.
5792 (propagate_unlikely_bbs_forward): Simplify.
5793 (determine_unlikely_bbs): Simplify.
5794 (estimate_bb_frequencies): Add hack to silence graphite issues.
5795 (compute_function_frequency): Use ipa counts.
5796 (pass_profile::execute): Update.
5797 (rebuild_frequencies): Use counts only.
5798 (force_edge_cold): Use counts only.
5799 * profile-count.c (profile_count::dump): Dump new count types.
5800 (profile_count::differs_from_p): Check compatiblity.
5801 (profile_count::to_frequency): New function.
5802 (profile_count::to_cgraph_frequency): New function.
5803 * profile-count.h (struct function): Declare.
5804 (enum profile_quality): Add profile_guessed_local and
5805 profile_guessed_global0.
5806 (class profile_proability): Decrease number of bits to 29;
5807 update from_reg_br_prob_note and to_reg_br_prob_note.
5808 (class profile_count: Update comment; decrease number of bits
5809 to 61. Check compatibility.
5810 (profile_count::compatible_p): New private member function.
5811 (profile_count::ipa_p): New member function.
5812 (profile_count::operator<): Handle global zero correctly.
5813 (profile_count::operator>): Handle global zero correctly.
5814 (profile_count::operator<=): Handle global zero correctly.
5815 (profile_count::operator>=): Handle global zero correctly.
5816 (profile_count::nonzero_p): New member function.
5817 (profile_count::force_nonzero): New member function.
5818 (profile_count::max): New member function.
5819 (profile_count::apply_scale): Handle IPA scalling.
5820 (profile_count::guessed_local): New member function.
5821 (profile_count::global0): New member function.
5822 (profile_count::ipa): New member function.
5823 (profile_count::to_frequency): Declare.
5824 (profile_count::to_cgraph_frequency): Declare.
5825 * profile.c (OVERLAP_BASE): Delete.
5826 (compute_frequency_overlap): Delete.
5827 (compute_branch_probabilities): Do not use compute_frequency_overlap.
5828 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
5829 * sched-ebb.c (rank): Use counts only.
5830 * shrink-wrap.c (handle_simple_exit): Use counts only.
5831 (try_shrink_wrapping): Use counts only.
5832 (place_prologue_for_one_component): Use counts only.
5833 * tracer.c (find_best_predecessor): Use to_frequency.
5834 (find_trace): Use to_frequency.
5835 (tail_duplicate): Use to_frequency.
5836 * trans-mem.c (expand_transaction): Do not update frequency.
5837 * tree-call-cdce.c: Do not update frequency.
5838 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
5839 (gimple_merge_blocks): Likewise.
5840 (gimple_split_edge): Likewise.
5841 (gimple_duplicate_sese_region): Likewise.
5842 (gimple_duplicate_sese_tail): Likewise.
5843 (move_sese_region_to_fn): Likewise.
5844 (gimple_account_profile_record): Likewise.
5845 (insert_cond_bb): Likewise.
5846 * tree-complex.c (expand_complex_div_wide): Likewise.
5847 * tree-eh.c (lower_resx): Update profile.
5848 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
5850 (initialize_cfun): Do not initialize frequencies
5851 (freqs_to_counts): Delete.
5852 (copy_cfg_body): Ignore count parameter.
5853 (copy_body): Update.
5854 (expand_call_inline): Update count_max.
5855 (optimize_inline_calls): Update count_max.
5856 (tree_function_versioning): Update count_max.
5857 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
5858 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
5860 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
5861 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
5862 (try_peel_loop): Likewise.
5863 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
5865 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
5866 (tree_transform_and_unroll_loop): Do not use frequencies
5867 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
5868 Use reliable prediction only.
5869 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
5870 * tree-ssa-sink.c (select_best_block): Use to_frequency.
5871 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
5872 probability scaling.
5873 * tree-ssa-threadupdate.c (create_block_for_threading): Do
5874 not update frequency
5875 (any_remaining_duplicated_blocks): Likewise.
5876 (update_profile): Likewise.
5877 (estimated_freqs_path): Delete.
5878 (freqs_to_counts_path): Delete.
5879 (clear_counts_path): Delete.
5880 (ssa_fix_duplicate_block_edges): Likewise.
5881 (duplicate_thread_path): Likewise.
5882 * tree-switch-conversion.c (gen_inbound_check): Use counts.
5883 * tree-tailcall.c (decrease_profile): Do not update frequency.
5884 (eliminate_tail_call): Likewise.
5885 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5886 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5887 (optimize_mask_stores): Likewise.
5888 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5889 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5890 (ubsan_expand_ptr_ifn): Update profile.
5891 * value-prof.c (gimple_ic): Simplify.
5892 * value-prof.h (gimple_ic): Update prototype.
5893 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5894 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5896 (want_inline_self_recursive_call_p): Likewise.
5897 (resolve_noninline_speculation): Only cummulate defined counts.
5898 (inline_small_functions): Use nonzero_p.
5899 (ipa_inline): Do not access freed node.
5901 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5903 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5904 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5906 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5908 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5909 non-legitimate address.
5911 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5913 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5914 (*lt0_<mode>di, *lt0_<mode>si): New.
5916 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5918 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5919 TARGET_PAIRED_FLOAT.
5921 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5922 Jim Wilson <jim.wilson@linaro.org>
5924 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5925 * config/aarch64/aarch64-tune.md: Regenerated.
5926 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5927 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5929 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5931 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5932 (arc_expand_prologue): Restore blink for millicode.
5933 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5935 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5938 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5939 gen_vec_duplicate after forcing the scalar into a register.
5941 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5943 * combine (try_combine): Print the insns input to try_combine to the
5946 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5949 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5950 Remove second argument from aarch64_process_target_attr call.
5951 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5953 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5955 (aarch64_handle_attr_arch): Remove second argument.
5956 (aarch64_handle_attr_cpu): Ditto.
5957 (aarch64_handle_attr_tune): Ditto.
5958 (aarch64_handle_attr_isa_flags): Ditto.
5959 (aarch64_process_one_target_attr): Ditto.
5960 (aarch64_process_target_attr): Ditto.
5961 (aarch64_option_valid_attribute_p): Remove second argument.
5962 on aarch64_process_target_attr call.
5964 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5966 * diagnostic.c: Include "selftest-diagnostic.h".
5967 (selftest::assert_location_text): New function.
5968 (selftest::test_diagnostic_get_location_text): New function.
5969 (selftest::diagnostic_c_tests): Call it.
5971 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5973 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5974 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5975 (class selftest::test_diagnostic_context): Move to...
5976 * selftest-diagnostic.c: New file.
5977 * selftest-diagnostic.h: New file.
5979 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5981 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5982 offset range for FT32B.
5983 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5984 * config/ft32/ft32.md: Add TARGET_NOPM.
5985 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5986 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5988 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5990 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5992 2017-11-02 Jeff Law <law@redhat.com>
5994 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5995 virtual keyword on FINAL OVERRIDE members.
5997 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5998 virtuals together. Add virtual destructor.
5999 (substitute_and_fold_engine): Similarly.
6001 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6003 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6005 2017-11-02 Richard Biener <rguenther@suse.de>
6007 PR tree-optimization/82795
6008 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6010 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6012 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6014 (gcc_GAS_CHECK_FEATURE): Remove.
6015 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6016 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6017 * configure: Regenerate.
6019 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6021 * config/arc/arc.c (hwloop_optimize): Account for empty
6024 2017-11-02 Richard Biener <rguenther@suse.de>
6027 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6028 Truncate ARRAY_REF index and element size.
6030 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6032 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6034 2017-11-01 Jeff Law <law@redhat.com>
6036 * tree-ssa-ccp.c (ccp_folder): New class derived from
6037 substitute_and_fold_engine.
6038 (ccp_folder::get_value): New member function.
6039 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6040 (ccp_fold_stmt): Remove prototype.
6041 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6042 * tree-ssa-copy.c (copy_folder): New class derived from
6043 substitute_and_fold_engine.
6044 (copy_folder::get_value): Renamed from get_value.
6045 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6046 * tree-vrp.c (vrp_folder): New class derived from
6047 substitute_and_fold_engine.
6048 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6049 (vrp_folder::get_value): New member function.
6050 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6051 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6052 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6053 provide a class interface to folder/substitute routines.
6054 (ssa_prop_fold_stmt_fn): Remove typedef.
6055 (ssa_prop_get_value_fn): Likewise.
6056 (subsitute_and_fold): Remove prototype.
6057 (replace_uses_in): Likewise.
6058 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6059 Renamed from replace_uses_in. Call the virtual member function
6060 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6061 (substitute_and_fold_dom_walker): Remove initialization of
6062 data member entries for calbacks. Add substitute_and_fold_engine
6063 member and initialize it.
6064 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6065 member functions for get_value, replace_phi_args_in c
6066 replace_uses_in, and fold_stmt calls.
6067 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6068 substitute_and_fold. Remove assert. Update ctor call.
6070 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6071 (ssa_prop_visit_phi_fn): Likewise.
6072 (class ssa_propagation_engine): New class to provide an interface
6074 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6076 (ssa_prop_visit_phi): Likewise.
6077 (ssa_propagation_engine::simulate_stmt): Moved into class.
6078 Call visit_phi/visit_stmt from the class rather than via
6079 file scoped static variables.
6080 (ssa_propagation_engine::simulate_block): Moved into class.
6081 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6082 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6083 set file scoped statics for the visit_stmt/visit_phi callbacks.
6084 * tree-complex.c (complex_propagate): New class derived from
6085 ssa_propagation_engine.
6086 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6087 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6088 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6090 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6091 ssa_propagation_engine.
6092 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6093 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6094 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6095 * tree-ssa-copy.c (copy_prop): New class derived from
6096 ssa_propagation_engine.
6097 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6098 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6099 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6100 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6101 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6102 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6103 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6105 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6107 PR rtl-optimization/82778
6108 PR rtl-optimization/82597
6109 * compare-elim.c (struct comparison): Add in_a_setter field.
6110 (find_comparison_dom_walker::before_dom_children): Remove killed
6111 bitmap and df_simulate_find_defs call, instead walk the defs.
6112 Compute last_setter and initialize in_a_setter. Merge definitions
6113 with first initialization for a few variables.
6114 (try_validate_parallel): Use insn_invalid_p instead of
6115 recog_memoized. Return insn rather than just the pattern.
6116 (try_merge_compare): Fix up comment. Don't uselessly test if
6117 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6119 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6122 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6123 Alan Hayward <alan.hayward@arm.com>
6124 David Sherwood <david.sherwood@arm.com>
6126 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6127 aarch64_hard_regno_nregs to get the number of registers
6130 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6131 Alan Hayward <alan.hayward@arm.com>
6132 David Sherwood <david.sherwood@arm.com>
6134 * config/aarch64/constraints.md (Upl): Rename to...
6136 * config/aarch64/aarch64.md
6137 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
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 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6145 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
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 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6153 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6154 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6156 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6157 of a single element.
6158 * config/aarch64/iterators.md: Add a comment above the permute
6159 unspecs to say that they are generated directly by
6160 aarch64_expand_vec_perm_const.
6161 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6163 2017-11-01 Nathan Sidwell <nathan@acm.org>
6165 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6167 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6169 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6170 and medany code models, and describe what they do.
6172 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6174 Revert accidental duplicate:
6176 * combine.c (can_change_dest_mode): Reject changes in
6177 REGMODE_NATURAL_SIZE.
6179 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6181 PR rtl-optimization/64682
6182 PR rtl-optimization/69567
6183 PR rtl-optimization/69737
6184 PR rtl-optimization/82683
6185 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6186 register mentioned in the note, drop the note, unless it came from I3,
6187 in which case it should go to I3 again.
6189 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6191 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6192 and return false if not.
6193 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6195 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6197 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6198 range is known to be empty.
6200 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6201 Alan Hayward <alan.hayward@arm.com>
6202 David Sherwood <david.sherwood@arm.com>
6204 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6205 and CONST_VECTOR_NUNITS instead of computing the number of units from
6206 the byte sizes of the vector and element.
6207 (simplify_binary_operation_1): Likewise.
6208 (simplify_const_binary_operation): Likewise.
6209 (simplify_ternary_operation): Likewise.
6211 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6212 Alan Hayward <alan.hayward@arm.com>
6213 David Sherwood <david.sherwood@arm.com>
6215 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6216 (int_mem_offset): ...this new function.
6217 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6218 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6219 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
6222 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6223 Alan Hayward <alan.hayward@arm.com>
6224 David Sherwood <david.sherwood@arm.com>
6226 * lower-subreg.c (interesting_mode_p): New function.
6227 (compute_costs, find_decomposable_subregs, decompose_register)
6228 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6229 (resolve_clobber, dump_choices): Use it.
6231 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6232 Alan Hayward <alan.hayward@arm.com>
6233 David Sherwood <david.sherwood@arm.com>
6235 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6237 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6238 Alan Hayward <alan.hayward@arm.com>
6239 David Sherwood <david.sherwood@arm.com>
6241 * alias.c (find_base_value, find_base_term): Only process integer
6242 truncations. Check the precision rather than the size.
6244 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6245 Alan Hayward <alan.hayward@arm.com>
6246 David Sherwood <david.sherwood@arm.com>
6248 * machmode.h (is_narrower_int_mode): New function
6249 * optabs.c (expand_float, expand_fix): Use it.
6250 * dwarf2out.c (rotate_loc_descriptor): Likewise.
6252 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6253 Alan Hayward <alan.hayward@arm.com>
6254 David Sherwood <david.sherwood@arm.com>
6256 * rtl.h (narrower_subreg_mode): New function.
6257 * ira-color.c (update_costs_from_allocno): Use it.
6259 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6260 Alan Hayward <alan.hayward@arm.com>
6261 David Sherwood <david.sherwood@arm.com>
6263 * optabs-query.h (convert_optab_p): New function, split out from...
6264 (convert_optab_handler): ...here.
6265 (widening_optab_handler): Delete.
6266 (find_widening_optab_handler): Remove permit_non_widening parameter.
6267 (find_widening_optab_handler_and_mode): Likewise. Provide an
6268 override that operates on mode class wrappers.
6269 * optabs-query.c (widening_optab_handler): Delete.
6270 (find_widening_optab_handler_and_mode): Remove permit_non_widening
6271 parameter. Assert that the two modes are the same class and that
6272 the "from" mode is narrower than the "to" mode. Use
6273 convert_optab_handler instead of widening_optab_handler.
6274 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
6275 instead of widening_optab_handler.
6276 * expr.c (expand_expr_real_2): Update calls to
6277 find_widening_optab_handler.
6278 * optabs.c (expand_widen_pattern_expr): Likewise.
6279 (expand_binop_directly): Take the insn_code as a parameter.
6280 (expand_binop): Only call find_widening_optab_handler for
6281 conversion optabs; use optab_handler otherwise. Update calls
6282 to find_widening_optab_handler and expand_binop_directly.
6283 Use convert_optab_handler instead of widening_optab_handler.
6284 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
6285 find_widening_optab_handler and use scalar_mode rather than
6287 (convert_plusminus_to_widen): Likewise.
6289 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6290 Alan Hayward <alan.hayward@arm.com>
6291 David Sherwood <david.sherwood@arm.com>
6293 * machmode.h (fixed_size_mode): New class.
6294 * rtl.h (get_pool_mode): Return fixed_size_mode.
6295 * gengtype.c (main): Add fixed_size_mode.
6296 * target.def (get_raw_result_mode): Return a fixed_size_mode.
6297 (get_raw_arg_mode): Likewise.
6298 * doc/tm.texi: Regenerate.
6299 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
6300 * targhooks.c (default_get_reg_raw_mode): Likewise.
6301 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
6302 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
6303 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
6304 (msp430_get_raw_result_mode): Likewise.
6305 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
6306 * dbxout.c (dbxout_parms): Require fixed-size modes.
6307 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
6308 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6309 * omp-low.c (lower_oacc_reductions): Likewise.
6310 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
6311 (simplify_subreg): Update accordingly.
6312 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
6313 (force_const_mem): Update accordingly. Return NULL_RTX for modes
6314 that aren't fixed-size.
6315 (get_pool_mode): Return a fixed_size_mode.
6316 (output_constant_pool_2): Take a fixed_size_mode.
6318 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6319 Alan Hayward <alan.hayward@arm.com>
6320 David Sherwood <david.sherwood@arm.com>
6322 * doc/rtl.texi (vec_series): Document.
6323 (const): Say that the operand can be a vec_series.
6324 * rtl.def (VEC_SERIES): New rtx code.
6325 * rtl.h (const_vec_series_p_1): Declare.
6326 (const_vec_series_p): New function.
6327 * emit-rtl.h (gen_const_vec_series): Declare.
6328 (gen_vec_series): Likewise.
6329 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
6330 (gen_vec_series): Likewise.
6331 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
6332 * simplify-rtx.c (simplify_unary_operation): Handle negations
6334 (simplify_binary_operation_series): New function.
6335 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
6336 (test_vector_ops_series): New function.
6337 (test_vector_ops): Call it.
6338 * config/powerpcspe/altivec.md (altivec_lvsl): Use
6339 gen_const_vec_series.
6340 (altivec_lvsr): Likewise.
6341 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
6343 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6344 Alan Hayward <alan.hayward@arm.com>
6345 David Sherwood <david.sherwood@arm.com>
6347 * doc/rtl.texi (const): Update description of address constants.
6348 Say that vector constants are allowed too.
6349 * common.md (E, F): Use CONSTANT_P instead of checking for
6351 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
6352 checking for CONST_VECTOR.
6353 * expmed.c (make_tree): Use build_vector_from_val for a CONST
6355 * expr.c (expand_expr_real_2): Check for vector modes instead
6356 of checking for CONST_VECTOR.
6357 * rtl.h (const_vec_p): New function.
6358 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
6359 (unwrap_const_vec_duplicate): Handle them here too.
6361 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6362 David Malcolm <dmalcolm@redhat.com>
6363 Alan Hayward <alan.hayward@arm.com>
6364 David Sherwood <david.sherwood@arm.com>
6366 * rtl.h (vec_duplicate_p): New function.
6367 * selftest-rtl.c (assert_rtx_eq_at): New function.
6368 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
6369 (assert_rtx_eq_at): Declare.
6370 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
6371 * selftest-run-tests.c (selftest::run_tests): Call it.
6372 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
6373 (simplify_unary_operation_1): Recursively handle vector duplicates.
6374 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
6376 (simplify_subreg): Handle subregs of vector duplicates.
6377 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
6378 (selftest::simplify_rtx_c_tests): New functions.
6380 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6381 Alan Hayward <alan.hayward@arm.com>
6382 David Sherwood <david.sherwood@arm.com>
6384 * emit-rtl.h (gen_const_vec_duplicate): Declare.
6385 (gen_vec_duplicate): Likewise.
6386 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
6388 (gen_const_vector): ...here.
6389 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
6390 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
6391 whose elements are all equal.
6392 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
6393 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
6394 (simplify_relational_operation): Likewise.
6395 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6397 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
6398 (aarch64_expand_vector_init): Likewise.
6399 * config/arm/arm.c (neon_vdup_constant): Likewise.
6400 (neon_expand_vector_init): Likewise.
6401 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
6402 (arm_block_set_unaligned_vect): Likewise.
6403 (arm_block_set_aligned_vect): Likewise.
6404 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
6405 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
6406 (expand_vec_perm_even_odd_pack): Likewise.
6407 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
6408 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
6409 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
6410 gen_const_vec_duplicate.
6411 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
6412 * config/mips/mips.c (mips_gen_const_int_vector): Use
6413 gen_const_vec_duplicate.
6414 (mips_expand_vector_init): Use CONST0_RTX.
6415 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
6416 (define_split): Use gen_const_vec_duplicate.
6417 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
6418 (define_split): Use gen_const_vec_duplicate.
6419 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
6420 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
6421 * config/spu/spu.c (spu_const): Likewise.
6423 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6424 Alan Hayward <alan.hayward@arm.com>
6425 David Sherwood <david.sherwood@arm.com>
6427 * combine.c (can_change_dest_mode): Reject changes in
6428 REGMODE_NATURAL_SIZE.
6430 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
6432 * configure.ac (--enable-libssp): New.
6433 (gcc_cv_libc_provides_ssp): Check for explicit setting before
6434 trying to determine target-specific default. Adjust indentation.
6435 * configure: Regenerated.
6436 * doc/install.texi (Configuration): Expand --disable-libssp
6439 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
6441 config/i386/i386.c (ix86_expand_epilogue): Correct stack
6444 2017-10-31 Martin Jambor <mjambor@suse.cz>
6447 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
6449 2017-10-31 David Malcolm <dmalcolm@redhat.com>
6451 * auto-profile.c (autofdo_source_profile::read): Use
6452 UNKNOWN_LOCATION rather than 0.
6453 * diagnostic-core.h (warning_at_rich_loc): Rename to...
6454 (warning_at): ...this overload.
6455 (warning_at_rich_loc_n): Rename to...
6456 (warning_n): ...this overload.
6457 (error_at_rich_loc): Rename to...
6458 (error_at): ...this overload.
6459 (pedwarn_at_rich_loc): Rename to...
6460 (pedwarn): ...this overload.
6461 (permerror_at_rich_loc): Rename to...
6462 (permerror): ...this overload.
6463 (inform_at_rich_loc): Rename to...
6464 (inform): ...this overload.
6465 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
6466 (diagnostic_n_impl_richloc): Rename to...
6467 (diagnostic_n_impl): ...this rich_location *-based decl.
6468 (inform_at_rich_loc): Rename to...
6469 (inform): ...this, and add an assertion.
6470 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
6471 (warning_at_rich_loc): Rename to...
6472 (warning_at): ...this, and add an assertion.
6473 (warning_at_rich_loc_n): Rename to...
6474 (warning_n): ...this, and add an assertion.
6475 (warning_n): Update location_t-based implementation for removal of
6476 location_t-based diagnostic_n_impl.
6477 (pedwarn_at_rich_loc): Rename to...
6478 (pedwarn): ...this, and add an assertion.
6479 (permerror_at_rich_loc): Rename to...
6480 (permerror): ...this, and add an assertion.
6481 (error_n): Update for removal of location_t-based diagnostic_n_impl.
6482 (error_at_rich_loc): Rename to...
6483 (error_at): ...this, and add an assertion.
6484 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
6485 (driver::do_spec_on_infiles): Likewise.
6486 * substring-locations.c (format_warning_va): Update for renaming
6487 of inform_at_rich_loc.
6489 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6491 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
6492 _Float<N>X built-in functions so that the variant without the
6493 "__builtin_" prefix is only enabled for the GNU C and Objective C
6494 languages when they are in non-strict ANSI/ISO mode.
6495 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
6496 * target.def (floatn_builtin_p): Add a target hook to control
6497 whether _Float<N> and _Float<N>X built-in functions without the
6498 "__builtin_" prefix are enabled, and return true for C and
6499 Objective C in the default hook. Include langhooks.h in
6501 * targhooks.h (default_floatn_builtin_p): Likewise.
6502 * targhooks.c (default_floatn_builtin_p): Likewise.
6503 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
6504 floatn_builtin_p target hook.
6505 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
6507 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
6508 Eric Botcazou <ebotcazou@adacore.com>
6510 PR rtl-optimization/81803
6511 * lra-constraints.c (curr_insn_transform): Also reload the whole
6512 register for a strict subreg no wider than a word if this is for
6513 a WORD_REGISTER_OPERATIONS target.
6515 2017-10-31 Jason Merrill <jason@redhat.com>
6517 * gdbinit.in: Skip over inlines from timevar.h.
6519 2017-10-31 Martin Liska <mliska@suse.cz>
6521 * doc/gcov.texi: Document new option.
6522 * gcov.c (print_usage): Likewise print it.
6523 (process_args): Support the argument.
6524 (format_count): New function.
6525 (format_gcov): Use the function.
6527 2017-10-31 Martin Liska <mliska@suse.cz>
6529 * gcov.c (struct name_map): do not use typedef.
6530 Define operator== and operator<.
6531 (name_search): Remove.
6532 (name_sort): Remove.
6533 (main): Do not allocate names.
6534 (process_file): Add vertical space.
6535 (generate_results): Use std::find.
6536 (release_structures): Do not release memory.
6537 (find_source): Use std::find.
6539 2017-10-31 Martin Liska <mliska@suse.cz>
6541 * gcov.c (struct line_info): Remove it's typedef.
6542 (line_info::line_info): Add proper ctor.
6543 (line_info::has_block): Do not use a typedef.
6544 (struct source_info): Do not use typedef.
6545 (circuit): Likewise.
6546 (get_cycles_count): Likewise.
6547 (output_intermediate_file): Iterate via vector iterator.
6548 (add_line_counts): Use std::vector methods.
6549 (accumulate_line_counts): Likewise.
6550 (output_lines): Likewise.
6552 2017-10-31 Martin Liska <mliska@suse.cz>
6554 * gcov.c (struct source_info): Remove typedef.
6555 (source_info::source_info): Add proper ctor.
6556 (accumulate_line_counts): Use struct, not it's typedef.
6557 (output_gcov_file): Likewise.
6558 (output_lines): Likewise.
6559 (main): Do not allocate an array.
6560 (output_intermediate_file): Use size of vector container.
6561 (process_file): Resize the vector.
6562 (generate_results): Do not preallocate, use newly added vector
6564 (release_structures): Do not release sources.
6565 (find_source): Use vector methods.
6566 (add_line_counts): Do not use typedef.
6568 2017-10-31 Martin Liska <mliska@suse.cz>
6570 * doc/gcov.texi: Document that.
6571 * gcov.c (add_line_counts): Mark lines with a non-executed
6573 (output_line_beginning): Handle such lines.
6574 (output_lines): Pass new argument.
6575 (output_intermediate_file): Print it in intermediate format.
6577 2017-10-31 Martin Liska <mliska@suse.cz>
6579 * color-macros.h: New file.
6580 * diagnostic-color.c: Factor out color related to macros to
6582 * doc/gcov.texi: Document -k option.
6583 * gcov.c (INCLUDE_STRING): Include string.h.
6584 (print_usage): Add -k option.
6585 (process_args): Parse it.
6586 (pad_count_string): New function.
6587 (output_line_beginning): Likewise.
6588 (DEFAULT_LINE_START): New macro.
6589 (output_lines): Support color output.
6591 2017-10-31 Martin Liska <mliska@suse.cz>
6593 PR gcov-profile/82633
6594 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
6595 their interaction with GCOV infrastructure.
6596 * configure.ac: Add -fkeep-{inline,static}-functions to
6598 * configure: Regenerate.
6600 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
6603 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
6605 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
6608 * config/rs6000/rs6000.md (allocate_stack): Force update interval
6609 into a register if it does not fit into an immediate offset field.
6611 2017-10-31 Olivier Hainque <hainque@adacore.com>
6613 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
6615 2017-10-31 Julia Koval <julia.koval@intel.com>
6617 * config.gcc: Add gfniintrin.h.
6618 * config/i386/gfniintrin.h: New.
6619 * config/i386/i386-builtin-types.def
6620 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6621 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6622 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6623 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6624 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6625 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
6626 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
6627 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
6628 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
6629 V64QI_FTYPE_V64QI_V64QI_INT): New types.
6630 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
6631 * config/i386/immintrin.h: Include gfniintrin.h.
6632 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
6634 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
6636 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
6638 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6640 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
6641 (arm_ashldi3_1bit): Remove pattern.
6642 (ashrdi3): Remove shift by 1 expansion.
6643 (arm_ashrdi3_1bit): Remove pattern.
6644 (lshrdi3): Remove shift by 1 expansion.
6645 (arm_lshrdi3_1bit): Remove pattern.
6646 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
6647 cost of ashldi3 by 1.
6648 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
6649 (<shift>di3_neon): Likewise.
6651 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6653 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
6654 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
6655 and (*aarch64_simd_mov<VQ:mode>).
6656 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
6657 pattern alternative.
6658 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
6659 attributes to match pattern alternative.
6661 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
6663 * config.gcc (powerpc*-*-*): Add emmintrin.h.
6664 * config/rs6000/emmintrin.h: New file.
6665 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
6667 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
6669 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
6670 * (movdi_vfp_cortexa8): Remove pattern.
6672 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6674 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
6676 (Specific, alpha*-dec-osf5.1): Remove.
6677 (Specific, mips-sgi-irix5): Remove.
6678 (Specific, mips-sgi-irix6): Remove.
6680 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6683 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
6684 arguments to clear_bit_region_be.
6686 2017-10-30 Jim Wilson <wilson@tuliptree.org>
6688 * gimplify.c: Include memmodel.h.
6690 2017-10-30 Martin Jambor <mjambor@suse.cz>
6692 * omp-grid.c (grid_attempt_target_gridification): Also insert a
6693 condition whether loop should be executed at all.
6695 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
6697 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6698 gimple folding of vec_madd() intrinsics.
6699 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
6700 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
6701 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
6703 2017-10-30 Richard Biener <rguenther@suse.de>
6705 PR tree-optimization/82762
6707 2017-10-23 Richard Biener <rguenther@suse.de>
6709 PR tree-optimization/82129
6711 2017-08-01 Richard Biener <rguenther@suse.de>
6713 PR tree-optimization/81181
6714 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6715 (compute_antic): ... end of iteration here.
6717 2017-10-30 Joseph Myers <joseph@codesourcery.com>
6719 * doc/invoke.texi (C Dialect Options): Document -std=c17,
6720 -std=iso9899:2017 and -std=gnu17.
6721 * doc/standards.texi (C Language): Document C17 support.
6722 * doc/cpp.texi (Overview): Mention -std=c17.
6723 (Standard Predefined Macros): Document C11 and C17 values of
6724 __STDC_VERSION__. Do not refer to C99 support as incomplete.
6725 * doc/extend.texi (Inline): Do not list individual options for
6726 standards newer than C99.
6727 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
6729 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
6732 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6734 * asan.c (asan_finish_file): Align asan globals array by shadow
6737 2017-10-30 Jakub Jelinek <jakub@redhat.com>
6740 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
6741 (struct store_immediate_info): Add bitregion_start and bitregion_end
6743 (store_immediate_info::store_immediate_info): Add brs and bre
6744 arguments and initialize bitregion_{start,end} from those.
6745 (struct merged_store_group): Add bitregion_start, bitregion_end,
6746 align_base and mask fields. Drop unnecessary struct keyword from
6747 struct store_immediate_info. Add do_merge method.
6748 (clear_bit_region_be): Use memset instead of loop storing zeros.
6749 (merged_store_group::do_merge): New method.
6750 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
6751 stores as long as the surrounding bitregions have no gaps.
6752 (merged_store_group::merge_overlapping): Use do_merge.
6753 (merged_store_group::apply_stores): Test that bitregion_{start,end}
6754 is byte aligned, rather than requiring that start and width are
6755 byte aligned. Drop unnecessary struct keyword from
6756 struct store_immediate_info. Allocate and populate also mask array.
6757 Make start of the arrays relative to bitregion_start rather than
6758 start and size them according to bitregion_{end,start} difference.
6759 (struct imm_store_chain_info): Drop unnecessary struct keyword from
6760 struct store_immediate_info.
6761 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
6762 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
6763 struct keyword from struct store_immediate_info.
6764 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
6765 between stores as long as the surrounding bitregions have no gaps.
6767 (struct split_store): Add orig non-static data member.
6768 (split_store::split_store): Initialize orig to false.
6769 (find_constituent_stmts): Return store_immediate_info *, non-NULL
6770 if there is exactly a single original stmt. Change stmts argument
6771 to pointer from reference, if NULL, don't push anything to it. Add
6772 first argument, use it to optimize skipping over orig stmts that
6773 are known to be before bitpos already. Simplify.
6774 (split_group): Return unsigned int count how many stores are or
6775 would be needed rather than a bool. Add allow_unaligned argument.
6776 Change split_stores argument from reference to pointer, if NULL,
6777 only do a dry run computing how many stores would be produced.
6778 Rewritten algorithm to use both alignment and misalign if
6779 !allow_unaligned and handle bitfield stores with gaps.
6780 (imm_store_chain_info::output_merged_store): Set start_byte_pos
6781 from bitregion_start instead of start. Compute allow_unaligned
6782 here, if true, do 2 split_group dry runs to compute which one
6783 produces fewer stores and prefer aligned if equal. Punt if
6784 new count is bigger or equal than original before emitting any
6785 statements, rather than during that. Remove no longer needed
6786 new_ssa_names tracking. Replace num_stmts with
6787 split_stores.length (). Use 32-bit stack allocated entries
6788 in split_stores auto_vec. Try to reuse original store lhs/rhs1
6789 if possible. Handle bitfields with gaps.
6790 (pass_store_merging::execute): Ignore bitsize == 0 stores.
6791 Compute bitregion_{start,end} for the stores and construct
6792 store_immediate_info with that. Formatting fixes.
6794 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
6797 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
6798 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
6800 2017-10-29 Jim Wilson <wilson@tuliptree.org>
6802 * gimplify.c: Include tm_p.h.
6804 * common.opt (gcoff): Re-add as ignored option.
6805 (gcoff1, gcoff2, gcoff3): Likewise.
6807 * Makefile.in (OBJS): Delete sdbout.o.
6808 (GTFILES): Delete $(srcdir)/sdbout.c.
6809 * debug.h: Delete sdb_debug_hooks.
6810 * final.c: Delete sdbout.h include.
6811 (final_scan_insn): Delete SDB_DEBUG check.
6812 (rest_of_clean_state): Likewise.
6813 * output.h: Delete sdb_begin_function_line.
6816 * toplev.c: Delete sdbout.h include.
6817 (process_options): Delete SDB_DEBUG check.
6818 * tree-core.h (tree_type_common): Delete pointer field of
6820 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
6821 TYPE_SYMTAB_POINTER.
6822 * tree.h (TYPE_SYMTAB_POINTER): Delete.
6823 (TYPE_SYMTAB_IS_POINTER): Delete.
6824 (TYPE_SYMTAB_IS_DIE): Renumber.
6825 * xcoffout.c: Refer to former sdbout.c file.
6826 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
6828 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
6829 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
6830 (Debugging Options): Delete -gcoff.
6831 (-gstabs): Delete SDB reference.
6833 (-gcoff@var{level}): Delete.
6834 * doc/passes.texi (Debugging information output): Delete SDB and
6835 sdbout.c references.
6836 * doc/tm.texi: Regenerate.
6837 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
6838 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
6840 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
6841 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
6842 (SDB_DEBUGGING_INFO): Delete.
6843 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
6844 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
6845 * target.def (output_source_filename): Delete COFF reference.
6847 * common.opt (gcoff): Delete.
6848 (gxcoff+): Update Negative chain.
6849 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
6851 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
6852 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
6853 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
6855 (expand_function_start): Change sdb reference to past tense.
6856 (expand_function_end): Change sdb reference to past tense.
6857 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
6858 * opts.c (debug_type_names): Delete coff entry.
6859 (common_handle_option): Delete OPT_gcoff case.
6860 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
6862 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
6863 * config/cris/cris.h: Delete SDB reference in comment.
6864 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
6865 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
6866 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
6867 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
6869 (ix86_expand_prologue): Likewise.
6870 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
6871 * config/ia64/ia64.h: Likewise.
6872 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
6873 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
6875 * config/mmix/mmix.h: Likewise.
6876 * config/nds32/nds32.c: Likewise.
6877 * config/stormy/storym16.h: Likewise.
6878 * config/visium/visium.h: Likewise.
6879 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
6881 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
6883 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
6884 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
6885 FRAME_POINTER_REGNUM point at high end of local var area.
6887 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6889 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6890 Move comment around. Do not reset best_edge for a copiable
6891 destination if the copy would cause a partition change.
6892 (better_edge_p): Remove redundant check.
6894 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6896 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6898 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6900 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6901 for math functions that have _Float<N> and _Float<N>X variants.
6902 (mathfn_built_in_2): Add support for math functions that have
6903 _Float<N> and _Float<N>X variants.
6904 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6905 (expand_builtin_mathfn_ternary): Add support for fma with
6906 _Float<N> and _Float<N>X variants.
6907 (expand_builtin): Likewise.
6908 (fold_builtin_3): Likewise.
6909 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6910 create math function _Float<N> and _Float<N>X variants as external
6912 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6913 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6914 the __builtin_ prefix and if not strict ansi, without the prefix.
6915 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6916 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6917 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6918 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6919 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6920 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6921 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6922 function signatures for fma _Float<N> and _Float<N>X variants.
6923 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6924 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6925 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6926 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6927 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6928 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6929 * gencfn-macros.c (print_case_cfn): Add support for math functions
6930 that have _Float<N> and _Float<N>X variants.
6931 (print_define_operator_list): Likewise.
6932 (fltfn_suffixes): Likewise.
6934 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6935 for math functions that have _Float<N> and _Float<N>X variants.
6936 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6937 and _Float<N>X variants.
6938 (COPYSIGN): Likewise.
6941 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6942 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6944 (integer_valued_read_call_p): Likewise.
6945 * fold-const-call.c (fold_const_call_ss): Likewise.
6946 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6947 _Float<N> and _Float<N>X variants.
6948 (fold_const_call_ssss): Add support for fma _Float<N> and
6949 _Float<N>X variants.
6950 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6951 support for copysign and fma _Float<N> and _Float<N>X variants.
6952 (backprop::process_builtin_call_use): Likewise.
6953 * tree-call-cdce.c (can_test_argument_range); Add support for
6954 sqrt _Float<N> and _Float<N>X variants.
6955 (edom_only_function): Likewise.
6956 (get_no_error_domain): Likewise.
6957 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6958 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6959 copysign _Float<N> and _Float<N>X variants.
6960 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6961 handled by machine independent code.
6962 (FMAF128): Likewise.
6963 * doc/cpp.texi (Common Predefined Macros): Document defining
6964 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6965 fma _Float<N> and _Float<N>X variants.
6967 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6970 * config/i386/i386-modes.def (CCFPU): Remove definition.
6971 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6972 (ix86_cc_modes_compatible): Ditto.
6973 (ix86_expand_carry_flag_compare): Ditto.
6974 (ix86_expand_int_movcc): Ditto.
6975 (ix86_expand_int_addcc): Ditto.
6976 (ix86_reverse_condition): Ditto.
6977 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6978 Return true/false for unordered/ordered fp comparisons.
6979 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6980 (ix86_prepare_fp_compare_args): Update for rename.
6981 (ix86_expand_fp_compare): Update for rename. Generate unordered
6982 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6983 (ix86_expand_sse_compare_and_jump): Ditto.
6984 * config/i386/predicates.md (fcmov_comparison_operator):
6985 Remove CCFPU mode handling.
6986 (ix86_comparison_operator): Ditto.
6987 (ix86_carry_flag_operator): Ditto.
6988 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6989 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6990 (*cmpu<mode>_cc_i387): Ditto.
6991 (FPCMP): Remove mode iterator.
6992 (unord): Remove mode attribute.
6993 (unord_subst): New define_subst transformation
6994 (unord): New define_subst attribute.
6996 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6997 (*cmpi<unord>xf_i387): Ditto.
6998 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6999 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7000 using unord_subst transformation.
7001 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7002 (round_saeonly): Also handle CCFP mode.
7003 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7004 Remove UNSPEC_SAHF unspec handling.
7006 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7008 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7010 2017-10-27 Jeff Law <law@redhat.com>
7012 * gimple-ssa-sprintf.c: Include domwalk.h.
7013 (class sprintf_dom_walker): New class, derived from dom_walker.
7014 (sprintf_dom_walker::before_dom_children): New function.
7015 (struct call_info): Moved into sprintf_dom_walker class
7016 (compute_formath_length, handle_gimple_call): Likewise.
7017 (sprintf_length::execute): Call the dominator walker rather
7018 than walking the statements.
7020 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7021 gimple statement locations.
7022 (check_array_bounds): Corresponding changes. Get the statement's
7023 location directly from wi->stmt.
7025 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7028 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7030 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7032 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7033 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7035 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7038 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7039 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7040 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7041 maybe_get_pool_constant.
7042 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7045 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7047 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7048 2.26 caveat. Update gas and gld versions.
7049 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7052 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7054 * cgraph.h (set_malloc_flag): Declare.
7055 * cgraph.c (set_malloc_flag_1): New function.
7056 (set_malloc_flag): Likewise.
7057 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7058 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7060 (read_ipa_call_summary): Add support for reading is_return_callee.
7061 (write_ipa_call_summary): Stream is_return_callee.
7062 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7063 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7064 ipa-prop.h, ipa-fnsummary.h.
7065 (pure_const_names): Change to static.
7066 (malloc_state_e): Define.
7067 (malloc_state_names): Define.
7068 (funct_state_d): Add field malloc_state.
7069 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7070 (check_retval_uses): New function.
7071 (malloc_candidate_p): Likewise.
7072 (analyze_function): Add support for malloc attribute.
7073 (pure_const_write_summary): Stream malloc_state.
7074 (pure_const_read_summary): Add support for reading malloc_state.
7075 (dump_malloc_lattice): New function.
7076 (propagate_malloc): New function.
7077 (warn_function_malloc): New function.
7078 (ipa_pure_const::execute): Call propagate_malloc and
7079 ipa_free_fn_summary.
7080 (pass_local_pure_const::execute): Add support for malloc attribute.
7081 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7082 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7084 2017-10-27 Martin Liska <mliska@suse.cz>
7086 PR gcov-profile/82457
7087 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7088 for fork-like functions to be properly instrumented.
7090 2017-10-27 Richard Biener <rguenther@suse.de>
7093 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7094 info when we redirected EH.
7096 2017-10-26 Michael Collison <michael.collison@arm.com>
7098 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7100 (<optab>_trunchf<GPI:mode>2: New pattern.
7101 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7102 * config/aarch64/iterators.md (wv): New mode attribute.
7103 (vf, VF): New mode attributes.
7104 (vgp, VGP): New mode attributes.
7105 (s): Update attribute with SImode and DImode prefixes.
7107 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7109 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7110 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7111 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7112 (nios2_option_overide): Initialize it. Don't allow R0-relative
7113 addressing with PIC.
7114 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7115 (nios2_symbolic_constant_p): Likewise.
7116 (nios2_legitimate_address_p): Likewise.
7117 (nios2_r0rel_section_name_p): New.
7118 (nios2_symbol_ref_in_r0rel_data_p): New.
7119 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7120 (r0rel_constant_p): New.
7121 (nios2_print_operand_address): Handle r0rel_constant_p.
7122 (nios2_cdx_narrow_form_p): Likewise.
7123 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7124 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7125 (Nios II Options): Document -mr0rel-sec.
7127 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7129 * config/nios2/nios2.c: Include xregex.h.
7130 (nios2_gprel_sec_regex): New.
7131 (nios2_option_overide): Initialize it. Don't allow GP-relative
7132 addressing with PIC.
7133 (nios2_small_section_name_p): Check for regex match.
7134 * config/nios2/nios2.opt (mgprel-sec=): New option.
7135 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7136 (Nios II Options): Document -mgprel-sec.
7138 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7140 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7142 2017-10-26 Tom de Vries <tom@codesourcery.com>
7144 PR tree-optimization/82707
7145 * gimple.c (gimple_copy): Fix unsharing of
7146 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7148 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7150 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7151 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7152 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7153 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7154 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7155 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7156 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7157 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7158 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7159 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7160 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7161 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7162 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7163 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7164 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7165 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7166 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7168 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7170 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7172 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7173 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7174 warning to rs6000.c. Remove the Undocumented flag, since it has
7176 (-mabi=ibmlongdouble): Likewise.
7177 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7178 already set, set the default format for long double.
7179 (rs6000_debug_reg_global): Print whether long double is IBM or
7181 (rs6000_option_override_internal): Rework setting long double
7182 format. Only warn if the user is changing the long double default
7183 and they did not use -Wno-psabi.
7184 * doc/invoke.texi (PowerPC options): Update the documentation for
7185 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7187 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7188 Alan Hayward <alan.hayward@arm.com>
7189 David Sherwood <david.sherwood@arm.com>
7191 * rtl.h (wider_subreg_mode): New function.
7192 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7193 rather than an unsigned int *.
7194 * ira-color.c (regno_max_ref_width): Replace with...
7195 (regno_max_ref_mode): ...this new variable.
7196 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7197 Use wider_subreg_mode.
7198 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7199 rather than an unsigned int *.
7200 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7201 (process_alt_operands): Likewise.
7202 (invariant_p): Likewise.
7203 * lra-spills.c (assign_mem_slot): Likewise.
7204 (add_pseudo_to_slot): Likewise.
7205 * lra.c (collect_non_operand_hard_regs): Likewise.
7206 (add_regs_to_insn_regno_info): Likewise.
7207 * reload1.c (regno_max_ref_width): Replace with...
7208 (regno_max_ref_mode): ...this new variable.
7209 (reload): Update accordingly. Update call to
7210 ira_sort_regnos_for_alter_reg.
7211 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7212 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7213 (scan_paradoxical_subregs): Likewise.
7215 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7217 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7218 (aarch64_frame): Add emit_frame_chain boolean.
7219 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7220 Move eh_return case to aarch64_layout_frame.
7221 (aarch64_layout_frame): Initialize emit_frame_chain.
7222 (aarch64_expand_prologue): Use emit_frame_chain.
7224 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7226 * config/aarch64/aarch64.c (aarch64_layout_frame):
7227 Ensure LR is always stored at the bottom of the callee-saves.
7228 Remove rarely used frame layout which saves callee-saves at top of
7229 frame, so the store of LR can be used as a valid probe in all cases.
7231 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7233 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7234 Improve unaligned TImode/TFmode base/offset split.
7236 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7237 Alan Hayward <alan.hayward@arm.com>
7238 David Sherwood <david.sherwood@arm.com>
7240 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7241 * combine.c (find_single_use_1): Likewise.
7242 (expand_field_assignment): Likewise.
7243 (move_deaths): Likewise.
7244 * lra-constraints.c (simplify_operand_subreg): Likewise.
7245 (curr_insn_transform): Likewise.
7246 * lra.c (collect_non_operand_hard_regs): Likewise.
7247 (add_regs_to_insn_regno_info): Likewise.
7248 * rtlanal.c (reg_referenced_p): Likewise.
7249 (covers_regno_no_parallel_p): Likewise.
7251 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7253 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
7254 Don't print any bits outside the precision of the value.
7255 * wide-int.cc (test_printing): Add some new tests.
7257 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7259 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
7260 supports -xbrace_comment option.
7261 * configure: Regenerate.
7262 * config.in: Regenerate.
7263 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
7264 (ASM_CPU_SPEC): Use it.
7266 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7268 * target.def (static_rtx_alignment): New hook.
7269 * targhooks.h (default_static_rtx_alignment): Declare.
7270 * targhooks.c (default_static_rtx_alignment): New function.
7271 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
7272 * doc/tm.texi: Regenerate.
7273 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
7274 instead of targetm.constant_alignment. Remove call to
7276 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7277 (cris_preferred_mininum_alignment): New function, split out from...
7278 (cris_constant_alignment): ...here.
7279 (cris_static_rtx_alignment): New function.
7280 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
7282 (ix86_constant_alignment): ...here.
7283 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7284 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7285 (mmix_static_rtx_alignment): New function.
7286 * config/spu/spu.c (spu_static_rtx_alignment): New function.
7287 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
7289 2017-10-26 Tamar Christina <tamar.christina@arm.com>
7292 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
7293 Add flag_trapping_math and flag_fp_int_builtin_inexact.
7295 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
7297 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
7298 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
7299 mark as a sign-extending load.
7300 (local_pic_load_u): Define.
7302 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
7305 * fold-const.c (operand_equal_for_comparison_p): Also return true
7306 if ARG0 is a simple variant of ARG1 with narrower precision.
7307 (fold_ternary_loc): Always pass unstripped operands to the predicate.
7309 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
7311 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
7313 * i386.h (processor_costs): Add gather_static, gather_per_elt,
7314 scatter_static, scatter_per_elt.
7315 * x86-tune-costs.h: Add new cost entries.
7317 2017-10-25 Richard Biener <rguenther@suse.de>
7319 * tree-ssa-sccvn.h (vn_eliminate): Declare.
7320 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
7321 class pass_fre): Move to ...
7322 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
7323 class pass_fre): ... here and adjust for statistics.
7325 2017-10-25 Jakub Jelinek <jakub@redhat.com>
7328 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
7329 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
7330 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
7331 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
7334 2017-10-25 Richard Biener <rguenther@suse.de>
7336 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
7337 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
7338 eliminate_push_avail, eliminate_insert): Move inside...
7339 (class eliminate_dom_walker): ... this class in preparation
7341 (fini_eliminate): Remove by merging with ...
7342 (eliminate): ... this function. Adjust for class changes.
7343 (pass_pre::execute): Remove fini_eliminate call.
7344 (pass_fre::execute): Likewise.
7346 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7349 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
7350 (VPERMI2, VPERMI2I): New mode iterators.
7351 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
7352 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
7354 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
7355 mode iterator. Remove 3 old define_insn patterns.
7356 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
7357 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
7358 VPERMI2 mode iterator, remove the other two expanders.
7359 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
7360 to use VPERMI2 mode iterator, add another alternative for vpermi2*
7361 instructions, remove the other two patterns.
7362 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
7363 mode iterator, remove the other two patterns.
7364 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
7365 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
7366 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
7367 and adjust argument order accordingly.
7368 (ix86_expand_vec_perm): Adjust caller.
7369 (expand_vec_perm_1): Likewise.
7370 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
7371 (expand_vec_perm_vpermt2_vpshub2): ... this.
7372 (ix86_expand_vec_perm_const_1): Adjust caller.
7373 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
7376 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
7377 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
7378 (vec_shl_<mode>): Remove unused expander.
7379 (avx512bw_<shift_insn><mode>3): New define_insn.
7380 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
7381 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
7383 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
7386 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
7389 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7391 PR rtl-optimization/82396
7392 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
7393 (autopref_multipass_init): Simplify initialization.
7394 (autopref_rank_data): Simplify sort order.
7395 * gcc/sched-int.h (autopref_multipass_data_): Remove
7396 multi_mem_insn_p, min_offset and max_offset.
7398 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
7401 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7402 Check special value of flag_omit_frame_pointer.
7403 (aarch64_can_eliminate): Likewise.
7404 (aarch64_override_options_after_change_1): Simplify handling of
7405 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
7407 2017-10-24 Richard Biener <rguenther@suse.de>
7409 PR tree-optimization/82697
7410 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
7411 zero for conditional load and unconditional store.
7413 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7415 * doc/install.texi: Document bootstrap-cet.
7417 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
7420 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
7421 ENDBR instruction at function entrance if function is only
7424 2017-10-24 Jakub Jelinek <jakub@redhat.com>
7427 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
7428 patterns to better describe from which operation the CF is computed.
7429 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
7430 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
7431 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
7432 is 0, use _0 suffixed expanders instead of emitting a comparison
7435 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
7437 * config/i386/i386.md(*movsf_internal, *movdf_internal):
7438 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
7440 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
7443 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
7444 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
7445 * loop-iv.c (iv_get_reaching_def): Likewise.
7446 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
7447 variable is promoted and the partition contains undefined values.
7449 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7451 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
7453 (nios2_address_cost): Define.
7454 (nios2_legitimize_address): Recognize (exp + constant) directly.
7455 (TARGET_ADDRESS_COST): Define.
7457 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7459 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
7460 (nios2_symbolic_memory_operand_p): Declare.
7461 (nios2_split_large_constant): Declare.
7462 (nios2_split_symbolic_memory_operand): Declare.
7463 * config/nios2/nios2.c: Adjust includes.
7464 (nios2_symbolic_constant_allowed): New.
7465 (nios2_symbolic_constant_p): New.
7466 (nios2_plus_symbolic_constant_p): New.
7467 (nios2_valid_addr_expr_p): Recognize addresses involving
7469 (nios2_legitimate_address_p): Likewise, also LO_SUM.
7470 (nios2_symbolic_memory_operand_p): New.
7471 (nios2_large_constant_p): New.
7472 (nios2_split_large_constant): New.
7473 (nios2_split_plus_large_constant): New.
7474 (nios2_split_symbolic_memory_operand): New.
7475 (nios2_legitimize_address): Code refactoring. Handle addresses
7476 involving symbolic constants.
7477 (nios2_emit_move_sequence): Likewise.
7478 (nios2_print_operand): Improve error output.
7479 (nios2_print_operand_address): Handle LO_SUM.
7480 (nios2_cdx_narrow_form_p): Likewise.
7481 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
7482 operands involving symbolic constants.
7483 (movhi_internal, movsi_internal): Likewise.
7484 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
7485 (extendhisi2, extendqi<mode>2): Likewise.
7487 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7489 * tree-pass.h (PROP_rtl_split_insns): Define.
7490 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
7492 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
7494 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
7496 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7499 * target.def (const_not_ok_for_debug_p): Default to
7500 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
7501 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
7502 * targhooks.c (default_const_not_ok_for_debug_p): New function.
7503 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
7504 which targetm.const_not_ok_for_debug_p returned true.
7505 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
7507 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
7509 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
7510 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
7511 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
7513 (ix86_const_not_ok_for_debug_p): New function.
7514 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
7515 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
7517 2017-10-23 David Malcolm <dmalcolm@redhat.com>
7520 * system.h: Conditionally include "unique-ptr.h" if
7521 INCLUDE_UNIQUE_PTR is defined.
7522 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
7523 of defining INCLUDE_UNIQUE_PTR before including "system.h".
7525 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
7527 * config/rl78/rl78.md: New define_expand "subdi3".
7529 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
7532 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
7533 DF_REF_INSN if DF_REF_INSN_INFO is false.
7535 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
7537 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
7538 xmm_move instead of sse_move.
7539 (sse_store_index): New function.
7540 (ix86_register_move_cost): Be more sensible about mismatch stall;
7541 model AVX moves correctly; make difference between sse->integer and
7543 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
7544 moves; make difference between SSE and AVX.
7545 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
7546 and zmm_move. Increase size of sse load and store tables;
7547 add unaligned load and store tables; add ssemmx_to_integer.
7548 * x86-tune-costs.h: Update all entries according to real
7549 move latencies from Agner Fog's manual and chip documentation.
7551 2017-10-23 Jakub Jelinek <jakub@redhat.com>
7554 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
7555 * config/i386/constraints.md (Wf): New constraint.
7556 * config/i386/i386.md (UNSPEC_SBB): New unspec.
7557 (cmp<dwi>_doubleword): Removed.
7558 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
7559 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
7560 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
7561 expand with cmp<dwi>_doubleword. For LTU and GEU use
7562 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
7564 * common.opt (gcolumn-info): Enable by default.
7565 * doc/invoke.texi (gcolumn-info): Document new default.
7567 2017-10-23 Richard Biener <rguenther@suse.de>
7569 PR tree-optimization/82672
7570 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
7571 Fold the stmt if we propagated into it.
7573 2017-10-23 Richard Biener <rguenther@suse.de>
7575 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
7576 (bitmap_remove_expr_from_set): ... this. All callers call this
7577 for non-constant values.
7578 (bitmap_set_subtract): Rename to...
7579 (bitmap_set_subtract_expressions): ... this. Adjust and
7581 (bitmap_set_contains_value): Remove superfluous check.
7582 (bitmap_set_replace_value): Inline into single caller ...
7583 (bitmap_value_replace_in_set): ... here and simplify.
7584 (dependent_clean): Merge into ...
7585 (clean): ... this using an overload. Adjust.
7586 (prune_clobbered_mems): Adjust.
7587 (compute_antic_aux): Likewise.
7588 (compute_partial_antic_aux): Likewise.
7590 2017-10-23 Richard Biener <rguenther@suse.de>
7592 PR tree-optimization/82129
7594 2017-08-01 Richard Biener <rguenther@suse.de>
7596 PR tree-optimization/81181
7597 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7598 (compute_antic): ... end of iteration here.
7600 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7602 * target.def (starting_frame_offset): New hook.
7603 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
7604 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
7605 * doc/tm.texi.in: Regenerate.
7606 * hooks.h (hook_hwi_void_0): Declare.
7607 * hooks.c (hook_hwi_void_0): New function.
7608 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
7609 STARTING_FRAME_OFFSET.
7610 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
7611 * reload1.c (reload): Likewise.
7612 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
7613 instead of STARTING_FRAME_OFFSET.
7614 * function.c (try_fit_stack_local): Likewise.
7615 (assign_stack_local_1): Likewise
7616 (instantiate_virtual_regs): Likewise.
7617 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
7618 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
7619 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
7620 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
7621 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
7622 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
7623 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
7624 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
7625 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
7626 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
7627 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
7628 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
7629 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
7630 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
7631 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
7632 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
7633 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
7634 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
7635 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
7636 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
7637 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
7638 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
7639 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
7640 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
7641 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
7642 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
7643 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
7644 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
7645 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
7646 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
7647 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
7648 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
7649 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
7650 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
7651 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
7652 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
7653 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
7654 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
7655 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
7656 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
7657 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
7658 * config/avr/avr.c (avr_starting_frame_offset): Make static and
7659 return a HOST_WIDE_INT.
7660 (avr_builtin_setjmp_frame_value): Use it instead of
7661 STARTING_FRAME_OFFSET.
7662 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7663 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
7664 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
7666 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7667 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
7668 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
7669 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7670 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
7671 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
7672 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7673 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
7674 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
7675 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7676 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
7677 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
7679 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7680 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
7681 * config/mips/mips.c (mips_compute_frame_info): Refer to
7682 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7683 (mips_starting_frame_offset): New function.
7684 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7685 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
7686 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
7687 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
7688 and return a HOST_WIDE_INT.
7689 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7690 (mmix_initial_elimination_offset): Refer to
7691 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
7692 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
7693 * config/pa/pa.c (pa_starting_frame_offset): New function.
7694 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
7695 (pa_expand_prologue): Likewise.
7696 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7697 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
7698 !FRAME_GROWS_DOWNWARD handling to...
7699 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7700 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
7701 !FRAME_GROWS_DOWNWARD handling to...
7702 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7703 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
7704 !FRAME_GROWS_DOWNWARD handling to...
7705 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7706 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
7708 (rs6000_starting_frame_offset): New function.
7709 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
7710 !FRAME_GROWS_DOWNWARD handling to...
7711 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7712 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
7713 !FRAME_GROWS_DOWNWARD handling to...
7714 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7715 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
7716 !FRAME_GROWS_DOWNWARD handling to...
7717 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
7718 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
7719 (rs6000_starting_frame_offset): New function.
7720 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
7721 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
7722 * config/vax/vax.c (vax_starting_frame_offset): New function.
7723 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
7724 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7725 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
7726 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
7727 (TARGET_STARTING_FRAME_OFFSET): Redefine.
7728 * system.h (STARTING_FRAME_OFFSET): Poison.
7730 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7732 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
7733 SCALAR_TYPE_MODE instead of TYPE_MODE.
7735 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7736 Alan Hayward <alan.hayward@arm.com>
7737 David Sherwood <david.sherwood@arm.com>
7739 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
7741 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7742 Alan Hayward <alan.hayward@arm.com>
7743 David Sherwood <david.sherwood@arm.com>
7745 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
7747 2017-10-23 Richard Biener <rguenther@suse.de>
7749 PR tree-optimization/82129
7750 * tree-ssa-pre.c (bitmap_set_and): Remove.
7751 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
7752 canonicalizing expressions in the set to those with lowest
7753 ID rather than taking that from the first edge.
7755 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
7757 * combine.c (rtx_equal_for_field_assignment_p): Use
7758 byte_lowpart_offset.
7760 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7761 Alan Hayward <alan.hayward@arm.com>
7762 David Sherwood <david.sherwood@arm.com>
7764 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
7765 to a SUBREG_PROMOTED_VAR.
7767 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7768 Alan Hayward <alan.hayward@arm.com>
7769 David Sherwood <david.sherwood@arm.com>
7771 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
7772 (expand_debug_source_expr): Likewise.
7773 * combine.c (combine_simplify_rtx): Likewise.
7774 * cse.c (fold_rtx): Likewise.
7775 * optabs.c (expand_float): Likewise.
7776 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7777 (simplify_binary_operation_1): Likewise.
7779 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7780 Alan Hayward <alan.hayward@arm.com>
7781 David Sherwood <david.sherwood@arm.com>
7783 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
7784 (record_promoted_value): Likewise.
7785 * expr.c (expand_expr_real_2): Likewise.
7786 * ree.c (update_reg_equal_equiv_notes): Likewise.
7787 (combine_set_extension): Likewise.
7788 * rtlanal.c (low_bitmask_len): Likewise.
7789 * simplify-rtx.c (neg_const_int): Likewise.
7790 (simplify_binary_operation_1): Likewise.
7792 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7793 Alan Hayward <alan.hayward@arm.com>
7794 David Sherwood <david.sherwood@arm.com>
7796 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
7797 * regcprop.c (maybe_mode_change): Likewise.
7798 * reload1.c (alter_reg): Likewise.
7800 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7802 * inchash.h (inchash::hash::add_wide_int): New function.
7803 * lto-streamer-out.c (hash_tree): Use it.
7805 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
7807 * inchash.h (inchash::hash::add_wide_int): Rename to...
7808 (inchash::hash::add_hwi): ...this.
7809 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
7810 (polymorphic_call_target_hasher::hash): Likewise.
7811 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
7812 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
7813 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
7814 * lto-streamer-out.c (hash_tree): Likewise.
7815 * optc-save-gen.awk: Likewise.
7816 * tree.c (add_expr): Likewise.
7818 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7821 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
7822 for ordered inequality comparisons even with TARGET_IEEE_FP.
7824 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
7827 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
7828 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7829 Expand with cmp<dwi>_doubleword.
7831 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7833 * extend.texi: Add x86 specific to 'nocf_check' attribute.
7834 List CET intrinsics.
7835 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
7836 specific to -fcf-protection option.
7838 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7840 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
7841 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
7842 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7843 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
7844 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
7845 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
7846 (extra_objs): Add cet.o for Linux/x86 targets.
7847 (tmake_file): Add i386/t-cet for Linux/x86 targets.
7848 * config/i386/cet.c: New file.
7849 * config/i386/cetintrin.h: Likewise.
7850 * config/i386/t-cet: Likewise.
7851 * config/i386/cpuid.h (bit_SHSTK): New.
7852 (bit_IBT): Likewise.
7853 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
7854 pass IBT and SHSTK bits.
7855 * config/i386/i386-builtin-types.def
7856 (VOID_FTYPE_UNSIGNED_PVOID): New.
7857 (VOID_FTYPE_UINT64_PVOID): Likewise.
7858 * config/i386/i386-builtin.def: Add CET intrinsics.
7859 * config/i386/i386-c.c (ix86_target_macros_internal): Add
7860 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
7861 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
7862 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
7864 * config/i386/i386.c (rest_of_insert_endbranch): New.
7865 (pass_data_insert_endbranch): Likewise.
7866 (pass_insert_endbranch): Likewise.
7867 (make_pass_insert_endbranch): Likewise.
7868 (ix86_notrack_prefixed_insn_p): Likewise.
7869 (ix86_target_string): Add -mibt, -mshstk flags.
7870 (ix86_option_override_internal): Add flag_cf_protection
7872 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
7873 (ix86_print_operand): Add 'notrack' prefix output.
7874 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
7875 (ix86_expand_builtin): Expand CET intrinsics.
7876 (x86_output_mi_thunk): Add 'endbranch' instruction.
7877 * config/i386/i386.h (TARGET_IBT): New.
7878 (TARGET_IBT_P): Likewise.
7879 (TARGET_SHSTK): Likewise.
7880 (TARGET_SHSTK_P): Likewise.
7881 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
7882 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
7883 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
7884 (builtin_setjmp_setup): New pattern.
7885 (builtin_longjmp): Likewise.
7886 (rdssp<mode>): Likewise.
7887 (incssp<mode>): Likewise.
7888 (saveprevssp): Likewise.
7889 (rstorssp): Likewise.
7890 (wrss<mode>): Likewise.
7891 (wruss<mode>): Likewise.
7892 (setssbsy): Likewise.
7893 (clrssbsy): Likewise.
7894 (nop_endbr): Likewise.
7895 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7897 * config/i386/immintrin.h: Include <cetintrin.h>.
7898 * config/i386/linux-common.h
7899 (file_end_indicate_exec_stack_and_cet): New prototype.
7900 (TARGET_ASM_FILE_END): New.
7902 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7904 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7905 latencies instead of having separate table; make difference between
7906 integer and float costs.
7907 * i386.h (processor_costs): Remove scalar_stmt_cost,
7908 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7909 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7911 * x86-tune-costs.h: Remove entries which has been removed in
7912 procesor_costs from all tables; make cond_taken_branch_cost
7913 and cond_not_taken_branch_cost COST_N_INSNS based.
7915 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7917 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7919 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7921 * config/i386/i386.md (isa): Remove fma_avx512f.
7922 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7923 <avx512>_fmadd_<mode>_mask3<round_name>,
7924 <avx512>_fmsub_<mode>_mask<round_name>,
7925 <avx512>_fmsub_<mode>_mask3<round_name>,
7926 <avx512>_fnmadd_<mode>_mask<round_name>,
7927 <avx512>_fnmadd_<mode>_mask3<round_name>,
7928 <avx512>_fnmsub_<mode>_mask<round_name>,
7929 <avx512>_fnmsub_<mode>_mask3<round_name>,
7930 <avx512>_fmaddsub_<mode>_mask<round_name>,
7931 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7932 <avx512>_fmsubadd_<mode>_mask<round_name>,
7933 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7934 (*vec_widen_umult_even_v16si<mask_name>,
7935 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7936 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7938 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7940 * extend.texi: Add 'nocf_check' documentation.
7941 * gimple.texi: Add second parameter to
7942 gimple_build_call_from_tree.
7943 * invoke.texi: Add -fcf-protection documentation.
7944 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7946 2017-10-20 Richard Biener <rguenther@suse.de>
7948 PR tree-optimization/82473
7949 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7950 the largest input type.
7952 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7954 * c-attribs.c (handle_nocf_check_attribute): New function.
7955 (c_common_attribute_table): Add 'nocf_check' handling.
7956 * gimple-parser.c: Add second argument NULL to
7957 gimple_build_call_from_tree.
7958 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7959 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7961 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7962 * common.opt: Add fcf-protection flag.
7963 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7964 * flag-types.h: Add enum cf_protection_level.
7965 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7966 Add 'nocf_check' attribute propagation to gimple call.
7967 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7968 (gimple_build_call_from_tree): Update prototype.
7969 (gimple_call_nocf_check_p): New function.
7970 (gimple_call_set_nocf_check): Likewise.
7971 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7972 * ipa-icf.c: Add nocf_check attribute in statement hash.
7973 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7974 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7975 * toplev.c (process_options): Add flag_cf_protection handling.
7977 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7979 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7981 2017-10-20 Richard Biener <rguenther@suse.de>
7983 PR tree-optimization/82603
7984 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7985 remove false predicated stores.
7987 2017-10-20 Richard Biener <rguenther@suse.de>
7989 * graphite-isl-ast-to-gimple.c
7990 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7991 Remove return value and simplify, dump copied stmt after lhs
7993 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7994 Reduce dump verbosity.
7995 (gsi_insert_earliest): Likewise.
7996 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7997 * graphite.c (print_global_statistics): Adjust dumping.
7998 (print_graphite_scop_statistics): Likewise.
7999 (print_graphite_statistics): Do not dump loops here.
8000 (graphite_transform_loops): But here.
8002 2017-10-20 Nicolas Roche <roche@adacore.com>
8004 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8005 * configure: Regenerate.
8007 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8010 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8011 functions when optimizing replace GIMPLE_RETURN stmts with
8012 calls to __builtin_unreachable ().
8015 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8016 for -fsanitize=thread link of executables.
8017 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8018 link of executables.
8021 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8023 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8024 define_insns for logical vector shifts to use VI248_AVX512BW
8025 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8026 condition, useless isa and prefix attributes. Change the first
8027 2 of these define_insns to ...
8028 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8029 define_insn for avx512vl.
8030 (<shift_insn><mode>3): ... and this, new define_insn without
8031 masking for non-avx512vl.
8034 * config/i386/sse.md (*andnot<mode>3,
8035 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8036 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8037 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8038 not applied use empty suffix even for TARGET_AVX512VL.
8039 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8040 is applied, supply evex,evex or evex,evex,evex instead of just
8043 2017-10-20 Julia Koval <julia.koval@intel.com>
8045 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8046 (OPTION_MASK_ISA_GFNI_UNSET): New.
8047 (ix86_handle_option): Handle OPT_mgfni.
8048 * config/i386/cpuid.h (bit_GFNI): New.
8049 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8050 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8051 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8052 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8053 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8054 * config/i386/i386.opt: Add mgfni.
8056 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8058 * config/msp430/msp430.c (msp430_option_override): Disable
8059 -fdelete-null-pointer-checks.
8060 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8062 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8064 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8065 of x87 and SSE instructions.
8067 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8069 * asan.c (create_cond_insert_point): Do not update edge count.
8070 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8071 (afdo_propagate_circuit): Likewise.
8072 (afdo_calculate_branch_prob): Likewise.
8073 (afdo_annotate_cfg): Likewise.
8074 * basic-block.h (struct edge_def): Remove count.
8075 (edge_def::count): New accessor.
8076 * bb-reorder.c (rotate_loop): Update.
8077 (find_traces_1_round): Update.
8078 (connect_traces): Update.
8079 (sanitize_hot_paths): Update.
8080 * cfg.c (unchecked_make_edge): Update.
8081 (make_single_succ_edge): Update.
8082 (check_bb_profile): Update.
8083 (dump_edge_info): Update.
8084 (update_bb_profile_for_threading): Update.
8085 (scale_bbs_frequencies_int): Update.
8086 (scale_bbs_frequencies_gcov_type): Update.
8087 (scale_bbs_frequencies_profile_count): Update.
8088 (scale_bbs_frequencies): Update.
8089 * cfganal.c (connect_infinite_loops_to_exit): Update.
8090 * cfgbuild.c (compute_outgoing_frequencies): Update.
8091 (find_many_sub_basic_blocks): Update.
8092 * cfgcleanup.c (try_forward_edges): Update.
8093 (try_crossjump_to_edge): Update
8094 * cfgexpand.c (expand_gimple_cond): Update
8095 (expand_gimple_tailcall): Update
8096 (construct_exit_block): Update
8097 * cfghooks.c (verify_flow_info): Update
8098 (redirect_edge_succ_nodup): Update
8099 (split_edge): Update
8100 (make_forwarder_block): Update
8101 (duplicate_block): Update
8102 (account_profile_record): Update
8103 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8104 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8105 * cfgloopmanip.c (scale_loop_profile): Update.
8107 (lv_adjust_loop_entry_edge): Update.
8108 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8109 (force_nonfallthru_and_redirect): Update.
8110 (purge_dead_edges): Update.
8111 (rtl_flow_call_edges_add): Update.
8112 * cgraphunit.c (init_lowered_empty_function): Update.
8113 (cgraph_node::expand_thunk): Update.
8114 * gimple-pretty-print.c (dump_probability): Update.
8115 (dump_edge_probability): Update.
8116 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8117 * haifa-sched.c (sched_create_recovery_edges): Update.
8118 * hsa-gen.c (convert_switch_statements): Update.
8119 * ifcvt.c (dead_or_predicable): Update.
8120 * ipa-inline-transform.c (inline_transform): Update.
8121 * ipa-split.c (split_function): Update.
8122 * ipa-utils.c (ipa_merge_profiles): Update.
8123 * loop-doloop.c (add_test): Update.
8124 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8125 * lto-streamer-in.c (input_cfg): Update.
8126 (input_function): Update.
8127 * lto-streamer-out.c (output_cfg): Update.
8128 * modulo-sched.c (sms_schedule): Update.
8129 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8130 * predict.c (maybe_hot_edge_p): Update.
8131 (unlikely_executed_edge_p): Update.
8132 (probably_never_executed_edge_p): Update.
8133 (dump_prediction): Update.
8134 (drop_profile): Update.
8135 (propagate_unlikely_bbs_forward): Update.
8136 (determine_unlikely_bbs): Update.
8137 (force_edge_cold): Update.
8138 * profile.c (compute_branch_probabilities): Update.
8139 * reg-stack.c (better_edge): Update.
8140 * shrink-wrap.c (handle_simple_exit): Update.
8141 * tracer.c (better_p): Update.
8142 * trans-mem.c (expand_transaction): Update.
8143 (split_bb_make_tm_edge): Update.
8144 * tree-call-cdce.c: Update.
8145 * tree-cfg.c (gimple_find_sub_bbs): Update.
8146 (gimple_split_edge): Update.
8147 (gimple_duplicate_sese_region): Update.
8148 (gimple_duplicate_sese_tail): Update.
8149 (gimple_flow_call_edges_add): Update.
8150 (insert_cond_bb): Update.
8151 (execute_fixup_cfg): Update.
8152 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8153 * tree-complex.c (expand_complex_div_wide): Update.
8154 * tree-eh.c (lower_resx): Update.
8155 (unsplit_eh): Update.
8156 (cleanup_empty_eh_move_lp): Update.
8157 * tree-inline.c (copy_edges_for_bb): Update.
8158 (freqs_to_counts): Update.
8159 (copy_cfg_body): Update.
8160 * tree-ssa-dce.c (remove_dead_stmt): Update.
8161 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8162 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8163 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8164 (unloop_loops): Update.
8165 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8166 * tree-ssa-loop-split.c (connect_loops): Update.
8167 (split_loop): Update.
8168 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8169 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8170 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8171 * tree-ssa-reassoc.c (branch_fixup): Update.
8172 * tree-ssa-tail-merge.c (replace_block_by): Update.
8173 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8174 (compute_path_counts): Update.
8175 (update_profile): Update.
8176 (recompute_probabilities): Update.
8177 (update_joiner_offpath_counts): Update.
8178 (estimated_freqs_path): Update.
8179 (freqs_to_counts_path): Update.
8180 (clear_counts_path): Update.
8181 (ssa_fix_duplicate_block_edges): Update.
8182 (duplicate_thread_path): Update.
8183 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8184 (case_bit_test_cmp): Update.
8185 (collect_switch_conv_info): Update.
8186 (gen_inbound_check): Update.
8187 (do_jump_if_equal): Update.
8188 (emit_cmp_and_jump_insns): Update.
8189 * tree-tailcall.c (decrease_profile): Update.
8190 (eliminate_tail_call): Update.
8191 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8192 (vect_do_peeling): Update.
8193 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8194 * ubsan.c (ubsan_expand_null_ifn): Update.
8195 (ubsan_expand_ptr_ifn): Update.
8196 * value-prof.c (gimple_divmod_fixed_value): Update.
8197 (gimple_mod_pow2): Update.
8198 (gimple_mod_subtract): Update.
8199 (gimple_ic): Update.
8200 (gimple_stringop_fixed_value): Update.
8202 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8205 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8207 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8209 PR rtl-optimization/82395
8210 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8211 based on non_spilled_static_chain_regno_p.
8213 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8215 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8216 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8217 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8219 2017-10-19 Martin Sebor <msebor@redhat.com>
8221 PR tree-optimization/82596
8222 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8224 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8226 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8227 (get_mem_refs_of_builtin_call): Likewise.
8228 * builtins.c (expand_builtin_apply): Adjust call to
8229 allocate_dynamic_stack_space.
8230 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8231 the third argument to allocate_dynamic_stack_space, otherwise -1.
8232 (expand_builtin): Deal with all alloca variants.
8233 (is_inexpensive_builtin): Likewise.
8234 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8235 * calls.c (special_function_p): Deal with all alloca variants.
8236 (initialize_argument_information): Adjust call to
8237 allocate_dynamic_stack_space.
8238 (expand_call): Likewise.
8239 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8240 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8241 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8242 use it for the stack usage computation.
8243 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8244 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8245 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8246 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8247 (in_loop_p): Remove first argument and useless check.
8248 (pass_walloca::execute): Remove useless test and adjust call to above.
8249 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
8250 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
8251 (gimplify_call_expr): Deal with all alloca variants.
8252 * hsa-gen.c (gen_hsa_alloca): Likewise.
8253 (gen_hsa_insns_for_call): Likewise.
8254 * ipa-pure-const.c (special_builtin_state): Likewise.
8255 * tree-chkp.c (chkp_build_returned_bound): Likewise.
8256 * tree-object-size.c (alloc_object_size): Likewise.
8257 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
8258 (call_may_clobber_ref_p_1): Likewise.
8259 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
8260 (ccp_fold_stmt): Likewise.
8261 (optimize_stack_restore): Likewise.
8262 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8263 (mark_all_reaching_defs_necessary_1): Likewise.
8264 (propagate_necessity): Likewise.
8265 (eliminate_unnecessary_stmts): Likewise.
8266 * tree.c (build_common_builtin_nodes): Build
8267 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
8268 (build_alloca_call_expr): New function.
8269 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
8270 (CASE_BUILT_IN_ALLOCA): Likewise.
8271 (build_alloca_call_expr): Declare.
8272 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
8274 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8277 * dwarf2out.c (new_die_raw): New static inline function.
8278 (new_die): Use it to create the DIE.
8279 (add_AT_external_die_ref): Likewise.
8280 (clone_die): Likewise.
8281 (clone_as_declaration): Likewise.
8282 (dwarf2out_vms_debug_main_pointer): Likewise.
8283 (base_type_die): Likewise. Remove early return for corner cases.
8284 Do not call add_pubtype on the DIE here.
8285 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
8286 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
8287 typedefs for base types with DW_AT_endianity. Make sure a DIE with
8288 native order exists for base types, attach the DIE manually and call
8289 add_pubtype on it. Do not equate a reverse order DIE to the type.
8291 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
8293 * config/arm/arm.c (align_ok_ldrd_strd): New function.
8294 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
8296 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
8298 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8300 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
8301 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
8302 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
8303 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
8304 * opts.c (sanitizer_opts): Add builtin.
8305 * ubsan.c (instrument_builtin): New function.
8306 (pass_ubsan::execute): Call it.
8307 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
8308 * doc/invoke.texi: Document -fsanitize=builtin.
8310 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
8311 builtins, store max (log2 (align), 0) into uchar field instead of
8312 align into uptr field.
8313 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
8314 store uchar 0 field instead of uptr 0 field.
8315 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
8316 instead of passing one address of struct with 2 locations pass
8317 two addresses of structs with 1 location each.
8318 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
8319 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8320 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
8321 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
8322 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
8323 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
8324 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
8325 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
8327 2017-10-19 Martin Liska <mliska@suse.cz>
8330 * file-find.c (remove_prefix): Remove.
8331 * file-find.h (remove_prefix): Likewise.
8332 * gcc-ar.c: Remove smartness of lookup.
8334 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
8336 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
8337 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
8338 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
8340 2017-10-19 Jakub Jelinek <jakub@redhat.com>
8343 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
8344 (setcc + and to xor + setcc): New peephole2.
8346 2017-10-19 Tom de Vries <tom@codesourcery.com>
8348 * doc/sourcebuild.texi (Test Directives, Variants of
8349 dg-require-support): Add dg-require-stack-size.
8351 2017-10-19 Martin Liska <mliska@suse.cz>
8354 * gimplify.c (gimplify_decl_expr): Do not instrument variables
8355 that have a large alignment.
8356 (gimplify_target_expr): Likewise.
8358 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
8360 PR rtl-optimization/82602
8361 * ira.c (rtx_moveable_p): Return false for volatile asm.
8363 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
8366 * config/i386/i386-modes.def (CCGZ): New CC mode.
8367 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
8368 * config/i386/predicates.md (ix86_comparison_operator):
8370 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8371 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
8372 with double-word subtraction.
8373 (put_condition_code): Handle CCGZmode.
8375 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
8377 * wide-int.cc (debug (const wide_int &)): New.
8378 (debug (const wide_int *)): New.
8379 (debug (const widest_int &)): New.
8380 (debug (const widest_int *)): New.
8382 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
8385 * lra-constraints.c (curr_insn_transform): Use non-input operand
8386 instead of output one for matched reload.
8388 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8390 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
8391 (tree-ssa-loop-ivopts.h): New header file.
8392 (struct builtin_info): New fields.
8393 (classify_builtin_1): Compute and record base and offset parts for
8394 memset builtin partition by calling strip_offset.
8395 (offset_cmp, fuse_memset_builtins): New functions.
8396 (finalize_partitions): Fuse adjacent memset partitions by calling
8398 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
8399 Expose the interface.
8400 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
8402 2017-10-18 Bin Cheng <bin.cheng@arm.com>
8404 PR tree-optimization/82574
8405 * tree-loop-distribution.c (find_single_drs): New parameter. Check
8406 that data reference must be executed exactly once per iteration
8407 against the outermost loop in nest.
8408 (classify_partition): Update call to above function.
8410 2017-10-18 Richard Biener <rguenther@suse.de>
8412 PR tree-optimization/82591
8413 * graphite.c (graphite_transform_loops): Move code gen message
8415 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8416 Here. Handle scop_to_isl_ast failing.
8417 (scop_to_isl_ast): Limit the number of ISL operations.
8419 2017-10-18 Richard Biener <rguenther@suse.de>
8421 * graphite-isl-ast-to-gimple.c
8422 (translate_isl_ast_to_gimple::set_rename): Simplify.
8423 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
8424 (graphite_copy_stmts_from_block): ... here.
8425 (copy_bb_and_scalar_dependences): Simplify.
8426 (add_parameters_to_ivs_params): Canonicalize.
8427 (generate_entry_out_of_ssa_copies): Simplify.
8428 * graphite-sese-to-poly.c (extract_affine_name): Simplify
8429 by passing in ISL dimension.
8430 (parameter_index_in_region_1): Rename to ...
8431 (parameter_index_in_region): ... this.
8432 (extract_affine): Adjust assert, pass down parameter index.
8433 (add_param_constraints): Use range-info when available.
8434 (build_scop_context): Adjust.
8435 * sese.c (new_sese_info): Adjust.
8436 (free_sese_info): Likewise.
8437 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
8438 Remove unused typedefs.
8439 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
8441 2017-10-18 Martin Liska <mliska@suse.cz>
8443 * combine.c (simplify_compare_const): Add gcc_fallthrough.
8445 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8447 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
8448 (s390_sched_init): Do not reset s390_sched_state if we entered the
8449 current basic block via a fallthru edge and all others are unlikely.
8451 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
8453 * config/s390/s390.c (NUM_SIDES): New variable.
8454 (LONGRUNNING_THRESHOLD): New variable.
8455 (LATENCY_FACTOR): New variable.
8456 (s390_sched_score): Decrease score for long-running instructions on
8458 (s390_sched_variable_issue): Perform bookkeeping for long-running
8461 2017-10-18 Richard Biener <rguenther@suse.de>
8463 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
8464 Simplify with removal of the parameter rename map.
8465 (set_rename): Likewise.
8466 (should_copy_to_new_region): Likewise.
8467 (graphite_copy_stmts_from_block): Likewise.
8468 (copy_bb_and_scalar_dependences): Remove initialization of
8469 unused copied_bb_map.
8471 (copy_internal_parameters): Likewise.
8472 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
8473 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
8474 Use INTEGRAL_TYPE_P.
8475 (parameter_index_in_region_1): Rename to ...
8476 (assign_parameter_index_in_region): ... this. Assert we have
8477 a parameter we handle.
8478 (scan_tree_for_params): Adjust.
8479 * sese.h (parameter_rename_map_t): Remove.
8480 (struct sese_info_t): Remove unused parameter_rename_map and
8481 copied_bb_map members.
8482 * sese.c (new_sese_info): Adjust.
8483 (free_sese_info): Likewise.
8485 2017-10-18 Martin Liska <mliska@suse.cz>
8488 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
8489 on an abnormal edge.
8491 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
8493 * doc/invoke.texi (ffunction-sections and fdata-sections):
8496 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8498 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
8499 the use statement can throw internally.
8501 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
8503 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
8504 any RTX present on the RHS of a SET.
8505 * compare-elim.c (try_eliminate_compare): Restore comment.
8507 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8509 * langhooks.h (struct lang_hooks): Document that tree_size langhook
8510 may be also called on tcc_type nodes.
8511 * langhooks.c (lhd_tree_size): Likewise.
8513 2017-10-17 David Malcolm <dmalcolm@redhat.com>
8515 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
8516 format_warning_at_substring.
8517 (maybe_warn): Convert source_range * param to a location_t. Pass
8518 UNKNOWN_LOCATION rather than NULL to fmtwarn.
8519 (format_directive): Remove code to extract source_ranges and
8520 source_range * in favor of just a location_t.
8521 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
8523 * substring-locations.c (format_warning_va): Convert
8524 source_range * param to a location_t.
8525 (format_warning_at_substring): Likewise.
8526 * substring-locations.h (format_warning_va): Likewise.
8527 (format_warning_at_substring): Likewise.
8529 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
8531 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
8533 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
8534 and scatter/gather ops.
8536 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
8537 vec_gather_load and vec_scatter_store.
8538 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
8539 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
8540 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
8541 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
8542 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
8543 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
8545 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
8547 * reg-stack.c (compare_for_stack_reg): Add bool argument.
8548 Detect FTST instruction and handle its register pops. Only pop
8549 second operand if can_pop_second_op is true.
8550 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
8551 set can_pop_second_op to false in the compare_for_stack_reg call.
8553 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
8554 output_fp_compare for stack register operands.
8555 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
8556 instructions here. Do not emit stack register pops here. Assert
8557 that FCOMPP pops next to top stack register. Rewrite function.
8559 2017-10-17 Nathan Sidwell <nathan@acm.org>
8562 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
8563 use DECL_ASSEMBLER_NAME_RAW.
8566 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
8569 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
8570 Wilco Dijkstra <wilco.dijkstra@arm.com>
8572 * builtins.c (expand_builtin_update_setjmp_buf): Add a
8573 converstion to Pmode from the buf_addr.
8575 2017-10-17 Richard Biener <rguenther@suse.de>
8577 * graphite-dependences.c (scop_get_reads_and_writes): Change
8578 output parameters to references.
8580 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
8582 PR 71026/tree-optimization
8583 * fold-const.c (distribute_real_division): Removed.
8584 (fold_binary_loc): Remove calls to distribute_real_divison.
8586 2017-10-17 Richard Biener <rguenther@suse.de>
8588 * graphite-scop-detection.c
8589 (scop_detection::stmt_has_simple_data_refs_p): Always use
8590 the full nest as region.
8591 (try_generate_gimple_bb): Likewise.
8592 * sese.c (scalar_evolution_in_region): Simplify now that
8593 SCEV can handle instantiation in regions.
8594 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
8595 in the non-loop part of a function if requested.
8597 2017-10-17 Richard Biener <rguenther@suse.de>
8599 PR tree-optimization/82563
8600 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
8602 (graphite_regenerate_ast_isl): Call it.
8603 * graphite-scop-detection.c (build_scops): Remove entry edge split.
8605 2017-10-17 Jakub Jelinek <jakub@redhat.com>
8607 PR tree-optimization/82549
8608 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
8609 Formatting fixes. Instead of calling make_bit_field_ref with negative
8612 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
8614 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
8615 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
8616 _mm_maskz_reduce_ss): New.
8617 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
8618 __builtin_ia32_reducess_mask): Ditto..
8619 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
8620 * config/i386/sse.md (reduces<mode>): Renamed to ...
8621 (reduces<mode><mask_scalar_name>): ... this.
8622 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
8624 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
8625 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
8627 2017-10-16 David Malcolm <dmalcolm@redhat.com>
8629 * Makefile.in (OBJS): Add unique-ptr-tests.o.
8630 * selftest-run-tests.c (selftest::run_tests): Call
8631 selftest::unique_ptr_tests_cc_tests.
8632 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
8633 * unique-ptr-tests.cc: New file.
8635 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
8638 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8640 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8641 (make_hard_regno_born, make_hard_regno_dead): Update
8642 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
8644 2017-10-16 Jeff Law <law@redhat.com>
8646 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8648 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
8650 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
8652 2017-10-16 Olivier Hainque <hainque@adacore.com>
8654 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
8655 with_cpu if we were configured for an e500v2 target cpu name.
8657 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
8659 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
8660 * doc/invoke.texi: Document +nodsp as a valid extension for
8663 2017-10-16 Martin Liska <mliska@suse.cz>
8665 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
8666 (test_set_range): Likewise.
8667 (test_range_functions): Rename to ...
8668 (test_bit_in_range): ... this.
8669 (sbitmap_c_tests): Add new test.
8671 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8673 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
8675 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
8676 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
8678 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8680 * config/aarch64/aarch64-builtins.c
8681 (aarch64_types_quadopu_lane_qualifiers): New.
8682 (TYPES_QUADOPU_LANE): New.
8683 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
8684 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
8685 (aarch64_<sur>dot_laneq<vsi2qi>): New.
8686 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
8687 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
8688 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
8689 (Vdottype, DOTPROD): New.
8690 (sur): Add SDOT and UDOT.
8692 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8694 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
8695 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
8696 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
8698 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
8699 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
8700 Enable TARGET_DOTPROD.
8701 (cortex-a75.cortex-a55): Likewise.
8702 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
8704 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8706 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
8707 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
8709 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
8711 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
8712 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
8713 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
8714 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
8715 * config/arm/types.md (neon_dot, neon_dot_q): New.
8716 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
8718 2017-10-16 Tamar Christina <tamar.christina@arm.com>
8720 * config/arm/arm.h (TARGET_DOTPROD): New.
8721 * config/arm/arm.c (arm_arch_dotprod): New.
8722 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
8723 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
8724 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
8725 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
8726 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
8727 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
8728 * doc/invoke.texi (armv8.2-a): Document dotprod
8730 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8732 * i386.c (ix86_vec_cost): New function.
8733 (ix86_rtx_costs): Handle vector operations better.
8734 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
8735 * x86-tune-costs.h: Add new costs to all tables.
8737 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
8739 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
8741 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
8742 divsd, sqrtss and sqrtsd
8743 * x86-tune-costs.h: Add new entries to all costs.
8744 (znver1_cost): Fix to match real instruction latencies.
8746 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8747 Michael Collison <michael.collison@arm.com>
8749 * compare-elim.c: Include emit-rtl.h.
8750 (can_merge_compare_into_arith): New function.
8751 (try_validate_parallel): Likewise.
8752 (try_merge_compare): Likewise.
8753 (try_eliminate_compare): Call the above when no previous clobber
8755 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
8758 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8762 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
8763 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
8767 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
8768 to be any operand_equal_p operands. For & (B - 1) require
8769 B to be power of 2. Recognize
8770 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
8772 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8775 * optabs.c (expand_memory_blockage): Fix call of
8776 targetm.have_memory_blockage.
8778 2017-10-14 Jakub Jelinek <jakub@redhat.com>
8781 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
8782 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
8783 objects to extra_objs instead of overwriting it.
8785 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8787 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
8788 Use any_fp_register_operand as operand[3] predicate. Simplify
8789 equality test for operands[2] and operands[4] memory location.
8790 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
8791 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
8792 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
8793 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
8794 any_fp_register_operand as operand[1] predicate. Simplify
8795 equality test for operands[0] and operands[3] memory location.
8796 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
8797 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
8798 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
8800 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
8802 * target-insns.def: Add memory_blockage.
8803 * optabs.c (expand_memory_blockage): New function.
8804 (expand_asm_memory_barrier): Rename ...
8805 (expand_asm_memory_blockage): ... to this.
8806 (expand_mem_thread_fence): Call expand_memory_blockage
8807 instead of expand_asm_memory_barrier.
8808 (expand_mem_singnal_fence): Ditto.
8809 (expand_atomic_load): Ditto.
8810 (expand_atomic_store): Ditto.
8811 * doc/md.texi (Standard Pattern Names For Generation):
8812 Document memory_blockage instruction pattern.
8814 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
8816 * config/rl78/rl78.c (rl78_emit_libcall): New function.
8817 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
8818 * config/rl78/rl78.md: New define_expand "adddi3".
8820 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8822 * cfghooks.c (verify_flow_info): Disable check that all probabilities
8825 2017-10-13 Jeff Law <law@redhat.com>
8827 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
8829 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8832 * internal-fn.c (expand_mul_overflow): If both operands have
8833 the same highpart of -1 or 0 and the topmost bit of lowpart
8834 is different, overflow is if res <= 0 rather than res < 0.
8836 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
8838 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
8839 TARGET_P9_VECTOR code for unaligned_load case.
8841 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
8843 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
8845 2017-10-13 Nathan Sidwell <nathan@acm.org>
8847 * tree-core.h (tree_contains_struct): Make bool.
8848 * tree.c (tree_contains_struct): Likewise.
8849 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
8850 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
8851 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
8852 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
8853 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
8855 2017-10-13 Richard Biener <rguenther@suse.de>
8857 * graphite-isl-ast-to-gimple.c
8858 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
8859 parameters and dominance check.
8860 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
8861 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
8862 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
8863 Do not update SSA form here or do intermediate IL verification.
8864 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
8865 (graphite_initialize): Remove check on the number of loops in
8866 the function and inline into graphite_transform_loops.
8867 (graphite_finalize): Inline into graphite_transform_loops.
8868 (graphite_transform_loops): Perform SSA update and IL verification
8870 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
8872 2017-10-13 Richard Biener <rguenther@suse.de>
8874 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
8875 graphite_expression_type_precision): Avoid global constructor
8877 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
8878 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
8879 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
8880 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
8881 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
8882 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
8884 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8887 * config/i386/i386.h (ix86_red_zone_size): New.
8888 * config/i386/i386.md (push peephole2s): Replace
8889 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8891 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8892 Alan Hayward <alan.hayward@arm.com>
8893 David Sherwood <david.sherwood@arm.com>
8895 * combine.c (can_change_dest_mode): Reject changes in
8896 REGMODE_NATURAL_SIZE.
8898 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8899 Alan Hayward <alan.hayward@arm.com>
8900 David Sherwood <david.sherwood@arm.com>
8902 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8903 (expand_debug_source_expr): Likewise.
8904 * combine.c (combine_simplify_rtx): Likewise.
8905 * cse.c (fold_rtx): Likewise.
8906 * fwprop.c (canonicalize_address): Likewise.
8907 * targhooks.c (default_shift_truncation_mask): Likewise.
8909 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8910 Alan Hayward <alan.hayward@arm.com>
8911 David Sherwood <david.sherwood@arm.com>
8913 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8914 (widened_mode): Likewise.
8915 (expand_unop): Likewise.
8916 * ree.c (transform_ifelse): Likewise.
8917 (merge_def_and_ext): Likewise.
8918 (combine_reaching_defs): Likewise.
8919 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8921 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8922 Alan Hayward <alan.hayward@arm.com>
8923 David Sherwood <david.sherwood@arm.com>
8925 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8926 * combine.c (gen_lowpart_for_combine): Likewise.
8927 * dwarf2out.c (rtl_for_decl_location): Likewise.
8928 * final.c (alter_subreg): Likewise.
8929 * rtlhooks.c (gen_lowpart_general): Likewise.
8930 (gen_lowpart_if_possible): Likewise.
8932 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8933 Alan Hayward <alan.hayward@arm.com>
8934 David Sherwood <david.sherwood@arm.com>
8936 * calls.c (expand_call): Use subreg_lowpart_offset.
8937 * cse.c (cse_insn): Likewise.
8938 * regcprop.c (copy_value): Likewise.
8939 (copyprop_hardreg_forward_1): Likewise.
8941 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8944 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8945 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8946 =Q constraints to +Q and into insn condition add check
8947 that operands[0] and operands[1] are equal.
8948 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8949 =Q constraints to +Q and into insn condition add check
8950 that operands[0] is equal to either operands[1] or operands[2].
8953 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8954 instead of handling MINUS_EXPR twice (once for each argument),
8955 canonicalize operand order and handle just once, use rtype where
8956 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8959 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8960 any values of __C while still being pattern recognizable as a simple
8963 2017-10-13 Richard Biener <rguenther@suse.de>
8965 PR tree-optimization/82451
8967 2017-10-02 Richard Biener <rguenther@suse.de>
8969 PR tree-optimization/82355
8970 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8971 a mapping for the enclosing loop but avoid generating one for
8973 (copy_bb_and_scalar_dependences): Remove premature codegen
8974 error on PHIs in blocks duplicated into multiple places.
8975 * graphite-scop-detection.c
8976 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8977 in the region use it as loop and nest to analyze the DR in.
8978 (try_generate_gimple_bb): Likewise.
8979 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8980 (add_loop_constraints): For blocks in a loop not in the region
8981 create a dimension with a single iteration.
8982 * sese.h (gbb_loop_at_index): Remove assert.
8984 * cfgloop.c (loop_preheader_edge): For the loop tree root
8985 return the single successor of the entry block.
8986 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8987 Reset the SCEV hashtable and niters.
8988 * graphite-scop-detection.c
8989 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8990 assert that we only have POLYNOMIAL_CHREC that vary in loops
8991 contained in the region.
8992 (scop_detection::graphite_can_represent_expr): Adjust.
8993 (scop_detection::stmt_has_simple_data_refs_p): For loops
8994 not in the region set loop to NULL. The nest is now the
8995 entry edge to the region.
8996 (try_generate_gimple_bb): Likewise.
8997 * sese.c (scalar_evolution_in_region): Adjust for
8998 instantiate_scev change.
8999 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9000 Make nest parameter the edge into the region.
9001 (create_data_ref): Likewise.
9002 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9003 entry edge into a region and adjust instantiate_scev calls.
9004 (create_data_ref): Likewise.
9005 (graphite_find_data_references_in_stmt): Likewise.
9006 (find_data_references_in_stmt): Pass the loop preheader edge
9007 from the nest argument.
9008 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9009 parameter the edge into the region.
9010 (instantiate_parameters): Use the loop preheader edge as entry.
9011 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9013 (get_instantiated_value_entry): Make instantiate_below parameter
9014 the edge into the region.
9015 (instantiate_scev_name): Likewise. Adjust dominance checks,
9016 when we cannot use loop-based instantiation instantiate by
9017 walking use-def chains.
9018 (instantiate_scev_poly): Adjust.
9019 (instantiate_scev_binary): Likewise.
9020 (instantiate_scev_convert): Likewise.
9021 (instantiate_scev_not): Likewise.
9022 (instantiate_array_ref): Remove.
9023 (instantiate_scev_3): Likewise.
9024 (instantiate_scev_2): Likewise.
9025 (instantiate_scev_1): Likewise.
9026 (instantiate_scev_r): Do not blindly handle N-operand trees.
9027 Do not instantiate array-refs. Handle all constants and invariants.
9028 (instantiate_scev): Make instantiate_below parameter
9029 the edge into the region.
9030 (resolve_mixers): Use the loop preheader edge for the region
9031 parameter to instantiate_scev_r.
9032 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9034 2017-10-13 Richard Biener <rguenther@suse.de>
9036 PR tree-optimization/82525
9037 * graphite-isl-ast-to-gimple.c
9038 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9040 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9041 Fail code generation when we cannot represent the isl integer.
9042 (binary_op_to_tree): Elide modulo operations that are no-ops
9043 in the type we code generate. Remove now superfluous code
9046 2017-10-13 Richard Biener <rguenther@suse.de>
9048 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9049 (scop_detection::harmful_loop_in_region): Remove premature
9050 IV type restriction.
9051 (scop_detection::graphite_can_represent_scev): We can handle
9052 pointer IVs just fine.
9054 2017-10-13 Alan Modra <amodra@gmail.com>
9056 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9057 "Clobbers and Scratch Registers". Add paragraph on
9058 alternative to clobbers for scratch registers and OpenBLAS
9061 2017-10-13 Alan Modra <amodra@gmail.com>
9063 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9064 example of a memory input for a string of known length. Move
9065 commentary out of table. Add a number of new examples
9066 covering array memory inputs.
9068 2017-10-12 Martin Liska <mliska@suse.cz>
9070 PR tree-optimization/82493
9071 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9072 (test_range_functions): New function.
9073 (sbitmap_c_tests): Likewise.
9074 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9075 * selftest.h (sbitmap_c_tests): New function.
9077 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9079 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9081 * config/rs6000/amo.h: Fix spacing issue.
9083 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9086 * config/i386/i386.md (*ashl<mode>3_mask_1,
9087 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9088 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9091 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9093 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9094 (profile_probability): Set max_probability
9095 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9097 * profile-count.c (profile_probability::differs_from_p): Do not
9098 rely on max_probaiblity == 10000
9100 2017-10-12 Jeff Law <law@redhat.com>
9102 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9105 2017-10-12 Martin Sebor <msebor@redhat.com>
9109 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9110 (handle_alias_pairs): Call it.
9111 * common.opt (-Wattribute-alias): New option.
9112 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9113 * doc/invoke.texi (-Wattribute-alias): Document.
9115 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9118 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9120 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9122 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9123 (make_hard_regno_born, make_hard_regno_dead): Update
9124 bb_killed_pseudos and bb_gen_pseudos.
9126 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9128 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9130 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9132 * config/alpha/alpha.c (alpha_split_conditional_move):
9133 Use std::swap instead of manually swapping.
9134 (alpha_stdarg_optimize_hook): Ditto.
9135 (alpha_canonicalize_comparison): Ditto.
9137 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9139 * tree-loop-distribution.c (struct builtin_info): New struct.
9140 (struct partition): Refactor fields into struct builtin_info.
9141 (partition_free): Free struct builtin_info.
9142 (build_size_arg_loc, build_addr_arg_loc): Delete.
9143 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9144 information from struct builtin_info.
9145 (find_single_drs): New function refactored from classify_partition.
9146 Also moved builtin validity checks to this function.
9147 (compute_access_range, alloc_builtin): New functions.
9148 (classify_builtin_st, classify_builtin_ldst): New functions.
9149 (classify_partition): Refactor code into functions find_single_drs,
9150 classify_builtin_st and classify_builtin_ldst.
9151 (distribute_loop): Don't do runtime alias check when distributing
9153 (find_seed_stmts_for_distribution): New function.
9154 (pass_loop_distribution::execute): Refactor code finding seed
9155 stmts into above function. Support distribution for the innermost
9156 two-level loop nest. Adjust dump information.
9158 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9160 * tree-loop-distribution.c: Adjust the general comment.
9161 (NUM_PARTITION_THRESHOLD): New macro.
9162 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9163 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9164 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9165 in call to build_partition_graph.
9166 (finalize_partitions): New parameter. Make loop distribution more
9167 conservative by fusing more partitions.
9168 (distribute_loop): Don't do runtime alias check in case of loop nest
9170 (find_seed_stmts_for_distribution): New function.
9171 (prepare_perfect_loop_nest): New function.
9172 (pass_loop_distribution::execute): Refactor code finding seed stmts
9173 and loop nest into above functions. Support loop nest distribution.
9174 Adjust dump information accordingly.
9176 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9178 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9179 and set PTYPE_SEQUENTIAL for merged partition.
9181 2017-10-12 Richard Biener <rguenther@suse.de>
9183 PR tree-optimization/69728
9185 2017-09-19 Richard Biener <rguenther@suse.de>
9187 PR tree-optimization/69728
9188 * graphite-sese-to-poly.c (schedule_error): New global.
9189 (add_loop_schedule): Handle empty domain by failing the
9191 (build_original_schedule): Handle schedule_error.
9193 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9194 domain by returning an unchanged schedule.
9196 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9198 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9199 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9201 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9203 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9206 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9209 * expr.c (store_field): Don't optimize away bitsize == 0 store
9210 from CALL_EXPR with addressable return type.
9212 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9214 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9215 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9216 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9217 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9218 TARGET_ISEL instead of TARGET_ISEL<sel>.
9220 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9222 * config/rs6000/rs6000.c
9223 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9225 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9227 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9229 (reg_or_cint_operand): Fix comment.
9230 (reg_or_zero_operand): New predicate.
9231 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9232 * config/rs6000/rs6000.c (output_isel): Delete.
9233 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9234 instead of reg_or_cint_operand. Output instruction directly (not via
9236 (isel_unsigned_<mode>): Ditto.
9237 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9238 gpc_reg_operand. Add an instruction alternative for this. Output
9239 instruction directly.
9240 (*isel_reversed_unsigned_<mode>): Ditto.
9242 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9244 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9245 (TARGET_CANONICALIZE_COMPARISON): Define.
9247 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
9250 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
9251 Check whether the dest is REG before adding REG_EQUIV note.
9253 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9256 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9258 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9259 (make_hard_regno_born, make_hard_regno_dead): Update
9260 bb_killed_pseudos and bb_gen_pseudos.
9262 2017-10-11 Nathan Sidwell <nathan@acm.org>
9264 * incpath.h (enum incpath_kind): Name enum, prefix values.
9265 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
9266 * incpath.c (heads, tails): Use INC_MAX.
9267 (add_env_var_paths, add_standard_paths): Use incpath_kind.
9268 (merge_include_chains, split_quote_chain,
9269 register_include_chains): Update incpath_kind names.
9270 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
9271 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
9273 (add_framework_path, darwin_register_objc_includes): Likewise.
9274 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
9276 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9278 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
9279 Do not use float_operator operator predicate.
9280 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
9281 * config/i386/predicates.md (float_operator): Remove predicate.
9283 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9285 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
9286 (*jccxf_i387): Ditto.
9287 (*jcc<mode>_i387): Ditto.
9288 (*jccu<mode>_i387): Ditto.
9289 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
9290 (*jcc_*_i387 splitters): Remove.
9291 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
9292 * config/i386/i386.c (ix86_split_fp_branch): Remove.
9293 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
9296 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9298 * profile-count.h (slow_safe_scale_64bit): New function.
9299 (safe_scale_64bit): New inline.
9300 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
9301 * profile-count.c: Include wide-int.h
9302 (slow_safe_scale_64bit): New.
9304 2017-10-11 Nathan Sidwell <nathan@acm.org>
9306 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
9307 HAS_DECL_ASSEMBLER_NAME_P.
9308 * gimple-expr.c (gimple_decl_printable_name: Check
9309 HAS_DECL_ASSEMBLER_NAME_P too.
9310 * ipa-utils.h (type_in_anonymous_namespace_p): Check
9311 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
9312 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
9313 * passes.c (rest_of_decl_compilation): Check
9314 HAS_DECL_ASSEMBLER_NAME_P too.
9315 * recog.c (verify_changes): Likewise.
9316 * tree-pretty-print.c (dump_decl_name): Likewise.
9317 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
9319 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
9320 (SET_DECL_ASSEMBLER_NAME): Use it.
9321 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
9322 (COPY_DECL_ASSEMBLER_NAME): Likewise.
9323 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
9325 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
9327 * config.gcc (i386, x86_64): Add extra objects.
9328 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
9329 (ix86_min_insn_size): Declare.
9330 (ix86_issue_rate): Declare.
9331 (ix86_adjust_cost): Declare.
9332 (ia32_multipass_dfa_lookahead): Declare.
9333 (ix86_macro_fusion_p): Declare.
9334 (ix86_macro_fusion_pair_p): Declare.
9335 (ix86_bd_has_dispatch): Declare.
9336 (ix86_bd_do_dispatch): Declare.
9337 (ix86_core2i7_init_hooks): Declare.
9338 (ix86_atom_sched_reorder): Declare.
9339 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
9340 (COSTS_N_BYTES): Move to x86-tune-costs.h.
9341 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
9342 (rip_relative_addr_p): Rename to ...
9343 (ix86_rip_relative_addr_p): ... this one; export.
9344 (memory_address_length): Update.
9345 (ix86_issue_rate): Move to x86-tune-sched.c.
9346 (ix86_flags_dependent): Move to x86-tune-sched.c.
9347 (ix86_agi_dependent): Move to x86-tune-sched.c.
9348 (exact_dependency_1): Move to x86-tune-sched.c.
9349 (exact_store_load_dependency): Move to x86-tune-sched.c.
9350 (ix86_adjust_cost): Move to x86-tune-sched.c.
9351 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
9352 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
9353 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
9354 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
9355 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
9356 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
9357 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
9358 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
9359 (min_insn_size): Rename to ...
9360 (ix86_min_insn_size): ... this one; export.
9361 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
9362 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
9363 (core2i7_first_cycle_multipass_backtrack): Move to
9364 x86-tune-sched-core.c.
9365 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
9366 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
9367 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
9368 (ix86_avoid_jump_mispredicts): Update.
9369 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
9370 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
9371 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
9372 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
9373 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
9374 (MAX_INSN): Move to ix86-tune-sched-bd.c.
9375 (MAX_IMM): Move to ix86-tune-sched-bd.c.
9376 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
9377 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
9378 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
9379 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
9380 (MAX_STORE): Move to ix86-tune-sched-bd.c.
9381 (BIG): Move to ix86-tune-sched-bd.c.
9382 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
9383 (enum insn_path): Move to ix86-tune-sched-bd.c.
9384 (get_mem_group): Move to ix86-tune-sched-bd.c.
9385 (is_cmp): Move to ix86-tune-sched-bd.c.
9386 (dispatch_violation): Move to ix86-tune-sched-bd.c.
9387 (is_branch): Move to ix86-tune-sched-bd.c.
9388 (is_prefetch): Move to ix86-tune-sched-bd.c.
9389 (init_window): Move to ix86-tune-sched-bd.c.
9390 (allocate_window): Move to ix86-tune-sched-bd.c.
9391 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
9392 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
9393 (process_end_window): Move to ix86-tune-sched-bd.c.
9394 (allocate_next_window): Move to ix86-tune-sched-bd.c.
9395 (find_constant): Move to ix86-tune-sched-bd.c.
9396 (get_num_immediates): Move to ix86-tune-sched-bd.c.
9397 (has_immediate): Move to ix86-tune-sched-bd.c.
9398 (get_insn_path): Move to ix86-tune-sched-bd.c.
9399 (get_insn_group): Move to ix86-tune-sched-bd.c.
9400 (count_num_restricted): Move to ix86-tune-sched-bd.c.
9401 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
9402 (add_insn_window): Move to ix86-tune-sched-bd.c.
9403 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
9404 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
9405 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
9406 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
9407 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
9408 (do_dispatch): Move to ix86-tune-sched-bd.c.
9409 (has_dispatch): Move to ix86-tune-sched-bd.c.
9410 * config/i386/t-i386: Add new object files.
9411 * config/i386/x86-tune-costs.h: New file.
9412 * config/i386/x86-tune-sched-atom.c: New file.
9413 * config/i386/x86-tune-sched-bd.c: New file.
9414 * config/i386/x86-tune-sched-core.c: New file.
9415 * config/i386/x86-tune-sched.c: New file.
9417 2017-10-11 Liu Hao <lh_mouse@126.com>
9419 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
9420 the generic version below instead.
9421 (should_colorize): Recognize Windows consoles as terminals
9423 * pretty-print.c [__MINGW32__] (write_all): New function.
9424 [__MINGW32__] (find_esc_head): Likewise.
9425 [__MINGW32__] (find_esc_terminator): Likewise.
9426 [__MINGW32__] (eat_esc_sequence): Likewise.
9427 [__MINGW32__] (mingw_ansi_fputs): New function that handles
9429 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
9432 2017-10-11 Richard Biener <rguenther@suse.de>
9434 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
9435 Properly call analyze_scalar_evolution with the loop of the stmt.
9437 2017-10-11 Richard Biener <rguenther@suse.de>
9439 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
9440 * tree-core.h (tree_base): Add chrec_var union member.
9441 * tree.h (CHREC_VAR): Remove.
9442 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
9443 * tree-chrec.h (build_polynomial_chrec): Adjust.
9444 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
9445 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
9447 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
9449 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
9450 * match.pd: ... here.
9451 ((T) X == (T) Y): Relax condition.
9453 2017-10-11 Bin Cheng <bin.cheng@arm.com>
9455 PR tree-optimization/82472
9456 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
9458 (break_alias_scc_partitions): Update postorder number.
9460 2017-10-11 Martin Liska <mliska@suse.cz>
9463 * opts.c (parse_no_sanitize_attribute): Do not use error_value
9465 * opts.h (parse_no_sanitize_attribute): Remove last argument.
9467 2017-10-11 Martin Liska <mliska@suse.cz>
9469 * print-rtl.c (print_insn): Move declaration of idbuf
9470 to same scope as name.
9472 2017-10-11 Martin Liska <mliska@suse.cz>
9477 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9480 2017-10-11 Martin Liska <mliska@suse.cz>
9483 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
9486 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
9488 * config.gcc (default_gnu_indirect_function): Default to yes for
9489 arm*-*-linux* with glibc.
9491 2017-10-11 Richard Biener <rguenther@suse.de>
9493 * tree-scalar-evolution.c (get_scalar_evolution): Handle
9494 default-defs and types we do not want to analyze.
9495 (interpret_loop_phi): Replace unreachable code with an assert.
9496 (compute_scalar_evolution_in_loop): Remove and inline ...
9497 (analyze_scalar_evolution_1): ... here, replacing condition with
9498 what makes the intent clearer. Remove handling of cases
9499 get_scalar_evolution now handles.
9501 2017-10-10 Jim Wilson <wilson@tuliptree.org>
9503 PR rtl-optimization/81434
9504 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
9505 comment for main loop. In sched_group_found if, also add checks for
9506 pass and min_cost_group.
9508 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
9510 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
9511 (rs6000_insn_cost): New function.
9512 * config/rs6000/rs6000.md (cost): New attribute.
9514 2017-10-10 Jakub Jelinek <jakub@redhat.com>
9515 H.J. Lu <hongjiu.lu@intel.com>
9519 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
9520 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
9521 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
9522 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
9523 to OPTION_MASK_ISA_AVX512VL - builtins that have both
9524 OPTION_MASK_ISA_MMX and some other bit set require both
9525 mmx and the ISAs without the mmx bit.
9526 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
9527 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
9528 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
9529 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
9530 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
9531 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
9532 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
9533 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
9534 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
9535 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
9536 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
9537 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
9538 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
9539 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
9540 Add OPTION_MASK_ISA_MMX.
9542 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
9544 * config.gcc (armv7*-*-freebsd*): New target.
9545 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
9547 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
9549 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
9550 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
9553 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
9555 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
9556 a template parameter.
9557 (WIDE_INT_REF_FOR): Update accordingly.
9558 * tree.h (wi::int_traits <const_tree>): Delete.
9559 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
9560 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
9561 (wi::tree_to_wide_ref): New typedef.
9562 (wi::to_wide): New function.
9563 * calls.c (get_size_range): Use wi::to_wide when operating on
9565 * cgraph.c (cgraph_node::create_thunk): Likewise.
9566 * config/i386/i386.c (ix86_data_alignment): Likewise.
9567 (ix86_local_alignment): Likewise.
9568 * dbxout.c (stabstr_O): Likewise.
9569 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
9570 * expr.c (const_vector_from_tree): Likewise.
9571 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
9572 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
9573 (fold_negate_expr_1, int_const_binop_1, const_binop)
9574 (fold_convert_const_int_from_real, optimize_bit_field_compare)
9575 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
9576 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
9577 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
9578 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
9579 (fold_not_const, round_up_loc): Likewise.
9580 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
9581 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
9582 (alloca_call_type): Likewise.
9583 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
9584 * godump.c (go_output_typedef): Likewise.
9585 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
9586 * internal-fn.c (get_min_precision): Likewise.
9587 * ipa-cp.c (ipcp_store_vr_results): Likewise.
9588 * ipa-polymorphic-call.c
9589 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
9590 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
9591 (ipa_modify_call_arguments): Likewise.
9592 * match.pd: Likewise.
9593 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
9594 * print-tree.c (print_node_brief, print_node): Likewise.
9595 * stmt.c (expand_case): Likewise.
9596 * stor-layout.c (layout_type): Likewise.
9597 * tree-affine.c (tree_to_aff_combination): Likewise.
9598 * tree-cfg.c (group_case_labels_stmt): Likewise.
9599 * tree-data-ref.c (dr_analyze_indices): Likewise.
9600 (prune_runtime_alias_test_list): Likewise.
9601 * tree-dump.c (dequeue_and_dump): Likewise.
9602 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
9603 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
9604 * tree-pretty-print.c (dump_generic_node): Likewise.
9605 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
9606 (simple_iv_with_niters): Likewise.
9607 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
9608 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
9609 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
9610 * tree-ssa-loop-niter.c (split_to_var_and_offset)
9611 (refine_value_range_using_guard, number_of_iterations_ne_max)
9612 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
9613 (get_cst_init_from_scev, record_nonwrapping_iv)
9614 (scev_var_range_cant_overflow): Likewise.
9615 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
9616 * tree-ssa-pre.c (compute_avail): Likewise.
9617 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
9618 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
9619 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
9620 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
9621 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
9622 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
9623 (dump_case_nodes, try_switch_expansion): Likewise.
9624 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
9625 (vect_do_peeling): Likewise.
9626 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
9627 * tree-vect-stmts.c (vectorizable_load): Likewise.
9628 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
9629 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
9630 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
9631 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
9632 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
9633 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
9634 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
9635 (evrp_dom_walker::before_dom_children): Likewise.
9636 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
9637 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
9638 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
9639 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
9640 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
9641 (get_range_pos_neg): Likewise.
9642 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
9643 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
9644 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
9645 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
9646 * config/avr/avr.c (avr_fold_builtin): Likewise.
9647 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
9648 * config/msp430/msp430.c (msp430_attr): Likewise.
9649 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
9650 * config/powerpcspe/powerpcspe-c.c
9651 (altivec_resolve_overloaded_builtin): Likewise.
9652 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
9653 (rs6000_expand_ternop_builtin): Likewise.
9654 * config/rs6000/rs6000-c.c
9655 (altivec_resolve_overloaded_builtin): Likewise.
9656 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
9657 (rs6000_expand_ternop_builtin): Likewise.
9658 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
9660 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9662 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
9663 when copying loop nest with only one inner loop.
9665 2017-10-10 Richard Biener <rguenther@suse.de>
9667 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9668 blocks if SCEV is active.
9669 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
9671 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
9672 (scev_initialize): Assert we are not yet initialized.
9674 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9676 * tree-loop-distribution.c (generate_loops_for_partition): Remove
9677 inner loop's exit stmt by making it always exit the loop, otherwise
9678 we would generate an infinite empty loop.
9680 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9682 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
9683 renaming variables in new preheader if it's deleted.
9685 2017-10-10 Bin Cheng <bin.cheng@arm.com>
9687 * tree-loop-distribution.c (struct partition): Remove unused field
9688 loops of the structure.
9689 (partition_alloc, partition_free): Ditto.
9690 (build_rdg_partition_for_vertex): Ditto.
9692 2017-10-09 Jeff Law <law@redhat.com>
9694 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
9695 return type to match prototype and documentation.
9697 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9699 * config/rs6000/rs6000.c (processor_costs): Move to ...
9700 * config/rs6000/rs6000.h: ... here.
9701 (rs6000_cost): Declare.
9703 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
9705 * except.c (setjmp_fn): New global variable.
9706 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
9707 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
9708 if DONT_USE_BUILTIN_SETJMP is defined.
9710 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9712 * target.def (insn_cost): New hook.
9713 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
9714 * doc/tm.texi: Regenerate.
9715 * rtlanal.c (insn_cost): Use the new hook.
9717 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9719 * combine.c (combine_validate_cost): Compute the new insn_cost,
9720 not just pattern_cost.
9721 (try_combine): Adjust comment.
9723 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
9725 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
9727 * combine.c (uid_insn_cost): Adjust comment.
9728 (combine_validate_cost): Adjust comment. Use pattern_cost instead
9730 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
9731 * dse.c (find_shift_sequence): Ditto.
9732 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
9733 (bb_valid_for_noce_process_p): Use pattern_cost.
9734 * rtl.h (insn_rtx_cost): Delete.
9735 (pattern_cost): New prototype.
9736 (insn_cost): New prototype.
9737 * rtlanal.c (insn_rtx_cost): Rename to...
9738 (pattern_cost): ... this.
9741 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
9743 * config/i386/i386.md (*jcc_2): Remove insn pattern.
9744 (*jcc<mode>_0_r_i387): Ditto.
9745 (*jccxf_r_i387): Ditto.
9746 (*jcc<mode>_r_i387): Ditto.
9747 (*jccu<mode>_r_i387): Ditto.
9748 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
9749 (*jcc): Rename from *jcc_1.
9751 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9753 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
9754 deferred rescans after the lvx/stvx recombination pre-pass.
9756 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9758 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
9759 memory operation instruction support.
9760 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
9761 (rs6000-ibm-aix[789]*): Likewise.
9762 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
9763 Document new functions.
9765 2017-10-09 Richard Biener <rguenther@suse.de>
9767 PR tree-optimization/82397
9768 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
9769 equality only for semantically equal trees.
9771 2017-10-09 Richard Biener <rguenther@suse.de>
9773 PR tree-optimization/82449
9774 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
9775 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
9776 allow constant addresses.
9777 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
9780 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9782 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
9785 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9788 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
9791 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9794 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
9796 2017-10-09 Jakub Jelinek <jakub@redhat.com>
9799 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
9800 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
9802 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
9804 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
9805 (WI_BINARY_PREDICATE_RESULT): Likewise.
9806 (wi::binary_traits::operator_result): New type.
9807 (wi::binary_traits::predicate_result): Likewise.
9808 (generic_wide_int::operator~, unary generic_wide_int::operator-)
9809 (generic_wide_int::operator==, generic_wide_int::operator!=)
9810 (generic_wide_int::operator&, generic_wide_int::and_not)
9811 (generic_wide_int::operator|, generic_wide_int::or_not)
9812 (generic_wide_int::operator^, generic_wide_int::operator+
9813 (binary generic_wide_int::operator-, generic_wide_int::operator*):
9815 (operator~, unary operator-, operator==, operator!=, operator&)
9816 (operator|, operator^, operator+, binary operator-, operator*): New
9818 * expr.c (get_inner_reference): Use wi::bit_and_not.
9819 * fold-const.c (fold_binary_loc): Likewise.
9820 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
9821 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
9822 (bit_value_binop): Likewise.
9823 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
9824 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
9825 (extract_range_from_binary_expr_1): Likewise.
9826 (masked_increment): Likewise.
9827 (simplify_bit_ops_using_ranges): Likewise.
9829 2017-10-09 Martin Jambor <mjambor@suse.cz>
9832 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
9833 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
9834 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
9836 (hsa_fixup_mov_insn_type): New function.
9837 (hsa_op_with_type::get_in_type): Use it.
9838 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
9839 immediates in an assert.
9840 (hsa_op_with_type::extend_int_to_32bit): New method.
9841 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
9842 types. Convert to dest type if necessary.
9843 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
9844 (reg_for_gimple_ssa): Pass false as min32int to
9845 hsa_type_for_scalar_tree_type.
9846 (gen_hsa_addr): Fixup type when creating addresable temporary.
9847 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
9848 (gen_hsa_unary_operation): Extend operands and convert to dest type if
9849 necessary. Call hsa_fixup_mov_insn_type.
9850 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
9851 extend operands and convert to dest type if necessary.
9852 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
9853 to dest type if necessary.
9854 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
9855 if conversion nt necessary and size matches.
9856 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
9857 to dest type if necessary.
9858 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
9859 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
9861 (gen_hsa_clrsb): Likewise.
9862 (gen_hsa_ffs): Likewise.
9863 (gen_hsa_divmod): Extend operands and convert to dest type if
9865 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
9867 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9869 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
9870 Remove empty default arguments. Use a brace block as output
9872 (conditional return): Ditto.
9874 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
9875 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
9877 (group_ending_nop): Ditto.
9878 (doloop_end): Ditto.
9879 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
9880 (splitters for those): Ditto.
9882 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
9884 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
9885 a conditional jump (and the compare for it) so that pc_rtx is the
9887 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9888 for the deleted and renamed ctr<mode>_internal[234] patterns.
9889 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9890 Delete second conditional return pattern.
9891 (ctr<mode>_internal2): Delete this second bdnz pattern.
9892 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9893 (ctr<mode>_internal4): Delete this second bdz pattern.
9895 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9897 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9898 (always_initialized_rtx_for_ssa_name_p): New predicate.
9899 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9900 (finish_out_of_ssa): Free new field of SA.
9901 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9902 * tree-ssa-coalesce.c: Include tree-ssa.h.
9903 (get_parm_default_def_partitions): Remove extern keyword.
9904 (get_undefined_value_partitions): New function.
9905 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9906 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9908 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9910 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9912 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9914 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9916 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9917 for targets that preffer 128bit.
9919 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9921 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9923 2017-10-08 Olivier Hainque <hainque@adacore.com>
9925 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9926 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9928 (thumb_set_return_address): Likewise.
9930 2017-10-08 Olivier Hainque <hainque@adacore.com>
9932 * common/config/arm/arm-common.c (arm_except_unwind_info):
9933 Handle DWARF2_UNWIND_INFO.
9935 2017-10-07 Michael Collison <michael.collison@arm.com>
9937 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9940 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9942 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9943 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9944 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9945 defined, force the creation of a new block for a dispatch label.
9947 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9949 * invoke.texi (Wsuggest-attribute=cold): Document.
9950 * common.opt (Wsuggest-attribute=cold): New
9951 * ipa-pure-const.c (warn_function_cold): New function.
9952 * predict.c (compute_function_frequency): Use it.
9953 * predict.h (warn_function_cold): Declare.
9955 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9957 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9960 2017-10-06 Martin Liska <mliska@suse.cz>
9962 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9963 keyword for member functions.
9964 (struct sanopt_tree_couple): New struct.
9965 (struct sanopt_tree_couple_hash): New function.
9966 (struct sanopt_ctx): Add new hash_map.
9967 (has_dominating_ubsan_ptr_check): New function.
9968 (record_ubsan_ptr_check_stmt): Likewise.
9969 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9970 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9971 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9973 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9976 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9977 aarch64_simd_valid_immediate on CONST_VECTORs.
9978 (aarch64_reg_or_bic_imm): Likewise.
9980 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9982 PR rtl-optimization/82396
9983 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9985 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9987 * graphite-dependences.c (scop_get_reads): Move code to...
9988 (scop_get_must_writes): Move code to...
9989 (scop_get_may_writes): Move code to...
9990 (scop_get_reads_and_writes): ... here.
9991 (scop_get_dependences): Call scop_get_reads_and_writes.
9993 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9995 PR tree-optimization/82434
9996 * fold-const.h (can_native_encode_type_p,
9997 can_native_encode_string_p): Remove.
9998 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9999 don't encode anything, just return what would be otherwise returned.
10000 (native_encode_fixed, native_encode_complex, native_encode_vector):
10002 (native_encode_string): Likewise. Inline by hand
10003 can_native_encode_string_p.
10004 (can_native_encode_type_p): Remove.
10005 (can_native_encode_string_p): Remove.
10006 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10007 STRING_CSTs using can_native_encode_string_p, test all
10008 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10009 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10010 argument from native_encode_expr.
10011 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10012 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10015 2017-10-06 Richard Biener <rguenther@suse.de>
10017 PR tree-optimization/82397
10018 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10019 operand_equal_p but rely on data_ref_compare_tree for detecting
10021 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10022 to match up with dr_group_sort_cmp.
10024 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10027 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10029 * config/s390/s390-builtin-types.def: Regenerate.
10031 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10034 * config/s390/s390-builtin-types.def: Regenerate.
10035 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10036 Change flag from B_VXE to B_VX.
10037 (s390_vec_min_dbl): Remove B_VXE flag.
10039 2017-10-06 Richard Biener <rguenther@suse.de>
10041 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10042 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10043 translate_isl_ast_to_gimple::is_valid_rename,
10044 translate_isl_ast_to_gimple::get_rename,
10045 translate_isl_ast_to_gimple::get_def_bb_for_const,
10046 translate_isl_ast_to_gimple::get_new_name,
10047 translate_isl_ast_to_gimple::collect_all_ssa_names,
10048 translate_isl_ast_to_gimple::copy_loop_phi_args,
10049 translate_isl_ast_to_gimple::collect_all_ssa_names,
10050 translate_isl_ast_to_gimple::copy_loop_phi_args,
10051 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10052 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10053 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10054 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10055 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10056 translate_isl_ast_to_gimple::copy_cond_phi_args,
10057 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10058 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10059 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10060 translate_isl_ast_to_gimple::rename_uses,
10061 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10062 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10063 (set_rename_for_each_def): Likewise.
10064 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10065 here. Handle rewriting SCEV analyzable uses here.
10066 (copy_bb_and_scalar_dependences): Generate code for PHI
10068 (graphite_regenerate_ast_isl): Adjust.
10069 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10070 (add_write, add_read): New functions.
10071 (build_cross_bb_scalars_def): Use it and simplify.
10072 (build_cross_bb_scalars_use): Likewise.
10073 (graphite_find_cross_bb_scalar_vars): Inline into...
10074 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10075 simulating out-of-SSA. Compute liveout and add dependencies.
10076 (build_scops): Force an empty entry block.
10077 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10079 (sese_build_liveouts): Declare.
10080 (sese_trivially_empty_bb_p): Likewise.
10081 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10082 compute liveout and debug_liveout.
10083 (sese_bad_liveouts_use): Remove.
10084 (sese_reset_debug_liveouts_bb): Likewise.
10085 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10086 (sese_build_liveouts): Build liveout and debug_liveout and store
10088 (new_sese_info): Adjust.
10089 (free_sese_info): Likewise.
10090 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10091 do not build liveout here.
10092 (move_sese_in_condition): Adjust region entry.
10093 (scev_analyzable_p): Match up with chrec_apply requirements.
10094 (sese_trivially_empty_bb_p): New.
10095 * tree-into-ssa.c (get_reaching_def): Properly support generating
10096 default-defs for incremental rewrite of anonymous names.
10098 2017-10-06 Richard Biener <rguenther@suse.de>
10100 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10101 precision do not perform modulo reduction.
10103 2017-10-06 Richard Biener <rguenther@suse.de>
10105 PR tree-optimization/82436
10106 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10107 conservatively choose the vectorization factor when checking
10108 whether we can perform the required load permutation.
10109 (vect_transform_slp_perm_load): Assert when we may not fail.
10111 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10113 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10114 message for incompatible -msdata=* and -mcall-* options.
10116 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10118 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10119 rate for post-reload scheduling.
10121 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10123 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10125 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10127 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10128 to improve monte carlo in scimark.
10130 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10132 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10133 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10134 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10135 pentium4_cost, nocona_cost): Set reassociation width to 1.
10136 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10137 width to 2 for fp operations and 1 otherwise.
10138 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10140 (atom_cost): Set reassociation width to 2.
10141 (slm_cost, generic_cost): Set fp reassociation width
10142 to 2 and 1 otherwise.
10143 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10144 (core_cost): Set fp reassociation width to 4 and vector to 2.
10145 (ix86_reassociation_width): Rewrite using cost table; special case
10146 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10147 and TARGET_AVX128_OPTIMAL.
10148 * config/i386/i386.h (processor_costs): Add
10149 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10150 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10151 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10152 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10153 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10154 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10156 2017-10-05 Nathan Sidwell <nathan@acm.org>
10158 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10160 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10162 * config/arm/arm.c (arm_test_fpu_data): New.
10163 (arm_run_selftests): Call arm_test_fpu_data.
10165 2017-10-04 Nathan Sidwell <nathan@acm.org>
10167 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10169 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10171 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10173 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10174 check type for aarch64_simd_valid_immediate.
10175 (aarch64_output_simd_mov_immediate): Update prototype.
10176 (aarch64_simd_valid_immediate): Update prototype.
10177 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10178 support for ORR-immediate.
10179 (and<mode>3): modified pattern to add support for BIC-immediate.
10180 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10181 now checks for valid immediate for BIC and ORR based on new enum
10183 (aarch64_output_simd_mov_immediate): Function now used to output
10184 BIC/ORR imm as well based on new enum argument.
10185 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10187 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10188 (aarch64_reg_or_bic_imm): Likewise.
10190 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10192 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10193 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10195 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10199 PR rtl-optimization/82396
10200 * haifa-sched.c (autopref_multipass_init): Simplify
10202 (autopref_rank_data): Simplify sort order.
10203 * sched-int.h (autopref_multipass_data_): Remove
10204 multi_mem_insn_p, min_offset and max_offset.
10206 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10208 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10210 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10212 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10213 vect_doubleint_cvt.
10215 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10217 * doc/sourcebuild.texi: Document vect_long_mult.
10219 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10221 PR tree-optimization/82413
10222 * fold-const.c (build_range_check): Use widest_int when comparing
10223 the maximum ETYPE value with HIGH.
10225 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10227 PR rtl-optimization/82396
10228 * haifa-sched.c (autopref_multipass_init): Simplify
10230 (autopref_rank_data): Simplify sort order.
10231 * sched-int.h (autopref_multipass_data_): Remove
10232 multi_mem_insn_p, min_offset and max_offset.
10234 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10236 PR tree-optimization/82381
10237 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10238 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10241 PR tree-optimization/82374
10242 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10243 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10244 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10245 current_function_decl to the new decl.
10247 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10249 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
10250 helper macro for IEEE float128 hardware built-in functions.
10251 (SQRTF128_ODD): Add built-in functions with the round-to-odd
10253 (TRUNCF128_ODD): Likewise.
10254 (ADDF128_ODD): Likewise.
10255 (SUBF128_ODD): Likewise.
10256 (MULF128_ODD): Likewise.
10257 (DIVF128_ODD): Likewise.
10258 (FMAF128_ODD): Likewise.
10259 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
10260 UNSPEC_TRUNC_ROUND_TO_ODD.
10261 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
10262 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
10263 floating point round to odd instructions.
10264 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
10265 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
10266 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
10267 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
10268 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
10269 (trunc<mode>sf2_hw): Change the truncate with round to odd
10270 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
10271 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
10272 to odd hardware instructions.
10273 (sub<mode>3_odd): Likewise.
10274 (mul<mode>3_odd): Likewise.
10275 (div<mode>3_odd): Likewise.
10276 (sqrt<mode>2_odd): Likewise.
10277 (fma<mode>4_odd): Likewise.
10278 (fms<mode>4_odd): Likewise.
10279 (nfma<mode>4_odd): Likewise.
10280 (nfms<mode>4_odd): Likewise.
10281 (trunc<mode>df2_odd): Change the truncate with round to odd
10282 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
10284 * doc/extend.texi (PowerPC built-in functions): Update documentation
10285 for existing IEEE float128-bit built-in functions. Add built-in
10286 functions that generate the IEEE 128-bit floating point round to
10289 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
10291 PR rtl-optimization/77729
10292 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
10293 to (X&(C1&~C2))|C2 transformations.
10295 2017-10-03 Martin Jambor <mjambor@suse.cz>
10297 PR tree-optimization/82363
10298 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
10299 mismatch, mark lacc written regardless of racc.
10301 2017-10-03 Jakub Jelinek <jakub@redhat.com>
10303 PR tree-optimization/82381
10304 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
10305 stmt_to_insert nor wheather SSA_NAMEs are default defs.
10306 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
10307 fallthrough into reassoc_stmt_dominates_stmt_p.
10310 * combine.c (combine_instructions): Don't combine in unreachable
10313 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10316 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
10317 function to not use the have_cpu variable. Do not set cpu_index,
10318 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
10319 or the default cpu.
10320 (rs6000_valid_attribute_p): Remove duplicate initializations of
10321 old_optimize and func_optimize.
10322 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
10323 (rs6000_activate_target_options): Make global.
10324 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
10327 2017-10-02 Jakub Jelinek <jakub@redhat.com>
10329 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
10330 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
10331 Set *poffset to 0 and *psize and *pmax_size to -1 if
10332 *poffset + *psize overflows in HOST_WIDE_INT.
10334 PR tree-optimization/82387
10335 PR tree-optimization/82388
10336 PR tree-optimization/82389
10337 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
10338 instead of live_bytes non-NULL.
10340 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
10343 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
10346 2017-10-02 Richard Biener <rguenther@suse.de>
10348 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
10349 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
10350 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
10352 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10354 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
10355 requested precision matches the type's.
10356 * calls.c (alloc_max_size): Calculate the new candidate size as
10357 a widest_int and use wi::to_widest when comparing it with the
10358 current candidate size.
10359 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
10360 zero rather than integer_zero_node.
10361 * match.pd: Check for a no-op conversion before using wi::add
10362 rather than after. Use tree_to_uhwi when summing small shift
10363 counts into an unsigned int.
10365 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
10366 Alan Hayward <alan.hayward@arm.com>
10367 David Sherwood <david.sherwood@arm.com>
10370 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
10371 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
10372 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
10373 POINTER_AND_FP_REGS.
10375 2017-10-02 Richard Biener <rguenther@suse.de>
10377 PR tree-optimization/82355
10378 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10379 a mapping for the enclosing loop but avoid generating one for
10380 the loop tree root.
10381 (copy_bb_and_scalar_dependences): Remove premature codegen
10382 error on PHIs in blocks duplicated into multiple places.
10383 * graphite-scop-detection.c
10384 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10385 in the region use it as loop and nest to analyze the DR in.
10386 (try_generate_gimple_bb): Likewise.
10387 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10388 (add_loop_constraints): For blocks in a loop not in the region
10389 create a dimension with a single iteration.
10390 * sese.h (gbb_loop_at_index): Remove assert.
10392 2017-10-01 Kevin Buettner <kevinb@redhat.com>
10394 * omp-expand.c (adjust_context_scope): New function.
10395 (expand_parallel_call): Call adjust_context_scope.
10397 2017-10-01 Jeff Law <law@redhat.com>
10399 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
10400 dom_opt_dom_walker class with direct access to private members.
10401 Add comments. Call test_for_singularity.
10402 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
10403 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
10404 m_dummy_cond anymore.
10405 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
10407 (pass_dominator:execute): Build the dummy_cond here and pass it
10408 to the dom_opt_dom_walker ctor.
10409 (test_for_singularity): New function.
10411 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
10412 Maya Rashish <coypu@sdf.org>
10414 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
10415 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
10416 (alpha*-*-netbsd*) Use nbsd_tm_file.
10417 (arm*-*-netbsdelf*) Likewise.
10418 (i[34567]86-*-netbsdelf*) Likewise.
10419 (x86_64-*-netbsd*) Likewise.
10420 (mips*-*-netbsd*) Likewise.
10421 (powerpc-*-netbsd*) Likewise.
10422 (sh*-*-netbsd*) Likewise.
10423 (sparc-*-netbsdelf*) Likewise.
10424 (sparc64-*-netbsd*) Likewise.
10425 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
10427 (vax-*-netbsdelf*) Likewise.
10428 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
10429 (UINT_FAST8_TYPE) Likewise.
10430 (INT_FAST16_TYPE) Check CHAR_FAST16.
10431 (UINT_FAST16_TYPE) Likewise.
10433 2017-09-30 Jakub Jelinek <jakub@redhat.com>
10436 * config/i386/i386.md
10437 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
10438 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
10439 *divmodsi4_zext_2): New define_insn_and_split.
10440 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
10441 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
10442 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
10443 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
10444 New define_insn_and_split.
10445 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
10446 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
10447 operands[1] having DImode when mode is SImode.
10449 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
10450 always SImode for DIV and MOD in REG_EQUAL notes.
10452 2017-09-29 Yury Gribov <tetra2005@gmail.com>
10454 PR middle-end/82319
10455 * match.pd: Fix handling of NaNs in pattern.
10457 2017-09-29 Jeff Law <law@redhat.com>
10459 * sbitmap.c (bitmap_bit_in_range_p): New function.
10460 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
10461 * tree-ssa-dse.c (live_bytes_read): New function.
10462 (dse_classify_store): Ignore reads of dead bytes.
10464 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
10465 typos and whitespace errors.
10466 * config/i386/predicates.md (address_no_seg_operand): Likewise.
10467 * config/s390/s390.c (s390_emit_prologue): Likewise.
10469 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10472 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
10473 with a symbol for LRA.
10475 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
10477 PR rtl-optimization/82338
10478 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
10480 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
10482 * genmodes.c (calc_wider_mode): Suppress qsort macro.
10483 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
10484 (qsort_chk): Declare.
10485 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
10486 (qsort_chk): New function.
10488 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10490 PR tree-optimization/82337
10491 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10492 phi definition if the PHI result appears in an abnormal PHI.
10493 (find_basis_for_base_expr): Don't record a basis if the LHS of the
10494 basis appears in an abnormal PHI.
10496 2017-09-29 Richard Biener <rguenther@suse.de>
10498 * graphite-isl-ast-to-gimple.c
10499 (translate_isl_ast_to_gimple::set_codegen_error): New function.
10500 (binary_op_to_tree): Use it.
10501 (get_rename_from_scev): Likewise.
10502 (copy_loop_phi_nodes): Likewise.
10503 (copy_bb_and_scalar_dependences): Likewise.
10504 (translate_pending_phi_nodes): Likewise.
10506 2017-09-29 Jakub Jelinek <jakub@redhat.com>
10509 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
10510 for movabsq $(i32 << shift), r64.
10512 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
10514 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
10515 index when encoding %esp as %rsp to avoid 0x67 prefix.
10517 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
10519 * config/i386/i386.md (*movsf_internal, *movdf_internal):
10520 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
10522 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
10524 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
10525 Extensions with more than 16 double VFP registers.
10526 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
10527 to_clear_mask and all code related to it. Replace the remaining
10528 entry by a sbitmap and adapt code accordingly.
10530 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
10532 * brig-builtins.def: Change pure attributes to const.
10534 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10536 * config.gcc (default_gnu_indirect_function): Default to yes for
10537 sparc*-*-linux* with glibc.
10539 2017-09-28 Joseph Myers <joseph@codesourcery.com>
10541 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
10542 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
10543 when creating .init_array and .fini_array sections with priority
10546 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
10549 * config/aarch64/aarch64.c
10550 (aarch64_builtin_support_vector_misalignment): Always return false
10551 when misalignment is unknown.
10553 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10555 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
10556 this function to return false if the definition used by the swap
10557 instruction is artificial, or if the memory address from which the
10558 constant value is loaded is not represented by a base address held
10559 in a register or if the base address register is a frame or stack
10560 pointer. Additionally, return false if the base address of the
10561 loaded constant is a SYMBOL_REF but is not considered to be a
10563 (replace_swapped_load_constant): New function.
10564 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
10565 loaded constant vector with a load of a swapped constant vector.
10567 2017-09-27 Carl Love <cel@us.ibm.com>
10569 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10570 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10571 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10574 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
10576 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
10577 first, always call autopref_rank_data otherwise.
10579 2017-09-27 Richard Biener <rguenther@suse.de>
10581 * graphite-scop-detection.c (find_scop_parameters): Move
10582 loop bound handling ...
10583 (gather_bbs::before_dom_children): ... here, avoiding the need
10584 to build scop_info->loop_nest.
10585 (record_loop_in_sese): Remove.
10586 * sese.h (sese_info_t::loop_nest): Remove.
10587 * sese.c (new_sese_info): Do not allocate loop_nest.
10588 (free_sese_info): Do not free loop_nest.
10590 2017-09-27 Jakub Jelinek <jakub@redhat.com>
10593 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
10594 lhs from calls if the lhs has addressable type.
10596 2017-09-27 Richard Biener <rguenther@suse.de>
10598 * graphite.h (scop::max_alias_set): New member.
10599 * graphite-scop-detection.c: Remove references to non-existing
10600 --param in comments.
10601 (build_alias_sets): Record the maximum alias set used for drs.
10602 (build_scops): Support zero as unlimited for
10603 --param graphite-max-arrays-per-scop.
10604 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
10605 and inline into ...
10606 (build_poly_sr_1): ... here. Compute alias set based on the
10607 maximum alias set used for drs rather than
10608 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
10610 2017-09-27 Richard Biener <rguenther@suse.de>
10612 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
10613 --param loop-block-tile-size=0 to disable tiling.
10615 2017-09-27 Richard Biener <rguenther@suse.de>
10617 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
10618 (graphite-max-nb-scop-params): Document special value zero.
10619 * domwalk.h (dom_walker::STOP): New symbolical constant.
10620 (dom_walker::dom_walker): Add optional parameter for bb to
10622 (dom_walker::~dom_walker): Declare.
10623 (dom_walker::before_dom_children): Document STOP return value.
10624 (dom_walker::m_user_bb_to_rpo): New member.
10625 (dom_walker::m_bb_to_rpo): Likewise.
10626 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
10627 mapping here if not provided by the user.
10628 (dom_walker::~dom_walker): Free bb to RPO mapping if not
10629 provided by the user.
10630 (dom_walker::STOP): Define.
10631 (dom_walker::walk): Do not compute bb to RPO mapping here.
10632 Support STOP return value from before_dom_children to stop
10634 * graphite-optimize-isl.c (optimize_isl): If the schedule
10635 is the same still generate code if -fgraphite-identity
10636 or -floop-parallelize-all are given.
10637 * graphite-scop-detection.c: Include cfganal.h.
10638 (gather_bbs::gather_bbs): Get and pass through bb to RPO
10640 (gather_bbs::before_dom_children): Return STOP for BBs
10642 (build_scops): Compute bb to RPO mapping and pass it to
10643 the domwalk. Treat --param graphite-max-nb-scop-params=0
10644 as not limiting the number of params.
10645 * graphite.c (graphite_initialize): Remove limit on the
10646 number of basic-blocks in a function.
10647 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
10648 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
10649 default value of 10.
10651 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10653 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
10654 Adjust code to eliminate needing to do the shift right 32-bits
10655 operation after XSCVDPSPN.
10657 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10659 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
10660 ((X / Y) != 0 -> X >= Y): Likewise.
10662 2017-09-26 Carl Love <cel@us.ibm.com>
10664 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
10665 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
10666 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
10667 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
10668 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
10669 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
10670 definitions and overloading.
10671 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
10672 statement for P9V_BUILTIN_XST_LEN_R.
10673 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
10674 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
10675 define_expand and define_insn for the instructions and builtins.
10676 * doc/extend.texi: Update the built-in documentation file for the new
10677 built-in functions.
10678 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
10679 define_insn for the instructions
10681 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
10684 * gcc/config/netbsd-protos.h: New file.
10685 * gcc/config/netbsd.c: New file.
10686 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
10687 * gcc/config/t-netbsd: New file.
10688 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
10689 (tmake_file) Add t-netbsd.
10690 (extra_objs) Add netbsd.o.
10692 2017-09-26 Janus Weil <janus@gcc.gnu.org>
10696 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
10698 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10700 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
10701 sign extension from a vector register to a GPR by doing a 32-bit
10702 direct move and then an EXTSW.
10703 (extendsi<mode>2 splitter): Likewise.
10704 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
10705 right or vector extract after doing XSCVDPSPN. Use
10706 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
10708 (movdi_from_sf_zero_ext): Likewise.
10709 (reload_gpr_from_vsxsf): Likewise.
10710 (p8_mfvsrd_4_disf): Delete, no longer used.
10711 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
10712 then needing to move the SFmode to a GPR to use the XSCVDPSP
10713 instruction instead of FRSP and XSCVDPSPN.
10714 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
10715 it is adjacent to the other XSCVSPDP insns.
10716 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
10717 SFmode to be in traditional Altivec registers.
10718 (vsx_xscvdpspn): Eliminate useless alternative constraint.
10719 (vsx_xscvspdpn): Likewise.
10720 (vsx_xscvspdpn_scalar): Likewise.
10722 2017-09-26 Martin Jambor <mjambor@suse.cz>
10724 * tree-sra.c (compare_access_positions): Put integral types first,
10725 stabilize sorting of integral types, remove conditions putting
10726 non-full-precision integers last.
10727 (sort_and_splice_var_accesses): Disable scalarization if a
10728 non-integert would be represented by a non-full-precision integer.
10730 2017-09-26 Joseph Myers <joseph@codesourcery.com>
10732 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
10733 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
10734 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
10735 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
10736 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
10737 conditionals inside the function instead of around it. Call
10738 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
10739 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
10741 2017-09-26 Richard Biener <rguenther@suse.de>
10743 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
10745 (scop_detection::build_scop_breadth): ... this. Removed.
10746 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
10747 (scop_detection::harmful_stmt_in_bb): Likewise.
10748 (scop_detection::graphite_can_represent_stmt): Likewise.
10749 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
10750 (scop_detection::can_represent_loop): Remove recursion, fold in ...
10751 (scop_detection::can_represent_loop_1): ... this. Removed.
10752 (scop_detection::harmful_loop_in_region): Simplify after inlining
10753 the above and remove more quadraticness.
10754 (build_scops): Adjust.
10755 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
10758 2017-09-26 Jakub Jelinek <jakub@redhat.com>
10761 * config/i386/i386.c (ix86_print_operand_address_as): Only test
10762 REGNO (base) == SP_REG if base is a REG.
10764 PR middle-end/35691
10765 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
10766 if it is different SSA_NAME.
10767 (optimize_range_tests_cmp_bitwise): New function.
10768 (optimize_range_tests): Call it.
10770 2017-09-26 Richard Biener <rguenther@suse.de>
10772 PR tree-optimization/82321
10773 * graphite.c (canonicalize_loop_closed_ssa): Properly check
10774 for the def being inside the loop.
10776 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10778 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
10780 * config/s390/s390-builtins.def: Fix constraint on op4.
10782 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10784 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
10785 independent expanders.
10786 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
10787 ("vec_ordered", "vec_unordered"): New expanders.
10789 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10791 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
10794 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10796 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
10797 vec_unpacks_lo_v16qi.
10798 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
10800 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10802 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
10803 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
10804 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
10806 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10808 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
10810 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
10812 ("*vec_slb<mode>"): New insn pattern.
10813 ("vec_shr_<mode>"): New expander.
10814 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
10815 and force the shift count operand to V16QImode.
10816 ("vec_srb<mode>"): Set shift count mode to V16QI.
10818 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10820 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
10821 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
10822 ("vec_widen_smult_hi_<mode>"): New expander definitions.
10824 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
10827 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
10829 2017-09-26 Richard Biener <rguenther@suse.de>
10831 PR tree-optimization/82320
10832 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
10835 2017-09-25 Jeff Law <law@redhat.com>
10837 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
10838 prototype for new argument.
10839 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
10840 mostly extracted from rs6000_emit_allocate_stack.
10841 (rs6000_emit_probe_stack_range_stack_clash): New function.
10842 (rs6000_emit_allocate_stack): Call
10843 rs6000_emit_probe_stack_range_stack_clash as needed.
10844 (rs6000_emit_probe_stack_range): Add additional argument
10845 to call to gen_probe_stack_range{si,di}.
10846 (output_probe_stack_range): New.
10847 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
10848 (output_probe_stack_range_stack_clash): New.
10849 (rs6000_emit_prologue): Emit notes into dump file as requested.
10850 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
10851 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
10852 Add additional operand and pass it to output_probe_stack_range.
10854 2017-09-25 Bin Cheng <bin.cheng@arm.com>
10856 PR tree-optimization/82163
10857 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
10858 (checking_verify_loop_closed_ssa): New parameter.
10859 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
10860 (check_loop_closed_ssa_stmt): Delete.
10861 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
10862 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
10863 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
10866 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
10868 * brig-builtins.def: Treat HSAIL barrier builtins as
10869 setjmp/longjump style functions.
10871 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10873 * target.def (constant_alignment): New hook.
10874 * defaults.h (CONSTANT_ALIGNMENT): Delete.
10875 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
10876 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
10877 * doc/tm.texi: Regenerate.
10878 * targhooks.h (default_constant_alignment): Declare.
10879 (constant_alignment_word_strings): Likewise.
10880 * targhooks.c (default_constant_alignment): New function.
10881 (constant_alignment_word_strings): Likewise.
10882 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
10883 instead of CONSTANT_ALIGNMENT.
10884 * varasm.c (align_variable, get_variable_align, build_constant_desc)
10885 (force_const_mem): Likewise.
10886 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10887 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10888 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10889 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10890 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10892 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10893 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10894 constant_alignment_word_strings.
10895 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10896 (CONSTANT_ALIGNMENT): Likewise.
10897 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10898 (arm_constant_alignment): New function.
10899 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10900 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10901 constant_alignment_word_strings.
10902 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10903 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10904 constant_alignment_word_strings.
10905 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10906 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10907 (cris_constant_alignment): New function.
10908 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10909 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10910 (epiphany_constant_alignment): New function.
10911 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10912 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10913 constant_alignment_word_strings.
10914 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10915 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10916 constant_alignment_word_strings.
10917 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10918 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10919 constant_alignment_word_strings.
10920 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10921 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10922 * config/i386/i386.c (ix86_constant_alignment): Make static.
10923 Use the same interface as the target hook.
10924 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10925 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10926 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10927 constant_alignment_word_strings.
10928 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10929 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10930 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10931 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10932 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10933 constant_alignment_word_strings.
10934 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10935 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10936 constant_alignment_word_strings.
10937 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10938 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10939 constant_alignment_word_strings.
10940 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10941 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10943 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10944 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10945 * config/mips/mips.c (mips_constant_alignment): New function.
10946 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10947 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10948 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10949 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10950 (mmix_constant_alignment): Make static. Use the same interface
10951 as the target hook.
10952 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10953 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10954 constant_alignment_word_strings.
10955 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10956 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10957 constant_alignment_word_strings.
10958 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10959 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10960 constant_alignment_word_strings.
10961 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10962 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10963 (rs6000_constant_alignment): New function.
10964 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10965 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10966 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10967 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10968 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10969 (rs6000_constant_alignment): New function.
10970 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10971 * config/s390/s390.c (s390_constant_alignment): New function.
10972 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10973 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10974 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10975 constant_alignment_word_strings.
10976 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10977 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10978 (sparc_constant_alignment): New function.
10979 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10980 * config/spu/spu.c (spu_constant_alignment): New function.
10981 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10982 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10983 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10984 constant_alignment_word_strings.
10985 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10986 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10987 constant_alignment_word_strings.
10988 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10989 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10990 constant_alignment_word_strings.
10991 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10992 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10993 (visium_constant_alignment): New function.
10994 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10995 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10996 (xtensa_constant_alignment): New function.
10997 * system.h (CONSTANT_ALIGNMENT): Poison.
10999 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11001 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11002 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11003 (rs6000_builtin_valid_without_lhs): New helper function.
11004 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11005 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11007 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11009 * target.h (vec_perm_indices): Use unsigned short rather than
11011 (auto_vec_perm_indices): Likewise.
11012 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11013 Use unsigned int rather than unsigned char.
11014 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11016 2017-09-25 Richard Biener <rguenther@suse.de>
11018 * cfgloop.h (sort_sibling_loops): Declare.
11019 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11020 (sort_sibling_loops): New function sorting the sibling loop list
11022 * graphite.c (graphite_transform_loops): Sort sibling loops.
11024 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11026 * target.def (vec_perm_const_ok): Change sel parameter to
11028 * optabs-query.c (can_vec_perm_p): Update accordingly.
11029 * doc/tm.texi: Regenerate.
11030 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11031 auto_vec_perm_indices and remove separate nelt field.
11032 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11033 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11034 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11035 (aarch64_expand_vec_perm_const): Update accordingly.
11036 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11037 to vec_perm_indices.
11038 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11039 auto_vec_perm_indices and remove separate nelt field.
11040 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11041 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11042 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11044 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11045 to vec_perm_indices.
11046 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11047 sel to vec_perm_indices.
11048 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11049 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11050 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11052 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11055 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11058 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11059 on the FUNCTION_DECL function context if it has a DIE that is a
11062 2017-09-25 Richard Biener <rguenther@suse.de>
11064 PR tree-optimization/82285
11065 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11068 2017-09-25 Tom de Vries <tom@codesourcery.com>
11072 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11075 2017-09-25 Richard Biener <rguenther@suse.de>
11077 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11078 ISL errors other than isl_error_quota happen. Dump if the
11079 schedule is the same.
11080 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11081 errors instead of aborting inside ISL.
11083 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11086 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11087 of libgcc_eh for m64.
11088 * config/i386/darwin64.h: Likewise.
11090 2017-09-25 Richard Biener <rguenther@suse.de>
11092 PR middle-end/82144
11093 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11094 attribute for incomplete types nor twice for complete ones.
11096 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11099 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11100 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11103 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11106 * config/i386/i386.opt (mprefer-avx256): Use
11107 ix86_target_flags variable.
11108 * config/i386/i386.c (ix86_target_string): Move
11109 -mprefer-avx256 to flag2_opts.
11111 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11113 PR middle-end/35691
11114 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11115 and x != -1 | y != -1 into (x & y) != -1.
11117 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11119 * config.gcc: Add new case statement to set
11120 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11121 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11122 s390x-*-linux* case statements. Added aarch64 to the list of
11123 supported architectures.
11125 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11127 PR tree-optimization/82289
11128 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11129 STMT_VINFO_RELEVANT_P.
11131 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11132 Alan Hayward <alan.hayward@arm.com>
11133 David Sherwood <david.sherwood@arm.com>
11135 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11136 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11137 (extract_range_from_binary_expr_1): Don't call
11138 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11140 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11141 Alan Hayward <alan.hayward@arm.com>
11142 David Sherwood <david.sherwood@arm.com>
11144 * target.def (preferred_vector_alignment): New hook.
11145 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11147 * doc/tm.texi: Regenerate.
11148 * targhooks.h (default_preferred_vector_alignment): Declare.
11149 * targhooks.c (default_preferred_vector_alignment): New function.
11150 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11152 (DR_TARGET_ALIGNMENT): New macro.
11153 (aligned_access_p): Update commentary.
11154 (vect_known_alignment_in_bytes): New function.
11155 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11157 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11158 Calculate the misalignment based on the target alignment rather than
11160 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11161 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11162 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11163 the target alignment, rather than masking the element misalignment
11164 with the number of elements in a vector. Also use the target
11165 alignment when calculating the maximum number of peels.
11166 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11167 instead of TYPE_ALIGN_UNIT.
11168 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11169 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11170 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11171 (vect_create_data_ref_ptr): Likewise.
11172 (vect_setup_realignment): Realign by ANDing with
11173 -DR_TARGET_MISALIGNMENT.
11174 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11175 the number of peels based on DR_TARGET_ALIGNMENT.
11176 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11177 with the guaranteed alignment boundary when deciding whether
11179 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11180 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11181 (ensure_base_align): Remove stmt_info parameter. Get the
11182 target base alignment from DR_TARGET_ALIGNMENT.
11183 (vectorizable_store): Update call accordingly. Interpret
11184 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11186 (vectorizable_load): Likewise.
11188 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11189 Alan Hayward <alan.hayward@arm.com>
11190 David Sherwood <david.sherwood@arm.com>
11192 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11193 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11194 (vect_enhance_data_refs_alignment): Likewise.
11196 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11198 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11199 error. Only quit immediately if parsing is complete.
11200 (BEGIN): Initialize fatal_err and parse_done.
11201 (begin fpu, end fpu): Check number of arguments.
11202 (begin arch, end arch): Likewise.
11203 (begin cpu, end cpu): Likewise.
11204 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11205 (optalias): Likewise.
11207 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11209 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11210 * config/arm/arm-isa.h: Delete. Move definitions to ...
11211 * arm-cpus.in: ... here. Use new feature and fgroup values.
11212 * config/arm/arm.c (arm_option_override): Use lower case for feature
11214 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11215 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11216 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11218 (print_isa_bits_for): New function.
11219 (gen_isa): New function.
11220 (gen_comm_data): Use print_isa_bits_for.
11221 (define feature): New keyword.
11222 (define fgroup): New keyword.
11223 * config/arm/t-arm (TM_H): Remove.
11224 (GTM_H): Add arm-isa.h.
11225 (arm-isa.h): Add rule to generate file.
11226 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11227 case for feature bit names.
11229 2017-09-22 Richard Biener <rguenther@suse.de>
11231 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11233 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11234 print of no dependency loops ...
11235 * graphite.c (graphite_transform_loops): ... here.
11236 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11238 (same_close_phi_node, remove_duplicate_close_phi,
11239 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11240 (canonicalize_loop_closed_ssa): ... here and simplify.
11241 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11242 (optimize_isl): Use dump_printf_loc to tell when we stopped
11243 optimizing because of an ISL timeout.
11245 2017-09-22 Richard Biener <rguenther@suse.de>
11247 PR tree-optimization/82291
11248 * tree-if-conv.c (predicate_mem_writes): Make sure to
11249 remove writes in blocks predicated with false.
11251 2017-09-22 Richard Biener <rguenther@suse.de>
11253 * sese.c: Include cfganal.h.
11254 (if_region_set_false_region): Remove.
11255 (create_if_region_on_edge): Likewise.
11256 (move_sese_in_condition): Re-implement without destroying
11259 2017-09-22 Richard Biener <rguenther@suse.de>
11261 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
11262 Verify both BBs contain loop PHI nodes before dispatching to
11263 copy_loop_phi_args.
11264 (graphite_regenerate_ast_isl): Do not recompute dominators,
11265 do not verify three times. Restructure for clarity.
11266 * graphite-scop-detection.c (same_close_phi_node,
11267 remove_duplicate_close_phi, make_close_phi_nodes_unique,
11268 defined_in_loop_p, canonicalize_loop_closed_ssa,
11269 canonicalize_loop_closed_ssa_form): Simplify, remove excess
11270 checking and SSA rewrite, move to ...
11271 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
11272 (graphite_initialize): Do not pass in ctx, do not reset the
11273 SCEV cache, compute only dominators.
11274 (graphite_transform_loops): Allocate ISL ctx after
11275 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
11276 Maintain post-dominators only around build_scops.
11277 * sese.c (if_region_set_false_region): Make static. Free
11278 and recompute dominators.
11279 (move_sese_in_condition): Assert we don't get called with
11280 post-dominators computed.
11281 * sese.h (if_region_set_false_region): Remove.
11283 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
11285 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
11286 mode attribute for TARGET_AVX512VL.
11288 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
11290 * config/i386/i386.opt (mprefer-avx256): New option.
11291 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
11293 (ix86_preferred_simd_mode): Return 256-bit AVX modes
11294 for TARGET_PREFER_AVX256.
11295 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
11297 2017-09-21 Jeff Law <law@redhat.com>
11299 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
11300 Fix dump output if the only stack space is for pushed registers.
11302 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11304 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
11307 2017-09-21 Martin Sebor <msebor@redhat.com>
11310 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
11311 code (in C++) or code that triggers warnings.
11313 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
11315 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
11317 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
11319 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
11320 * sched-rgn.c: Ditto.
11321 * sel-sched-ir.c: Ditto.
11323 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
11325 * toplev.h (set_random_seed): Adjust return type.
11326 * toplev.c (init_local_tick): Move eager initialization of random_seed
11327 to get_random_seed. Adjust comment.
11328 (init_random_seed): Inline to get_random_seed, delete.
11329 (get_random_seed): Initialize random_seed lazily.
11330 (set_random_seed): Do not return previous value.
11331 (print_switch_value): Do not call get_random_seed.
11333 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
11335 * cgraph.c (delete_function_version): New, broken out from...
11336 (cgraph_node::delete_function_version): ...here. Rename to
11337 cgraph_node::delete_function_version_by_decl. Update all uses.
11338 (cgraph_node::remove): Call delete_function_version.
11340 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11343 * tree-inline.c (expand_call_inline): Emit clobber stmts for
11344 VAR_DECLs to which addressable non-volatile parameters are mapped
11345 and for id->retvar after the return value assignment. Clear
11346 id->retval and id->retbnd after inlining.
11348 2017-09-21 Richard Biener <rguenther@suse.de>
11350 PR tree-optimization/82276
11351 PR tree-optimization/82244
11352 * tree-vrp.c (build_assert_expr_for): Set
11353 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
11355 (remove_range_assertions): Revert earlier change.
11357 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
11360 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
11362 2017-09-21 Richard Biener <rguenther@suse.de>
11364 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11365 Restore valid IL after code generation errors.
11366 * graphite.c (graphite_transform_loops): Diagnose code
11367 generation issues as MSG_MISSED_OPTIMIZATION and continue
11368 with processing SCOPs.
11370 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11371 Alan Hayward <alan.hayward@arm.com>
11372 David Sherwood <david.sherwood@arm.com>
11374 * calls.c (compute_argument_addresses): Use simplify_gen_binary
11375 rather than choosing between plus_constant and gen_rtx_<CODE>.
11376 * expr.c (emit_push_insn): Likewise.
11377 (expand_expr_real_2): Likewise.
11379 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11380 Alan Hayward <alan.hayward@arm.com>
11381 David Sherwood <david.sherwood@arm.com>
11383 * loop-unroll.c (split_iv): Call copy_rtx on the step.
11385 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11386 Alan Hayward <alan.hayward@arm.com>
11387 David Sherwood <david.sherwood@arm.com>
11389 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
11390 calling tree_to_uhwi.
11392 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11393 Alan Hayward <alan.hayward@arm.com>
11394 David Sherwood <david.sherwood@arm.com>
11396 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
11397 INTEGER_CST rather than a negative test for ADDR_EXPR.
11399 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
11400 Alan Hayward <alan.hayward@arm.com>
11401 David Sherwood <david.sherwood@arm.com>
11403 * tree-vrp.c (extract_range_from_binary_expr_1): Check
11404 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
11406 2017-09-21 Richard Biener <rguenther@suse.de>
11408 PR tree-optimization/71351
11409 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
11410 graphite_create_new_loop_guard): Remove, fold remaining parts
11412 (translate_isl_ast_node_for): ... here and simplify.
11414 2017-09-21 Jakub Jelinek <jakub@redhat.com>
11417 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
11418 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
11419 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
11420 alternative always use QI mode, for -Os imov (=R,R) alternative
11421 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
11424 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11425 Jeff Law <law@redhat.com>
11427 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
11428 (allocate_stack_space): New function, partially extracted from
11429 s390_emit_prologue.
11430 (s390_emit_prologue): Track offset to most recent stack probe.
11431 Code to allocate space moved into allocate_stack_space.
11432 Dump actions when no stack is allocated.
11433 (s390_prologue_plus_offset): New function.
11434 (s390_emit_stack_probe): Likewise.
11436 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
11438 * common.opt (Wa, Wl, Wp, g, gz=): Add
11440 (gno-column-info): Remove.
11441 (gcolumn-info): Drop RejectNegative.
11442 (gno-): New prefix.
11443 (gno-record-gcc-switches): Remove.
11444 (grecord-gcc-switches): Drop RejectNegative.
11445 (gno-split-dwarf): Remove.
11446 (gsplit-dwarf): Drop RejectNegative.
11447 (gno-strict-dwarf): Remove.
11448 (gstrict-dwarf): Drop RejectNegative.
11449 * config/darwin.opt (gfull, gused): Add RejectNegative.
11450 * dwarf2out.c (gen_producer_string): Drop
11451 gno-record-gcc-switches handler.
11452 * optc-gen.awk: Add g to prefixes with negative forms.
11453 * opts-common.c (remapping_prefix_p): New.
11454 (find_opt): Check it.
11455 (generate_canonical_option): Test g prefix.
11456 (option_map): Add -gno- mapping.
11457 (add_misspelling_candidates): Check remapping_prefix_p.
11459 2017-09-20 Jeff Law <law@redhat.com>
11461 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
11462 thinko in stack clash protection support.
11464 * explow.c (compute_stack_clash_protection_loop_data): Use
11465 CONST_INT_P instead of explicit test. Verify object is a
11466 CONST_INT_P before looking at INTVAL.
11467 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
11468 instead of explicit test.
11470 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
11473 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
11474 address instead of to r1 and r11.
11476 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
11478 * config.gcc: Support "knm".
11479 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
11480 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11482 * config/i386/i386.c (m_KNM): Define.
11483 (processor_target_table): Add "knm".
11485 (ix86_option_override_internal): Add "knm".
11486 (ix86_issue_rate): Add PROCESSOR_KNM.
11487 (ix86_adjust_cost): Ditto.
11488 (ia32_multipass_dfa_lookahead): Ditto.
11489 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
11490 (fold_builtin_cpu): Add M_INTEL_KNM.
11491 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
11492 (processor_type): Add PROCESSOR_KNM.
11493 * config/i386/x86-tune.def: Add m_KNM.
11494 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
11496 2017-09-20 Richard Biener <rguenther@suse.de>
11498 PR tree-optimization/80213
11499 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
11500 are allowed in empty BBs as well.
11501 (canonicalize_loop_closed_ssa): Also look for other complex
11503 (scop_detection::get_sese): Include the loop-closed PHI block
11505 (scop_detection::merge_sese): Remove code adding extra blocks.
11506 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
11507 (build_scops): Assert the final returned scop is invalid.
11509 2017-09-20 Richard Biener <rguenther@suse.de>
11511 PR tree-optimization/82264
11512 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
11514 (vn_phi_lookup): Likewise.
11515 (vn_phi_insert): Likewise.
11517 2017-09-20 Jakub Jelinek <jakub@redhat.com>
11519 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
11520 that fits into uhwi or shwi, add DW_AT_const_value regardless
11521 of early_dwarf without going through RTL, using add_AT_unsigned
11524 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
11525 (DEBUG_LTO_ABBREV_SECTION): Likewise.
11526 (DEBUG_LTO_MACINFO_SECTION): Likewise.
11527 (DEBUG_MACRO_SECTION): Likewise.
11528 (DEBUG_LTO_MACRO_SECTION): Likewise.
11529 (DEBUG_STR_DWO_SECTION): Likewise.
11530 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
11531 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
11532 (DEBUG_LTO_DWO_LINE_SECTION): Define.
11533 (DEBUG_LTO_LINE_STR_SECTION): Define.
11534 (init_sections_and_labels): Initialize debug_line_str_section
11535 variable. Initialize debug_loc_section for -gdwarf-5 to
11536 DEBUG_LOCLISTS_SECTION. Formatting fixes.
11538 2017-09-20 Richard Biener <rguenther@suse.de>
11540 * graphite-sese-to-poly.c (extract_affine): Properly handle
11541 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
11543 2017-09-20 Richard Biener <rguenther@suse.de>
11545 PR tree-optimization/81373
11546 * graphite-scop-detection.c (build_cross_bb_scalars_def):
11547 Force SESE live-out defs to be handled even if they are
11550 2017-09-19 Jeff Law <law@redhat.com>
11552 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
11553 nothing for stack adjustments with REG_STACK_CHECK.
11554 * sched-deps.c (parse_add_or_inc): Reject insns with
11555 REG_STACK_CHECK from dependency breaking.
11556 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
11557 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
11558 * reg-notes.def (STACK_CHECK): New note.
11560 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
11561 (ix86_expand_prologue): Dump stack clash info as needed.
11562 Call ix86_adjust_stack_and_probe_stack_clash as needed.
11564 * function.c (dump_stack_clash_frame_info): New function.
11565 * function.h (dump_stack_clash_frame_info): Prototype.
11566 (enum stack_clash_probes): New enum.
11568 * config/alpha/alpha.c (alpha_expand_prologue): Also check
11569 flag_stack_clash_protection.
11570 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
11571 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
11572 (arm_frame_pointer_required): Likewise.
11573 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
11574 (ia64_expand_prologue): Likewise.
11575 * config/mips/mips.c (mips_expand_prologue): Likewise.
11576 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
11577 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11578 (sparc_flat_expand_prologue): Likewise.
11579 * config/spu/spu.c (spu_expand_prologue): Likewise.
11581 * explow.c: Include "params.h".
11582 (anti_adjust_stack_and_probe_stack_clash): New function.
11583 (get_stack_check_protect): Likewise.
11584 (compute_stack_clash_protection_loop_data): Likewise.
11585 (emit_stack_clash_protection_loop_start): Likewise.
11586 (emit_stack_clash_protection_loop_end): Likewise.
11587 (allocate_dynamic_stack_space): Use get_stack_check_protect.
11588 Use anti_adjust_stack_and_probe_stack_clash.
11589 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
11590 (emit_stack_clash_protection_loop_start): Likewise.
11591 (emit_stack_clash_protection_loop_end): Likewise.
11592 * rtl.h (get_stack_check_protect): Prototype.
11593 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
11594 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
11595 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
11597 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
11599 * doc/tm.texi: Rebuilt.
11600 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
11601 get_stack_check_protect.
11602 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
11603 * config/arm/arm.c (arm_expand_prologue): Likewise.
11604 (arm_frame_pointer_required): Likewise.
11605 * config/i386/i386.c (ix86_expand_prologue): Likewise.
11606 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
11607 * config/mips/mips.c (mips_expand_prologue): Likewise.
11608 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
11609 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
11610 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
11611 (sparc_flat_expand_prologue): Likewise.
11613 * common.opt (-fstack-clash-protection): New option.
11614 * flag-types.h (enum stack_check_type): Note difference between
11615 -fstack-check= and -fstack-clash-protection.
11616 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
11617 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
11618 * toplev.c (process_options): Issue warnings/errors for cases
11619 not handled with -fstack-clash-protection.
11620 * doc/invoke.texi (-fstack-clash-protection): Document new option.
11621 (-fstack-check): Note additional problem with -fstack-check=generic.
11622 Note that -fstack-check is primarily for Ada and refer users
11623 to -fstack-clash-protection for stack-clash-protection.
11624 Document new params for stack clash protection.
11626 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
11628 * config/i386/i386.c (ix86_split_long_move): Do not handle
11629 address used for LEA in a special way.
11631 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
11633 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
11635 2017-09-19 Martin Sebor <msebor@redhat.com>
11638 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
11639 of incompatible types.
11641 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
11643 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11644 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
11645 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11646 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
11648 2017-09-19 Richard Biener <rguenther@suse.de>
11650 PR tree-optimization/82244
11651 * tree-vrp.c (remove_range_assertions): Do not propagate
11652 a constant to abnormals but replace the assert with a copy.
11654 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
11656 PR rtl-optimization/57878
11657 PR rtl-optimization/68988
11658 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
11659 avoidance test involving non_reload_pseudos. Move frequency test
11660 below the general fragmentation avoidance test.
11662 2017-09-19 Richard Biener <rguenther@suse.de>
11664 PR tree-optimization/69728
11665 * graphite-sese-to-poly.c (schedule_error): New global.
11666 (add_loop_schedule): Handle empty domain by failing the
11668 (build_original_schedule): Handle schedule_error.
11670 2017-09-19 Richard Biener <rguenther@suse.de>
11672 * graphite-scop-detection.c (scop_detection::can_represent_loop):
11673 Do not iterate to sibling loops but only to siblings of inner
11676 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
11679 * config/m68k/m68k.md (moveq feeding equality comparison): Check
11680 that the registers are different.
11682 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
11684 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
11685 to processor_model and "amdfam17h" to arch_names_table.
11686 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
11688 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11691 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
11693 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11694 Alan Hayward <alan.hayward@arm.com>
11695 David Sherwood <david.sherwood@arm.com>
11697 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
11699 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
11701 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
11702 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
11703 vect_schedule_slp_instance.
11704 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
11705 Update call to vect_slp_analyze_node_operations. Simplify return
11707 (vect_slp_analyze_bb_1): Update call accordingly.
11708 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
11709 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
11710 (vect_schedule_slp): Update call accordingly.
11712 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11713 Alan Hayward <alan.hayward@arm.com>
11714 David Sherwood <david.sherwood@arm.com>
11716 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
11717 with types that aren't in fact scalar.
11719 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11721 * tree-vect-slp.c (vect_record_max_nunits): New function,
11723 (vect_build_slp_tree_1): ...here.
11724 (vect_build_slp_tree_2): Call it for phis too.
11726 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11728 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
11729 to vect_get_vec_def_for_operand when getting the mask operand.
11731 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11732 Alan Hayward <alan.hayward@arm.com>
11733 David Sherwood <david.sherwood@arm.com>
11735 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
11738 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11739 Alan Hayward <alan.hayward@arm.com>
11740 David Sherwood <david.sherwood@arm.com>
11742 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
11743 calculation for vector booleans.
11745 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
11746 Alan Hayward <alan.hayward@arm.com>
11747 David Sherwood <david.sherwood@arm.com>
11749 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
11751 (vect_transform_stmt): ...here.
11752 (vect_analyze_stmt): Use it instead of calling
11753 vectorizable_live_operation directly.
11755 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
11757 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
11758 non-SIMT targets in acc vector loops.
11760 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
11762 * configure.ac: Add arc and check if assembler supports gdwarf2.
11763 * configure: Regenerate.
11765 2017-09-18 Richard Biener <rguenther@suse.de>
11767 PR tree-optimization/82220
11768 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
11769 epilogue niters from the min_profitable_iters compute.
11771 2017-09-18 Jakub Jelinek <jakub@redhat.com>
11774 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
11775 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
11776 (ix86_init_pic_reg): Revert 2017-09-01 changes.
11778 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
11781 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
11782 switching to a new text section.
11784 2017-09-18 Richard Biener <rguenther@suse.de>
11786 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
11788 (build_alias_set): Reject aliases with no access function.
11790 2017-09-18 Richard Biener <rguenther@suse.de>
11792 PR tree-optimization/79622
11793 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
11795 (build_cross_bb_scalars_use): Likewise.
11797 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
11799 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
11801 2017-09-18 Alan Modra <amodra@gmail.com>
11804 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
11805 stack_pointer_rtx for count 0. Update comments. Break up
11806 large rtl expression.
11808 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
11810 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
11811 Increase to 20 bytes.
11812 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
11813 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
11814 or avx version of the stub.
11816 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
11819 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
11820 compute the minimum stack alignment. Also update preferred stack
11821 boundary for leaf functions.
11823 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
11825 PR tree-optimization/82228
11826 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
11829 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11831 * common/config/nds32/nds32-common.c
11832 (nds32_option_optimization_table): Refine formatting.
11833 (nds32_option_optimization_table): Use -fsched-pressure and
11834 -fomit-frame-pointer for specific optimization level.
11836 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
11838 * config/nds32/nds32.c: Refine formatting and comments.
11839 * config/nds32/nds32.h: Likewise.
11840 * config/nds32/nds32.md: Likewise.
11841 * config/nds32/nds32-cost.c: Likewise.
11842 * config/nds32/nds32-isr.c: Likewise.
11843 * config/nds32/nds32-md-auxiliary.c: Likewise.
11844 * config/nds32/nds32-multiple.md: Likewise.
11845 * config/nds32/nds32-predicates.c: Likewise.
11847 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
11848 Jakub Jelinek <jakub@redhat.com>
11850 Add support for -std=c++2a.
11851 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
11853 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
11855 2017-09-15 Steve Ellcey <sellcey@cavium.com>
11858 * doc/extend.texi (Common Function Attributes): Add
11859 references to ARM, AArch64, and S/390 specific attributes.
11860 (Function Specific Option Pragmas): Add AArch64 and S/390
11861 to list of back ends that support the target pragma.
11863 2017-09-15 Nathan Sidwell <nathan@acm.org>
11865 * doc/standards.texi: Fix C++17 description. Update URLs for
11868 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11870 * common.opt (Wcast-align=strict): New warning option.
11871 * doc/invoke.texi: Document -Wcast-align=strict.
11873 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
11875 * cgraph.h (cgraph_thunk_info): Add comments.
11876 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
11877 assert for VIRTUAL_* arguments stricter.
11879 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
11881 PR tree-optimization/71026
11882 * match.pd: Move RDIV patterns from fold-const.c
11883 * fold-const.c (distribute_real_division): Removed.
11884 (fold_binary_loc): Remove calls to distribute_real_divison.
11886 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11888 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11889 c++1z and gnu++1z as deprecated. Change other references to
11890 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11891 Change -Wc++1z-compat to -Wc++17-compat.
11892 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11893 * dwarf2out.c (highest_c_language): Handle C++17.
11894 (gen_compile_unit_die): Likewise.
11896 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11898 PR rtl-optimization/82192
11899 * combine.c (make_extraction): Don't look through non-paradoxical
11900 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11903 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11904 Alan Hayward <alan.hayward@arm.com>
11905 David Sherwood <david.sherwood@arm.com>
11907 * target.def (function_arg_offset): New hook.
11908 * targhooks.h (default_function_arg_offset): Declare.
11909 * targhooks.c (default_function_arg_offset): New function.
11910 * function.c (locate_and_pad_parm): Use
11911 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11912 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11913 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11914 * doc/tm.texi: Regenerate.
11915 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11916 * config/spu/spu.c (spu_function_arg_offset): New function.
11917 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11918 * system.h (FUNCTION_ARG_OFFSET): Poison.
11920 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11921 Alan Hayard <alan.hayward@arm.com>
11922 David Sherwood <david.sherwood@arm.com>
11924 * target.def (truly_noop_truncation): New hook.
11925 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11926 than TRULY_NOOP_TRUNCATION.
11927 * hooks.h (hook_bool_uint_uint_true): Declare.
11928 * hooks.c (hook_bool_uint_uint_true): New function.
11929 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11930 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11931 * doc/tm.texi: Regenerate.
11932 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11933 rather than TRULY_NOOP_TRUNCATION in comments.
11934 (simplify_comparison): Likewise.
11935 (record_truncated_value): Likewise.
11936 * expmed.c (extract_bit_field_1): Likewise.
11937 (extract_split_bit_field): Likewise.
11938 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11939 instead of TRULY_NOOP_TRUNCATION.
11940 * function.c (assign_parm_setup_block): Likewise.
11941 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11942 * rtlhooks.c: Include target.h.
11943 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11944 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11945 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11946 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11947 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11948 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11949 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11950 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11951 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11952 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11953 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11954 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11955 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11956 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11957 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11958 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11959 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11960 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11961 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11962 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11963 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11964 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11965 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11966 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11967 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11968 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11969 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11970 rather than TRULY_NOOP_TRUNCATION in comments.
11971 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11972 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11973 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11974 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11975 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11976 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11977 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11978 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11979 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11980 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11981 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11982 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11983 rather than TRULY_NOOP_TRUNCATION in comments.
11984 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11985 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11986 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11987 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11988 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11989 TRULY_NOOP_TRUNCATION condition.
11990 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11991 (TRULY_NOOP_TRUNCATION): Delete.
11992 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11993 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11994 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11995 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11996 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11997 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11998 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11999 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12000 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12001 rather than TRULY_NOOP_TRUNCATION in comments.
12002 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12003 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12004 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12005 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12006 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12007 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12009 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12012 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12013 (*cmp_ior): Likewise.
12014 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12015 (*ior_scc_scc_cmp): Likewise.
12016 (*and_scc_scc): Likewise.
12017 (*and_scc_scc_cmp): Likewise.
12019 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12020 Alan Hayard <alan.hayward@arm.com>
12021 David Sherwood <david.sherwood@arm.com>
12023 * target.def (can_change_mode_class): New hook.
12024 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12025 (hard_regno_nregs): Likewise.
12026 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12027 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12028 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12029 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12030 (LOAD_EXTEND_OP): Update accordingly.
12031 * doc/tm.texi: Regenerate.
12032 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12033 CANNOT_CHANGE_MODE_CLASS.
12034 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12035 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12036 * combine.c (simplify_set): Update accordingly.
12037 * emit-rtl.c (validate_subreg): Likewise.
12038 * recog.c (general_operand): Likewise.
12039 * regcprop.c (mode_change_ok): Likewise.
12040 * reload1.c (choose_reload_regs): Likewise.
12041 (inherit_piecemeal_p): Likewise.
12042 * rtlanal.c (simplify_subreg_regno): Likewise.
12043 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12044 instead of CANNOT_CHANGE_MODE_CLASS.
12045 (reload_cse_simplify_operands): Likewise.
12046 * reload.c (push_reload): Use targetm.can_change_mode_class
12047 instead of CANNOT_CHANGE_MODE_CLASS.
12048 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12049 REG_CANNOT_CHANGE_MODE_P.
12050 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12051 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12052 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12053 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12054 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12055 (arm_can_change_mode_class): New function.
12056 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12057 than CANNOT_CHANGE_MODE_CLASS in comments.
12058 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12059 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12060 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12061 (ix86_can_change_mode_class): ...this new function, inverting the
12062 sense of the return value.
12063 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12064 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12065 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12066 (ia64_can_change_mode_class): New function.
12067 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12068 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12069 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12070 (m32c_can_change_mode_class): ...this new function, inverting the
12071 sense of the return value.
12072 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12073 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12074 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12075 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12076 (mips_can_change_mode_class): ...this new function, inverting the
12077 sense of the return value.
12078 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12079 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12080 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12081 (msp430_can_change_mode_class): New function.
12082 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12083 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12084 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12085 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12086 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12087 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12088 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12089 (pa_cannot_change_mode_class): Replace with...
12090 (pa_can_change_mode_class): ...this new function, inverting the
12091 sense of the return value.
12092 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12093 than CANNOT_CHANGE_MODE_CLASS in comments.
12094 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12095 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12096 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12097 (pdp11_cannot_change_mode_class): Replace with...
12098 (pdp11_can_change_mode_class): ...this new function, inverting the
12099 sense of the return value.
12100 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12101 * config/powerpcspe/powerpcspe-protos.h
12102 (rs6000_cannot_change_mode_class_ptr): Delete.
12103 * config/powerpcspe/powerpcspe.c
12104 (rs6000_cannot_change_mode_class_ptr): Delete.
12105 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12106 (rs6000_option_override_internal): Assign to
12107 targetm.can_change_mode_class instead of
12108 rs6000_cannot_change_mode_class_ptr.
12109 (rs6000_cannot_change_mode_class): Replace with...
12110 (rs6000_can_change_mode_class): ...this new function, inverting the
12111 sense of the return value.
12112 (rs6000_debug_cannot_change_mode_class): Replace with...
12113 (rs6000_debug_can_change_mode_class): ...this new function.
12114 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12115 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12116 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12117 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12118 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12120 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12121 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12122 (rs6000_option_override_internal): Assign to
12123 targetm.can_change_mode_class instead of
12124 rs6000_cannot_change_mode_class_ptr.
12125 (rs6000_cannot_change_mode_class): Replace with...
12126 (rs6000_can_change_mode_class): ...this new function, inverting the
12127 sense of the return value.
12128 (rs6000_debug_cannot_change_mode_class): Replace with...
12129 (rs6000_debug_can_change_mode_class): ...this new function.
12130 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12131 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12132 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12133 (s390_can_change_mode_class): ...this new function, inverting the
12134 sense of the return value.
12135 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12136 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12137 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12138 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12139 (sh_cannot_change_mode_class): Replace with...
12140 (sh_can_change_mode_class): ...this new function, inverting the
12141 sense of the return value.
12142 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12143 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12144 (sparc_can_change_mode_class): New function.
12145 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12146 * config/spu/spu.c (spu_can_change_mode_class): New function.
12147 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12148 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12149 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12150 (visium_can_change_mode_class): New function.
12151 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12153 2017-09-15 Richard Biener <rguenther@suse.de>
12155 PR tree-optimization/82217
12156 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12157 but not undefined case.
12159 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12162 * postreload.c (reload_cse_simplify_operands): Skip
12163 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12165 2017-09-15 Richard Biener <rguenther@suse.de>
12167 PR tree-optimization/68823
12168 * graphite-scop-detection.c (build_alias_set): If we have a
12169 possible dependence check whether we can handle them by just
12170 looking at the DRs DR_ACCESS_FNs.
12171 (build_scops): If build_alias_set fails, fail the SCOP.
12173 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12175 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12176 to support float128 built-in functions that require the ISA 3.0
12178 (BU_FLOAT128_3_HW): Likewise.
12179 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12180 built-in functions.
12181 (FMAF128): Likewise.
12183 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12184 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12185 floating point instructions.
12186 (rs6000_invalid_builtin): Likewise.
12187 (rs6000_builtin_mask_names): Likewise.
12188 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12189 (RS6000_BTM_FLOAT128_HW): Likewise.
12190 (RS6000_BTM_COMMON): Likewise.
12191 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12193 * doc/extend.texi (RS/6000 built-in functions): Document the
12194 IEEE 128-bit floating point square root and fused multiply-add
12195 built-in functions.
12197 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12199 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12200 reg (r2) isn't in the set of registers defined in the prologue.
12202 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12203 Alan Hayward <alan.hayward@arm.com>
12204 David Sherwood <david.sherwood@arm.com>
12206 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12207 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12208 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12209 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12210 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12212 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12213 max_vectorization_factor.
12214 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12216 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12217 Alan Hayward <alan.hayward@arm.com>
12218 David Sherwood <david.sherwood@arm.com>
12220 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12221 (vect_worthwhile_without_simd_p): Declare.
12222 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12223 (vectorizable_reduction): Use it.
12224 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12225 (vectorizable_operation): Likewise.
12227 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12228 Alan Hayward <alan.hayward@arm.com>
12229 David Sherwood <david.sherwood@arm.com>
12231 * tree-vectorizer.h (vect_get_num_copies): New function.
12232 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12233 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12234 (vectorizable_induction): Likewise.
12235 (vectorizable_live_operation): Likewise.
12236 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12237 (vectorizable_bswap): Likewise.
12238 (vectorizable_call): Likewise.
12239 (vectorizable_conversion): Likewise.
12240 (vectorizable_assignment): Likewise.
12241 (vectorizable_shift): Likewise.
12242 (vectorizable_operation): Likewise.
12243 (vectorizable_store): Likewise.
12244 (vectorizable_load): Likewise.
12245 (vectorizable_condition): Likewise.
12246 (vectorizable_comparison): Likewise.
12247 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12249 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12250 Alan Hayward <alan.hayward@arm.com>
12251 David Sherwood <david.sherwood@arm.com>
12253 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
12254 of vect_init_vector.
12256 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12257 Alan Hayward <alan.hayward@arm.com>
12258 David Sherwood <david.sherwood@arm.com>
12260 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
12261 an inline wrapper that provides a location.
12262 (gimple_build_vector): Likewise.
12263 * gimple-fold.c (gimple_build_vector_from_val): New function.
12264 (gimple_build_vector): Likewise.
12265 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
12266 functions to build the initial value. Always return a gimple value.
12267 (get_initial_defs_for_reduction): Likewise. Only compute
12269 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
12270 vect_init_vector on the results from get_initial_def(s)_for_reduction.
12271 (vectorizable_induction): Use gimple_build_vector rather than
12274 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12275 Alan Hayward <alan.hayward@arm.com>
12276 David Sherwood <david.sherwood@arm.com>
12278 * target.h (vec_perm_indices): New typedef.
12279 (auto_vec_perm_indices): Likewise.
12280 * optabs-query.h: Include target.h
12281 (can_vec_perm_p): Take a vec_perm_indices *.
12282 * optabs-query.c (can_vec_perm_p): Likewise.
12283 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
12284 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12285 * tree-vect-generic.c (lower_vec_perm): Likewise.
12286 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
12287 (vect_grouped_load_supported): Likewise.
12288 (vect_shift_permute_load_chain): Likewise.
12289 (vect_permute_store_chain): Use auto_vec_perm_indices.
12290 (vect_permute_load_chain): Likewise.
12291 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
12292 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
12293 Update uses of can_vec_perm_p.
12294 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
12295 mode with a number of elements. Take a vec_perm_indices *.
12296 (vect_create_epilog_for_reduction): Update accordingly.
12297 Use auto_vec_perm_indices.
12298 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
12299 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
12300 (vect_transform_slp_perm_load): Likewise.
12301 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
12302 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
12303 (vect_gen_perm_mask_checked): Likewise.
12304 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
12305 (vect_gen_perm_mask_checked): Likewise.
12306 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
12307 (vectorizable_store): Likewise.
12308 (vectorizable_load): Likewise.
12309 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
12310 (vectorizable_bswap): Likewise.
12312 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12313 Alan Hayward <alan.hayward@arm.com>
12314 David Sherwood <david.sherwood@arm.com>
12316 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
12317 * tree.c (build_vector): Likewise.
12318 (build_vector_from_ctor): Update accordingly.
12319 (build_vector_from_val): Likewise.
12320 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12321 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
12322 * tree-vect-generic.c (add_rshift): Likewise.
12323 (expand_vector_divmod): Likewise.
12324 (optimize_vector_constructor): Likewise.
12325 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
12326 (vect_transform_slp_perm_load): Likewise.
12327 (vect_schedule_slp_instance): Likewise.
12328 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
12329 (vectorizable_call): Likewise.
12330 (vect_gen_perm_mask_any): Likewise. Add elements in order.
12331 * expmed.c (make_tree): Likewise.
12332 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
12333 a vector passed to build_vector.
12334 (fold_convert_const): Likewise.
12335 (exact_inverse): Likewise.
12336 (fold_ternary_loc): Likewise.
12337 (fold_relational_const): Likewise.
12338 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
12339 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
12340 (const_unop): Likewise. Store the reduction accumulator in a
12341 variable rather than an array.
12342 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
12343 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
12344 the new vector, rather than constructing it after the input arrays.
12345 (native_interpret_vector): Use auto_vec<tree> when building
12346 a vector passed to build_vector. Add elements in order.
12347 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
12348 auto_vec<tree> when building a vector passed to build_vector.
12349 (vect_create_epilog_for_reduction): Likewise.
12350 (vectorizable_induction): Likewise.
12351 (get_initial_def_for_reduction): Likewise. Fix indentation of
12353 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
12355 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
12356 passed to build_vector.
12358 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12359 Alan Hayward <alan.hayward@arm.com>
12360 David Sherwood <david.sherwood@arm.com>
12362 * tree-core.h (tree_base::u): Add an "nelts" field.
12363 (tree_vector): Use VECTOR_CST_NELTS as the length.
12364 * tree.c (tree_size): Likewise.
12365 (make_vector): Initialize VECTOR_CST_NELTS.
12366 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
12367 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
12368 TYPE_VECTOR_SUBPARTS.
12369 * expr.c (const_vector_mask_from_tree): Consistently use "units"
12370 as the number of units, setting it from VECTOR_CST_NELTS.
12371 (const_vector_from_tree): Likewise.
12372 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
12373 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
12374 (fold_negate_expr_1): Likewise.
12375 (fold_convert_const): Likewise.
12376 (const_binop): Likewise. Differentiate the number of output and
12378 (const_unop): Likewise.
12379 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
12380 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
12381 in cases that did the opposite.
12383 2017-09-14 Richard Biener <rguenther@suse.de>
12385 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
12388 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
12390 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
12392 2017-09-14 Jakub Jelinek <jakub@redhat.com>
12395 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
12396 if and where to split a bb, except for splitting before debug insn
12397 sequences followed by non-label real insn. Delete debug insns
12398 in between basic blocks.
12400 * combine.c (make_compound_operation_int): Formatting fixes.
12402 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
12403 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12404 * config/netbsd.h (LINK_EH_SPEC): Likewise.
12405 * config/sol2.h (LINK_EH_SPEC): Likewise.
12406 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12407 * config/s390/linux.h (LINK_SPEC): Likewise.
12408 * config/freebsd.h (LINK_EH_SPEC): Likewise.
12409 * config/openbsd.h (LINK_EH_SPEC): Likewise.
12410 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12411 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
12412 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
12413 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12414 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
12415 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
12417 2017-09-13 Jakub Jelinek <jakub@redhat.com>
12419 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
12421 (ENDFILE_LINUX_SPEC): Likewise.
12422 (LINK_EH_SPEC): Likewise.
12423 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
12424 (LINK_OS_LINUX_SPEC32): Likewise.
12425 (LINK_OS_LINUX_SPEC64): Likewise.
12426 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
12427 (LINK_OS_LINUX_SPEC): Likewise.
12429 2017-09-13 Martin Liska <mliska@suse.cz>
12431 PR middle-end/82154
12432 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
12433 CASE_HIGH is NULL_TREE.
12435 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12436 Alan Hayward <alan.hayward@arm.com>
12437 David Sherwood <david.sherwood@arm.com>
12439 * target.def (secondary_memory_needed): New hook.
12440 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
12441 instead of SECONDARY_MEMORY_NEEDED.
12442 (secondary_memory_needed_mode): Likewise.
12443 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
12444 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
12445 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
12446 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
12447 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
12448 * doc/tm.texi: Regenerate.
12449 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
12450 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
12451 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12452 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
12453 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
12454 * config/i386/i386.c (inline_secondary_memory_needed): Put the
12455 mode argument first and change the reg_class arguments to reg_class_t.
12456 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
12457 Make static. Update the call to inline_secondary_memory_needed.
12458 (ix86_register_move_cost): Update the call to
12459 inline_secondary_memory_needed.
12460 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12461 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
12463 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
12464 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12466 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
12467 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
12468 * config/mips/mips.c (mips_secondary_memory_needed): Make static
12469 and match hook interface. Add comment from mips.h.
12470 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12471 * config/mmix/mmix.md (truncdfsf2): Refer to
12472 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12474 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
12475 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
12476 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12477 (pa_secondary_memory_needed): New function.
12478 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
12479 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
12480 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12481 (pdp11_secondary_memory_needed): Make static and match hook interface.
12482 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
12483 * config/powerpcspe/powerpcspe-protos.h
12484 (rs6000_secondary_memory_needed_ptr): Delete.
12485 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
12487 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12488 (rs6000_option_override_internal): Assign to
12489 targetm.secondary_memory_needed rather than
12490 rs6000_secondary_memory_needed_ptr.
12491 (rs6000_secondary_memory_needed): Match hook interface.
12492 (rs6000_debug_secondary_memory_needed): Likewise.
12493 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
12494 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
12495 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
12496 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12497 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
12498 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
12500 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
12501 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12502 (rs6000_option_override_internal): Assign to
12503 targetm.secondary_memory_needed rather than
12504 rs6000_secondary_memory_needed_ptr.
12505 (rs6000_secondary_memory_needed): Match hook interface.
12506 (rs6000_debug_secondary_memory_needed): Likewise.
12507 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
12508 * config/s390/s390.c (s390_secondary_memory_needed): New function.
12509 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12510 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
12511 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
12512 (sparc_secondary_memory_needed): New function.
12513 * lra-constraints.c (check_and_process_move): Refer to
12514 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
12516 (curr_insn_transform): Likewise.
12517 (process_alt_operands): Use targetm.secondary_memory_needed
12518 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12519 (check_secondary_memory_needed_p): Likewise.
12520 (choose_split_class): Likewise.
12521 * reload.c: Unconditionally include code that was previously
12522 conditional on SECONDARY_MEMORY_NEEDED.
12523 (push_secondary_reload): Use targetm.secondary_memory_needed
12524 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12525 (push_reload): Likewise.
12526 * reload1.c: Unconditionally include code that was previously
12527 conditional on SECONDARY_MEMORY_NEEDED.
12528 (choose_reload_regs): Use targetm.secondary_memory_needed
12529 instead of TARGET_SECONDARY_MEMORY_NEEDED.
12530 (gen_reload): Likewise.
12531 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
12533 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
12534 Alan Hayward <alan.hayward@arm.com>
12535 David Sherwood <david.sherwood@arm.com>
12537 * target.def (secondary_memory_needed_mode): New hook:
12538 * targhooks.c (default_secondary_memory_needed_mode): Declare.
12539 * targhooks.h (default_secondary_memory_needed_mode): New function.
12540 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
12541 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
12542 * doc/tm.texi: Regenerate.
12543 * lra-constraints.c (check_and_process_move): Use
12544 targetm.secondary_memory_needed_mode instead of
12545 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
12546 (curr_insn_transform): Likewise.
12547 * reload.c (get_secondary_mem): Likewise.
12548 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12549 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
12551 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12552 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12553 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
12554 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12555 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
12557 * config/powerpcspe/powerpcspe-protos.h
12558 (rs6000_secondary_memory_needed_mode): Delete.
12559 * config/powerpcspe/powerpcspe.c
12560 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12561 (rs6000_secondary_memory_needed_mode): Make static.
12562 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12563 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
12565 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12567 (rs6000_secondary_memory_needed_mode): Make static.
12568 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12569 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
12570 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
12571 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
12572 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
12574 (sparc_secondary_memory_needed_mode): New function.
12575 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
12577 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
12579 * config/aarch64/constraints.md (Umq): New constraint.
12580 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
12582 (mov<mode>): Update condition.
12584 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12586 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
12587 when bitposition is the same.
12589 2017-09-13 Richard Biener <rguenther@suse.de>
12591 * dwarf2out.c (output_die_symbol): Remove.
12592 (output_die): Do not output a DIEs symbol.
12594 2017-09-13 Richard Biener <rguenther@suse.de>
12596 PR middle-end/82128
12597 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
12598 default-def to avoid breaking iterator update with the weird
12599 interaction with cgraph_update_edges_for_call_stmt_node.
12601 2017-09-13 Richard Biener <rguenther@suse.de>
12603 * tree-cfg.c (verify_gimple_assign_binary): Add verification
12604 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
12605 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
12606 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
12608 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12610 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
12611 Disable pc relative literal load irrespective of
12612 TARGET_FIX_ERR_A53_84341 for default.
12614 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
12616 * config/sparc/sparc.c (output_return): Output the source location of
12617 the insn in the delay slot, if any.
12618 (output_sibcall): Likewise.
12620 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
12623 * common.opt (-static-pie): New alias.
12624 (shared): Negate static-pie.
12625 (-no-pie): Update help text.
12627 (static-pie): New option.
12628 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
12629 -static-pie support.
12630 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
12631 (LINK_EH_SPEC): Likewise.
12632 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
12633 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12634 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
12635 * gcc.c (LINK_COMMAND_SPEC): Likewise.
12636 (init_gcc_specs): Likewise.
12637 (init_spec): Likewise.
12638 (display_help): Update help message for -pie.
12639 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
12642 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
12644 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
12645 (movdi_aarch64): Likewise.
12646 (movti_aarch64): Likewise.
12648 2017-09-12 Simon Wright <simon@pushface.org>
12651 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
12652 calculation of the minor version, always output as 0.
12654 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12657 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12658 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
12659 on it early, rather than manual conversion late. For
12660 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
12661 instead of performing manual conversion.
12663 2017-09-12 Carl Love <cel@us.ibm.com>
12665 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
12666 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
12668 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12669 VMULOSW): Add definitions.
12670 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12671 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12672 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12673 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12674 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12676 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12678 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
12680 (movti_aarch64): Likewise.
12681 (movdf_aarch64): Likewise.
12682 (movtf_aarch64): Likewise.
12683 (load_pairdi): Likewise.
12684 (store_pairdi): Likewise.
12685 (load_pairdf): Likewise.
12686 (store_pairdf): Likewise.
12687 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
12688 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
12689 (ldr_got_small_<mode>): Likewise.
12690 (ldr_got_small_28k_<mode>): Likewise.
12691 (ldr_got_tiny): Likewise.
12692 * config/aarch64/iterators.md (ldst_sz): New.
12693 (ldpstp_sz): Likewise.
12694 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
12696 (thunderx_load): Split load_8 to load_16.
12697 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
12699 (thunderx2t99_storepair_basic): Split store_8 to store_16.
12700 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
12701 (xgene1_store_pair): Split store_8 to store_16.
12702 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
12703 (falkor_st_0_st_sd): Split store_8 to store_16.
12705 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
12707 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
12708 and store1/2/3/4 to store_4/8/12/16.
12709 * config/aarch64/aarch64.md: Update for rename.
12710 * config/arm/arm.md: Likewise.: Likewise.
12711 * config/arm/arm.c: Likewise.
12712 * config/arm/thumb1.md: Likewise.
12713 * config/arm/thumb2.md: Likewise.
12714 * config/arm/vfp.md: Likewise.
12715 * config/arm/arm-generic.md: Likewise.
12716 * config/arm/arm1020e.md: Likewise.
12717 * config/arm/arm1026ejs.md: Likewise.
12718 * config/arm/arm1136jfs.md: Likewise.
12719 * config/arm/arm926ejs.md: Likewise.
12720 * config/arm/cortex-a15.md: Likewise.
12721 * config/arm/cortex-a17.md: Likewise.
12722 * config/arm/cortex-a5.md: Likewise.
12723 * config/arm/cortex-a53.md: Likewise.
12724 * config/arm/cortex-a57.md: Likewise.
12725 * config/arm/cortex-a7.md: Likewise.
12726 * config/arm/cortex-a8.md: Likewise.
12727 * config/arm/cortex-a9.md: Likewise.
12728 * config/arm/cortex-m4.md: Likewise.
12729 * config/arm/cortex-m7.md: Likewise.
12730 * config/arm/cortex-r4.md: Likewise.
12731 * config/arm/exynos-m1.md: Likewise.
12732 * config/arm/fa526.md: Likewise.
12733 * config/arm/fa606te.md: Likewise.
12734 * config/arm/fa626te.md: Likewise.
12735 * config/arm/fa726te.md: Likewise.
12736 * config/arm/fmp626.md: Likewise.
12737 * config/arm/iwmmxt.md: Likewise.
12738 * config/arm/ldmstm.md: Likewise.
12739 * config/arm/marvell-pj4.md: Likewise.
12740 * config/arm/xgene1.md: Likewise.
12741 * config/aarch64/thunderx.md: Likewise.
12742 * config/aarch64/thunderx2t99.md: Likewise.
12743 * config/aarch64/falkor.md: Likewise.
12745 2017-09-12 Martin Liska <mliska@suse.cz>
12747 * attribs.c (private_lookup_attribute): New function.
12748 * attribs.h (private_lookup_attribute): Declared here.
12749 (lookup_attribute): Called from this place.
12751 2017-09-12 Richard Biener <rguenther@suse.de>
12753 PR tree-optimization/82157
12754 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
12755 stmts with side-effects.
12757 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12758 Alan Hayward <alan.hayward@arm.com>
12759 David Sherwood <david.sherwood@arm.com>
12761 * target.def (hard_regno_nregs): New hook.
12762 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
12763 * targhooks.h (default_hard_regno_nregs): Declare.
12764 * targhooks.c (default_hard_regno_nregs): New function.
12765 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
12766 (TARGET_HARD_REGNO_NREGS): ...this hook.
12767 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
12768 (CLASS_MAX_NREGS): Likewise.
12769 * doc/tm.texi: Regenerate.
12770 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
12771 instead of HARD_REGNO_NREGS.
12772 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
12773 HARD_REGNO_NREGS in the comment.
12774 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
12775 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
12776 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
12777 Return an unsigned int.
12778 (TARGET_HARD_REGNO_NREGS): Redefine.
12779 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
12780 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
12781 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12782 (arc_hard_regno_nregs): New function.
12783 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
12784 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
12785 (arm_hard_regno_nregs): New function.
12786 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
12787 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
12788 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
12789 (TARGET_HARD_REGNO_NREGS): Redefine.
12790 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
12791 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
12792 (HARD_REGNO_NREGS): Delete.
12793 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
12794 (cr16_hard_regno_nregs): New function.
12795 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
12796 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
12797 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
12798 (cris_hard_regno_nregs): New function.
12799 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
12800 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
12801 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12802 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
12803 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
12804 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
12805 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
12806 (frv_hard_regno_nregs): Make static. Take and return an
12808 (frv_class_max_nregs): Remove outdated copy of documentation.
12809 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
12810 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
12811 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
12812 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
12813 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
12814 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
12815 (TARGET_HARD_REGNO_NREGS): Redefine.
12816 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
12817 (CLASS_MAX_NREGS): Update comment.
12818 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
12819 (ia64_hard_regno_nregs): New function.
12820 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
12821 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
12822 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
12823 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
12824 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
12826 (m32c_hard_regno_nregs): Likewise. Make static.
12827 (TARGET_HARD_REGNO_NREGS): Redefine.
12828 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
12829 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
12830 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
12831 (m68k_hard_regno_nregs): New function.
12832 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
12833 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
12834 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
12835 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
12836 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
12837 Take and return an unsigned int.
12838 (TARGET_HARD_REGNO_NREGS): Redefine.
12839 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
12840 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
12841 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
12842 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
12843 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
12844 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
12845 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
12846 (msp430_hard_regno_nregs): Make static. Take and return an
12848 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
12849 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
12850 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
12851 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
12852 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
12853 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
12854 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
12855 (TARGET_HARD_REGNO_NREGS): Redefine.
12856 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
12857 (PA_HARD_REGNO_NREGS): ...this.
12858 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
12859 (PA_HARD_REGNO_NREGS): ...this.
12860 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12861 (pa_hard_regno_nregs): New function.
12862 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
12863 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
12864 (pdp11_hard_regno_nregs): New function.
12865 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
12866 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
12867 (rs6000_hard_regno_nregs_hook): New function.
12868 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
12869 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
12870 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
12871 Take and return an unsigned int. Move earlier in file.
12872 (TARGET_HARD_REGNO_NREGS): Redefine.
12873 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
12874 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
12875 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
12876 (rl78_hard_regno_nregs): Make static. Take and return an
12878 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
12879 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
12880 (rs6000_hard_regno_nregs_hook): New function.
12881 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
12882 * config/rx/rx.c (rx_hard_regno_nregs): New function.
12883 (TARGET_HARD_REGNO_NREGS): Redefine.
12884 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
12885 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12886 instead of HARD_REGNO_NREGS.
12887 (s390_hard_regno_nregs): New function.
12888 (s390_hard_regno_mode_ok): Add comment from s390.h.
12889 (TARGET_HARD_REGNO_NREGS): Redefine.
12890 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12891 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12892 (sh_hard_regno_nregs): New function.
12893 (sh_pass_in_reg_p): Use it.
12894 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12895 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12896 (sparc_hard_regno_nregs): New function.
12897 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12898 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12899 (spu_function_arg_advance): Use it, supplying a valid register number.
12900 (TARGET_HARD_REGNO_NREGS): Redefine.
12901 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12902 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12903 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12904 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12905 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12906 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12907 (CLASS_MAX_NREGS): Remove copy of old documentation.
12908 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12909 (visium_hard_regno_nregs): New function.
12910 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12911 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12912 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12913 xtensa_hard_regno_nregs): New function.
12914 * system.h (HARD_REGNO_NREGS): Poison.
12916 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12918 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12919 hard_regno_nregs instead of HARD_REGNO_NREGS.
12920 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12921 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12922 (c6x_expand_epilogue): Likewise.
12923 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12924 (frv_read_iacc_argument): Likewise.
12925 * config/sh/sh.c: Include regs.h.
12926 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12927 (regs_used): Likewise.
12928 (output_stack_adjust): Likewise.
12929 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12930 * expmed.c: Include regs.h.
12931 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12932 * ree.c: Include regs.h.
12933 (combine_reaching_defs): Use hard_regno_nregs instead of
12935 (add_removable_extension): Likewise.
12937 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12939 * regs.h (hard_regno_nregs): Turn into a function.
12940 (end_hard_regno): Update accordingly.
12941 * caller-save.c (setup_save_areas): Likewise.
12942 (save_call_clobbered_regs): Likewise.
12943 (replace_reg_with_saved_mem): Likewise.
12944 (insert_restore): Likewise.
12945 (insert_save): Likewise.
12946 * combine.c (can_change_dest_mode): Likewise.
12947 (move_deaths): Likewise.
12948 (distribute_notes): Likewise.
12949 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12950 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12951 (rs6000_split_multireg_move): Likewise.
12952 (rs6000_register_move_cost): Likewise.
12953 (rs6000_memory_move_cost): Likewise.
12954 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12955 (rs6000_split_multireg_move): Likewise.
12956 (rs6000_register_move_cost): Likewise.
12957 (rs6000_memory_move_cost): Likewise.
12958 * cselib.c (cselib_reset_table): Likewise.
12959 (cselib_lookup_1): Likewise.
12960 * emit-rtl.c (set_mode_and_regno): Likewise.
12961 * function.c (aggregate_value_p): Likewise.
12962 * ira-color.c (setup_profitable_hard_regs): Likewise.
12963 (check_hard_reg_p): Likewise.
12964 (calculate_saved_nregs): Likewise.
12965 (assign_hard_reg): Likewise.
12966 (improve_allocation): Likewise.
12967 (calculate_spill_cost): Likewise.
12968 * ira-emit.c (modify_move_list): Likewise.
12969 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12970 (ira_hard_reg_in_set_p): Likewise.
12971 * ira.c (setup_reg_mode_hard_regset): Likewise.
12972 (clarify_prohibited_class_mode_regs): Likewise.
12973 (check_allocation): Likewise.
12974 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12975 (lra_setup_reg_renumber): Likewise.
12976 (setup_try_hard_regno_pseudos): Likewise.
12977 (spill_for): Likewise.
12978 (assign_hard_regno): Likewise.
12979 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12980 * lra-constraints.c (in_class_p): Likewise.
12981 (lra_constraint_offset): Likewise.
12982 (simplify_operand_subreg): Likewise.
12983 (lra_constraints): Likewise.
12984 (split_reg): Likewise.
12985 (split_if_necessary): Likewise.
12986 (invariant_p): Likewise.
12987 (inherit_in_ebb): Likewise.
12988 * lra-lives.c (process_bb_lives): Likewise.
12989 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12990 (get_hard_regs): Likewise.
12991 (do_remat): Likewise.
12992 * lra-spills.c (assign_spill_hard_regs): Likewise.
12993 * mode-switching.c (create_pre_exit): Likewise.
12994 * postreload.c (reload_combine_recognize_pattern): Likewise.
12995 * recog.c (peep2_find_free_register): Likewise.
12996 * regcprop.c (kill_value_regno): Likewise.
12997 (set_value_regno): Likewise.
12998 (copy_value): Likewise.
12999 (maybe_mode_change): Likewise.
13000 (find_oldest_value_reg): Likewise.
13001 (copyprop_hardreg_forward_1): Likewise.
13002 * regrename.c (check_new_reg_p): Likewise.
13003 (regrename_do_replace): Likewise.
13004 * reload.c (push_reload): Likewise.
13005 (combine_reloads): Likewise.
13006 (find_dummy_reload): Likewise.
13007 (operands_match_p): Likewise.
13008 (find_reloads): Likewise.
13009 (find_equiv_reg): Likewise.
13010 (reload_adjust_reg_for_mode): Likewise.
13011 * reload1.c (count_pseudo): Likewise.
13012 (count_spilled_pseudo): Likewise.
13013 (find_reg): Likewise.
13014 (clear_reload_reg_in_use): Likewise.
13015 (free_for_value_p): Likewise.
13016 (allocate_reload_reg): Likewise.
13017 (choose_reload_regs): Likewise.
13018 (reload_adjust_reg_for_temp): Likewise.
13019 (emit_reload_insns): Likewise.
13020 (delete_output_reload): Likewise.
13021 * rtlanal.c (subreg_get_info): Likewise.
13022 * sched-deps.c (sched_analyze_reg): Likewise.
13023 * sel-sched.c (init_regs_for_mode): Likewise.
13024 (mark_unavailable_hard_regs): Likewise.
13025 (choose_best_reg_1): Likewise.
13026 (verify_target_availability): Likewise.
13027 * valtrack.c (dead_debug_insert_temp): Likewise.
13028 * var-tracking.c (track_loc_p): Likewise.
13029 (emit_note_insn_var_location): Likewise.
13030 * varasm.c (make_decl_rtl): Likewise.
13031 * reginfo.c (choose_hard_reg_mode): Likewise.
13032 (init_reg_modes_target): Refer directly to
13033 this_target_regs->x_hard_regno_nregs.
13035 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13037 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13038 instead of hard_regno_nregs.
13040 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13042 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13043 end_hard_regno instead of hard_regno_nregs.
13044 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13045 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13046 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13047 * ira-color.c (improve_allocation): Likewise.
13048 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13049 * lra-lives.c (mark_regno_live): Likewise.
13050 (mark_regno_dead): Likewise.
13051 * lra-remat.c (operand_to_remat): Likewise.
13052 * lra.c (collect_non_operand_hard_regs): Likewise.
13053 * postreload.c (reload_combine_note_store): Likewise.
13054 (move2add_valid_value_p): Likewise.
13055 * reload.c (regno_clobbered_p): Likewise.
13057 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13059 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13061 * config/v850/v850.c (v850_reorg): Likewise.
13062 * reload.c (refers_to_regno_for_reload_p): Likewise.
13063 (find_equiv_reg): Likewise.
13064 * reload1.c (reload_reg_reaches_end_p): Likewise.
13066 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13068 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13070 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13071 * config/arm/arm.c (output_move_neon): Likewise.
13072 (arm_attr_length_move_neon): Likewise.
13073 (neon_split_vcombine): Likewise.
13074 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13075 (c6x_mark_reg_written): Likewise.
13076 (c6x_dwarf_register_span): Likewise.
13077 * config/i386/i386.c (ix86_save_reg): Likewise.
13078 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13079 (rws_access_reg): Likewise.
13080 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13081 * mode-switching.c (create_pre_exit): Likewise.
13082 * ree.c (combine_reaching_defs): Likewise.
13083 (add_removable_extension): Likewise.
13084 * regcprop.c (find_oldest_value_reg): Likewise.
13085 (copyprop_hardreg_forward_1): Likewise.
13086 * reload.c (reload_inner_reg_of_subreg): Likewise.
13087 (push_reload): Likewise.
13088 (combine_reloads): Likewise.
13089 (find_dummy_reload): Likewise.
13090 (reload_adjust_reg_for_mode): Likewise.
13091 * reload1.c (find_reload_regs): Likewise.
13092 (forget_old_reloads_1): Likewise.
13093 (reload_reg_free_for_value_p): Likewise.
13094 (reload_adjust_reg_for_temp): Likewise.
13095 (emit_reload_insns): Likewise.
13096 (delete_output_reload): Likewise.
13097 * sel-sched.c (choose_best_reg_1): Likewise.
13098 (choose_best_pseudo_reg): Likewise.
13100 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13101 Alan Hayward <alan.hayward@arm.com>
13102 David Sherwood <david.sherwood@arm.com>
13104 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13105 * target.def (slow_unaligned_access): New hook.
13106 * targhooks.h (default_slow_unaligned_access): Declare.
13107 * targhooks.c (default_slow_unaligned_access): New function.
13108 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13109 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13110 * doc/tm.texi: Regenerate.
13111 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13112 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13113 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13115 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13116 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13118 (rs6000_slow_unaligned_access): New function.
13119 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13120 (expand_block_compare): Likewise.
13121 (expand_strn_compare): Likewise.
13122 (rs6000_rtx_costs): Likewise.
13123 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13124 (riscv_slow_unaligned_access): Likewise.
13125 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13126 (riscv_slow_unaligned_access_p): ...this and make static.
13127 (riscv_option_override): Update accordingly.
13128 (riscv_slow_unaligned_access): New function.
13129 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13130 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13131 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13132 (rs6000_slow_unaligned_access): New function.
13133 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13134 (rs6000_rtx_costs): Likewise.
13135 * config/rs6000/rs6000-string.c (expand_block_compare)
13136 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13137 of SLOW_UNALIGNED_ACCESS.
13138 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13139 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13140 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13141 of SLOW_UNALIGNED_ACCESS.
13142 * expmed.c (simple_mem_bitfield_p): Likewise.
13143 * expr.c (alignment_for_piecewise_move): Likewise.
13144 (emit_group_load_1): Likewise.
13145 (emit_group_store): Likewise.
13146 (copy_blkmode_from_reg): Likewise.
13147 (emit_push_insn): Likewise.
13148 (expand_assignment): Likewise.
13149 (store_field): Likewise.
13150 (expand_expr_real_1): Likewise.
13151 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13152 * lra-constraints.c (simplify_operand_subreg): Likewise.
13153 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13154 * gimple-ssa-store-merging.c: Likewise in block comment at start
13156 * tree-ssa-strlen.c: Include target.h.
13157 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13158 of SLOW_UNALIGNED_ACCESS.
13159 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13161 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13163 PR rtl-optimization/82185
13164 * expmed.c (emit_store_flag_int): Only test tem if it has been
13167 2017-09-12 Richard Biener <rguenther@suse.de>
13169 PR middle-end/82149
13170 * match.pd ((FTYPE) N CMP CST): Fix typo.
13172 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13174 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13176 (mips_near_type_p): Add 'short_call' attribute as a synonym
13178 * doc/extend.texi (short_call): Document new function attribute.
13180 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13183 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13184 assertion check that in the condition.
13185 (get_atomic_generic_size): Likewise. Before testing if parameter
13186 has pointer type, if it has array type, call for C++
13187 default_conversion to perform array-to-pointer conversion.
13189 2017-09-12 Richard Biener <rguenther@suse.de>
13191 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13192 for operations we cannot scalarize.
13194 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13196 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13197 vectors heap vectors. Clean up comments.
13198 Make visited_bbs a reference.
13199 (profitable_jump_thread_path): Make GC
13200 vectors heap vectors. Clean up comments.
13202 (convert_and_register_jump_thread_path): Make GC vectors heap
13204 (check_subpath_and_update_thread_path): Same. Clean up comments.
13205 Make visited_bbs a reference.
13206 (handle_phi): Abstract common code to to
13207 register_jump_thread_path_if_profitable.
13208 Rename VAR_BB to DEF_BB.
13210 Make GC vectors heap vectors.
13211 Make visited_bbs a reference.
13212 (handle_assignment): Same.
13213 (register_jump_thread_path_if_profitable): New.
13214 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13216 Make GC vectors heap vectors. Clean up comments.
13217 Make visited_bbs a reference.
13218 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13219 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13222 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13225 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13226 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13228 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13230 Revert r251800 and r251799.
13232 2017-09-11 Martin Jambor <mjambor@suse.cz>
13235 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13236 arguments in advance.
13237 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13238 use it to find predecessor edges.
13239 (naive_outof_ssa): Collect vector of predecessors.
13241 2017-09-08 Jason Merrill <jason@redhat.com>
13243 PR c++/70029 - ICE with ref-qualifier and -flto
13244 * langhooks.h (struct lang_hooks_for_types): Add
13245 copy_lang_qualifiers.
13246 * attribs.c (build_type_attribute_qual_variant): Use it.
13247 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13249 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
13250 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
13252 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
13255 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
13256 (*mulsi3_sp64): New instruction.
13257 (mulsi3): New expander.
13259 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
13261 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
13263 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13265 * sancov.c: Include memmodel.h.
13267 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
13270 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
13273 2017-09-07 Carl Love <cel@us.ibm.com>
13275 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
13276 the sldi instruction.
13278 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
13280 * sancov.c: Include tm_p.h.
13282 2017-09-07 Jakub Jelinek <jakub@redhat.com>
13285 * output.h (switch_to_other_text_partition): New declaration.
13286 * varasm.c (switch_to_other_text_partition): New function.
13287 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
13288 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
13289 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13290 to the other text partition before emitting LCL label and switch back
13291 after emitting the word after it.
13293 2017-09-07 Richard Biener <rguenther@suse.de>
13295 * passes.def (pass_split_crit_edges): Remove instance before PRE.
13296 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
13297 critical edges here, after loop init.
13298 (pass_data_pre): Remove PROP_no_crit_edges flags.
13299 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
13300 for valueization of call args to avoid leaking VN_TOP.
13301 (visit_use): Assert we do not visit default defs.
13302 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
13303 Use error_mark_node to more easily detect leaking VN_TOP.
13304 All default-defs are varying, not VN_TOP. Mark them visited.
13305 (run_scc_vn): Make code match comment.
13307 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13309 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
13310 OPTION_MASK_FLOAT128_KEYWORD.
13311 (POWERPC_MASKS): Likewise.
13312 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
13313 support for the -mfloat128-type option, and make -mfloat128
13314 default on PowerPC Linux systems. Define or undefine
13315 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
13316 Define __float128 to be __ieee128 if IEEE 128-bit support is
13317 enabled, or undefine it.
13318 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
13319 Delete defining __FLOAT128_TYPE__.
13320 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
13321 -mfloat128-type option and make -mfloat128 default on PowerPC
13323 (TARGET_FLOAT128_TYPE): Likewise.
13324 (-mfloat128-type): Likewise.
13325 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13326 Delete the -mfloat128-type option and make -mfloat128 default on
13327 PowerPC Linux systems. Always use __ieee128 to be the keyword for
13328 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
13329 128-bit floating point is enabled. Change tests from using
13330 -mfloat128-type to -mfloat128.
13331 (rs6000_mangle_type): Use the correct mangling for the __float128
13332 type even if normal long double is restricted to 64-bits.
13333 (floatn_mode): Enable the _Float128 type by default on VSX Linux
13335 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
13336 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
13337 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
13339 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
13340 documentation for -mfloat128.
13342 2017-09-06 Olivier Hainque <hainque@adacore.com>
13344 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
13346 2017-09-06 Wish Wu <wishwu007@gmail.com>
13347 Jakub Jelinek <jakub@redhat.com>
13349 * asan.c (initialize_sanitizer_builtins): Add
13350 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
13351 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
13352 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
13353 BT_FN_VOID_UINT64_PTR variables.
13354 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
13355 (BT_FN_VOID_UINT16_UINT16): Likewise.
13356 (BT_FN_VOID_UINT32_UINT32): Likewise.
13357 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13358 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
13359 (BT_FN_VOID_UINT64_PTR): Likewise.
13360 * common.opt (flag_sanitize_coverage): New variable.
13361 (fsanitize-coverage=trace-pc): Remove.
13362 (fsanitize-coverage=): Add.
13363 * flag-types.h (enum sanitize_coverage_code): New enum.
13364 * fold-const.c (fold_range_test): Disable non-short-circuit
13365 optimization if flag_sanitize_coverage.
13366 (fold_truth_andor): Likewise.
13367 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13368 * opts.c (COVERAGE_SANITIZER_OPT): Define.
13369 (coverage_sanitizer_opts): New array.
13370 (get_closest_sanitizer_option): Add OPTS argument, handle also
13371 OPT_fsanitize_coverage_.
13372 (parse_sanitizer_options): Adjusted to also handle
13373 OPT_fsanitize_coverage_.
13374 (common_handle_option): Add OPT_fsanitize_coverage_.
13375 * sancov.c (instrument_comparison, instrument_switch): New function.
13376 (sancov_pass): Add trace-cmp support.
13377 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
13378 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
13379 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
13380 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
13381 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
13382 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
13383 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
13384 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
13385 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
13386 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
13388 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13390 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13391 error. Only quit immediately if parsing is complete.
13392 (BEGIN): Initialize fatal_err and parse_done.
13393 (begin fpu, end fpu): Check number of arguments.
13394 (begin arch, end arch): Likewise.
13395 (begin cpu, end cpu): Likewise.
13396 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13397 (optalias): Likewise.
13399 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
13401 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13402 * config/arm/arm-isa.h: Delete. Move definitions to ...
13403 * arm-cpus.in: ... here. Use new feature and fgroup values.
13404 * config/arm/arm.c (arm_option_override): Use lower case for feature
13406 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13407 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13408 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13410 (print_isa_bits_for): New function.
13411 (gen_isa): New function.
13412 (gen_comm_data): Use print_isa_bits_for.
13413 (define feature): New keyword.
13414 (define fgroup): New keyword.
13415 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
13416 (arm-isa.h): Add rule to generate file.
13417 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13418 case for feature bit names.
13420 2017-09-06 Richard Biener <rguenther@suse.de>
13422 * tree-ssa-pre.c (NECESSARY): Remove.
13423 (create_expression_by_pieces): Do not touch pass-local flags.
13424 (insert_into_preds_of_block): Likewise.
13425 (do_pre_regular_insertion): Likewise.
13426 (eliminate_insert): Likewise.
13427 (eliminate_dom_walker::before_dom_children): Likewise.
13428 (fini_eliminate): Do not look at inserted_exprs.
13429 (mark_operand_necessary): Remove.
13430 (remove_dead_inserted_code): Replace with simple work-list
13431 algorithm based on inserted_exprs and SSA uses.
13432 (pass_pre::execute): Re-order fini_eliminate and
13433 remove_dead_inserted_code.
13435 2017-09-06 Olivier Hainque <hainque@adacore.com>
13437 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
13438 for VxWorks 7. Adjust surrounding comments.
13440 2017-09-06 Richard Biener <rguenther@suse.de>
13442 * gimple-ssa-strength-reduction.c
13443 (find_candidates_dom_walker::before_dom_children): Also allow
13446 2017-09-06 Richard Biener <rguenther@suse.de>
13448 PR tree-optimization/82108
13449 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
13450 for gap in the non-permutation SLP case.
13452 2017-09-06 Martin Jambor <mjambor@suse.cz>
13454 PR tree-optimization/82078
13455 * tree-sra.c (sort_and_splice_var_accesses): Move call to
13456 add_access_to_work_queue...
13457 (build_accesses_from_assign): ...here.
13458 (propagate_all_subaccesses): Make sure racc is the group
13459 representative, if there is one.
13461 2017-09-06 Jakub Jelinek <jakub@redhat.com>
13463 PR middle-end/82095
13464 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
13467 2017-09-06 Richard Biener <rguenther@suse.de>
13469 * gimple-ssa-strength-reduction.c
13470 (find_candidates_dom_walker::before_doom_children): Use a
13471 type and not a mode check.
13473 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
13476 * config/arm/predicates.md (arm_general_adddi_operand): Create new
13478 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
13480 2017-09-05 Jeff Law <law@redhat.com>
13482 PR tree-optimization/64910
13483 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
13484 cases where we have 3 or more operands.
13486 2017-09-05 Jakub Jelinek <jakub@redhat.com>
13488 PR middle-end/81768
13489 * omp-low.c (lower_omp_for): Recompute tree invariant if
13490 gimple_omp_for_initial/final is ADDR_EXPR.
13492 PR middle-end/81768
13493 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
13494 into gimple val before gimplification fo the COND_EXPR.
13496 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
13498 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
13499 REGION_COPY argument.
13500 (thread_through_all_blocks): Remove unused argument to
13501 duplicate_thread_path.
13503 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13504 Alan Hayward <alan.hayward@arm.com>
13505 David Sherwood <david.sherwood@arm.com>
13507 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
13508 Take a scalar_mode rather than a machine_mode.
13509 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13510 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
13511 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13512 (aarch64_gen_adjusted_ldpstp): Likewise.
13513 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
13515 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13516 Alan Hayward <alan.hayward@arm.com>
13517 David Sherwood <david.sherwood@arm.com>
13519 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
13520 Take a scalar_int_mode instead of a machine_mode.
13521 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13522 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13523 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13524 (aarch64_simd_attr_length_rglist): Delete.
13525 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
13526 a scalar_int_mode instead of a machine_mode.
13527 (aarch64_add_offset): Likewise.
13528 (aarch64_internal_mov_immediate): Likewise
13529 (aarch64_add_constant_internal): Likewise.
13530 (aarch64_add_constant): Likewise.
13531 (aarch64_movw_imm): Likewise.
13532 (aarch64_rtx_arith_op_extract_p): Likewise.
13533 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
13534 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
13535 Remove assert that the mode isn't a vector.
13536 (aarch64_output_scalar_simd_mov_immediate): Likewise.
13537 (aarch64_expand_mov_immediate): Update calls after above changes.
13538 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
13539 (aarch64_and_bitmask_imm): Check for scalar integer modes.
13540 (aarch64_move_imm): Likewise.
13541 (aarch64_can_const_movi_rtx_p): Likewise.
13542 (aarch64_strip_extend): Likewise.
13543 (aarch64_extr_rtx_p): Likewise.
13544 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
13545 a CONST_INT when the mode parameter is VOIDmode.
13546 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
13548 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13550 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
13551 * stor-layout.c (bitwise_mode_for_mode): Likewise.
13552 (bitwise_type_for_mode): Update accordingly.
13554 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13556 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
13557 * stor-layout.c (mode_for_size_tree): Likewise.
13558 (mode_for_array): Update accordingly.
13559 (layout_decl): Likewise.
13560 (compute_record_mode): Likewise. Only set the mode once.
13562 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13564 * target.def (get_mask_mode): Change return type to opt_mode.
13566 * doc/tm.texi: Regenerate.
13567 * targhooks.h (default_get_mask_mode): Return an opt_mode.
13568 * targhooks.c (default_get_mask_mode): Likewise.
13569 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
13570 * optabs-query.c (can_vec_mask_load_store_p): Update use of
13571 targetm.get_mask_mode.
13572 * tree.c (build_truth_vector_type): Likewise.
13574 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13576 * machmode.h (mode_for_vector): Return an opt_mode.
13577 * stor-layout.c (mode_for_vector): Likewise.
13578 (mode_for_int_vector): Update accordingly.
13579 (layout_type): Likewise.
13580 * config/i386/i386.c (emit_memmov): Likewise.
13581 (ix86_expand_set_or_movmem): Likewise.
13582 (ix86_expand_vector_init): Likewise.
13583 (ix86_get_mask_mode): Likewise.
13584 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
13586 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
13587 * expmed.c (extract_bit_field_1): Likewise.
13588 * expr.c (expand_expr_real_2): Likewise.
13589 * optabs-query.c (can_vec_perm_p): Likewise.
13590 (can_vec_mask_load_store_p): Likewise.
13591 * optabs.c (expand_vec_perm): Likewise.
13592 * targhooks.c (default_get_mask_mode): Likewise.
13593 * tree-vect-stmts.c (vectorizable_store): Likewise.
13594 (vectorizable_load): Likewise.
13595 (get_vectype_for_scalar_type_and_size): Likewise.
13597 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13599 * machmode.h (mode_for_int_vector): New function.
13600 * stor-layout.c (mode_for_int_vector): Likewise.
13601 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
13602 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
13603 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
13604 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
13605 (s390_expand_vcond): Likewise.
13607 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13609 * machmode.h (opt_machine_mode): New type.
13610 (opt_mode<T>): Allow construction from anything that can be
13612 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
13613 (mode_for_size): Return an opt_machine_mode.
13614 * stor-layout.c (mode_for_size): Likewise.
13615 (mode_for_size_tree): Update call accordingly.
13616 (bitwise_mode_for_mode): Likewise.
13617 (make_fract_type): Likewise.
13618 (make_accum_type): Likewise.
13619 * caller-save.c (replace_reg_with_saved_mem): Update call
13621 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13622 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13623 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13624 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
13625 * expmed.c (extract_bit_field_1): Likewise.
13626 * reload.c (get_secondary_mem): Likewise.
13627 * varasm.c (assemble_integer): Likewise.
13628 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
13631 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13633 * machmode.h (decimal_float_mode_for_size): New function.
13634 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
13635 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
13636 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
13637 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
13638 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
13639 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
13641 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13643 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
13644 (get_builtin_sync_mode): Likewise.
13645 (expand_ifn_atomic_compare_exchange): Likewise.
13646 (expand_builtin_atomic_clear): Likewise.
13647 (expand_builtin_atomic_test_and_set): Likewise.
13648 (fold_builtin_atomic_always_lock_free): Likewise.
13649 * calls.c (compute_argument_addresses): Likewise.
13650 (emit_library_call_value_1): Likewise.
13651 (store_one_arg): Likewise.
13652 * combine.c (combine_instructions): Likewise.
13653 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
13654 * config/arm/arm.c (arm_function_value): Likewise.
13655 (aapcs_allocate_return_reg): Likewise.
13656 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
13657 * config/i386/i386.c (construct_container): Likewise.
13658 (ix86_gimplify_va_arg): Likewise.
13659 (ix86_expand_sse_cmp): Likewise.
13660 (emit_memmov): Likewise.
13661 (emit_memset): Likewise.
13662 (expand_small_movmem_or_setmem): Likewise.
13663 (ix86_expand_pextr): Likewise.
13664 (ix86_expand_pinsr): Likewise.
13665 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
13666 * config/microblaze/microblaze.c (microblaze_block_move_straight):
13668 * config/mips/mips.c (mips_function_value_1) Likewise.
13669 (mips_block_move_straight): Likewise.
13670 (mips_expand_ins_as_unaligned_store): Likewise.
13671 * config/powerpcspe/powerpcspe.c
13672 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13673 (rs6000_darwin64_record_arg_flush): Likewise.
13674 * config/rs6000/rs6000.c
13675 (rs6000_darwin64_record_arg_advance_flush): Likewise.
13676 (rs6000_darwin64_record_arg_flush): Likewise.
13677 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
13678 (sparc_function_value_1): Likewise.
13679 * config/spu/spu.c (adjust_operand): Likewise.
13680 (spu_emit_branch_or_set): Likewise.
13681 (arith_immediate_p): Likewise.
13682 * emit-rtl.c (gen_lowpart_common): Likewise.
13683 * expr.c (expand_expr_real_1): Likewise.
13684 * function.c (assign_parm_setup_block): Likewise.
13685 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
13686 * reload1.c (alter_reg): Likewise.
13687 * stor-layout.c (mode_for_vector): Likewise.
13688 (layout_type): Likewise.
13690 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13692 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
13693 (spu_convert_move): Likewise.
13694 * lower-subreg.c (resolve_simple_move): Likewise.
13696 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13699 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
13700 define_insn to a define_expand.
13701 (altivec_vsum2sws_direct): New define_insn.
13702 (altivec_vsumsws): Convert from a define_insn to a define_expand.
13704 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13706 * config/arm/arm.c (arm_option_params_internal): Improve setting of
13709 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
13714 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
13715 realign stack if stack alignment needed is less than incoming
13718 2017-09-05 Marek Polacek <polacek@redhat.com>
13721 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
13724 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
13726 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
13728 2017-09-05 Richard Biener <rguenther@suse.de>
13730 PR tree-optimization/82084
13731 * fold-const.c (can_native_encode_string_p): Handle wide characters.
13733 2017-09-05 Richard Biener <rguenther@suse.de>
13735 PR tree-optimization/82102
13736 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
13738 2017-09-05 Martin Liska <mliska@suse.cz>
13740 PR tree-optimization/82032
13741 * tree-cfg.c (generate_range_test): New function.
13742 * tree-cfg.h (generate_range_test): Declared here.
13743 * tree-cfgcleanup.c (convert_single_case_switch): New function.
13744 (cleanup_control_expr_graph): Use it.
13745 * tree-switch-conversion.c (try_switch_expansion): Remove
13747 (emit_case_nodes): Use generate_range_test.
13749 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13752 * config/i386/i386.md (*<btsc><mode>_mask): Add
13753 TARGET_USE_BT to insn constraint.
13754 (*btr<mode>_mask): Ditto.
13756 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
13758 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
13759 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
13761 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
13764 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
13765 TARGET_NEON and TARGET_IWMMXT.
13766 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
13767 TARGET_NEON and TARGET_IWMMXT.
13768 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
13770 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
13772 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
13773 (ix86_rewrite_tls_address): Ditto.
13774 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
13775 (ix86_rewrite_tls_address_1): Ditto.
13776 (ix86_rewrite_tls_address): Ditto.
13777 * config/i386/predicates.md (tls_address_pattern): New predicate.
13778 * config/i386/i386.md (TLS address splitter): New splitter.
13780 2017-09-04 Richard Biener <rguenther@suse.de>
13782 PR tree-optimization/82084
13783 * fold-const.h (can_native_encode_string_p): Declare.
13784 * fold-const.c (can_native_encode_string_p): Factor out from ...
13785 (native_encode_string): ... here.
13786 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
13787 vectorizing stores from constants we later cannot handle.
13789 2017-09-04 Marek Polacek <polacek@redhat.com>
13792 * doc/invoke.texi: Update -Wtautological-compare documentation.
13794 2017-09-04 Jeff Law <law@redhat.com>
13796 PR tree-optimization/64910
13797 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
13798 swap the first and last operand if the last is a constant.
13800 2017-09-04 Marek Polacek <polacek@redhat.com>
13803 * convert.c (do_narrow): When sanitizing signed integer overflows,
13804 bail out for signed types.
13805 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
13807 2017-09-04 Richard Biener <rguenther@suse.de>
13809 PR tree-optimization/82060
13810 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13811 Move devirtualization after stmt folding and before EH/AB/noreturn
13812 cleanup to get the stmt refs canonicalized. Use a bool instead
13813 of gimple_modified_p since that doesn't work for NOPs. Schedule
13814 NOPs generated by folding for removal.
13816 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13817 Alan Hayward <alan.hayward@arm.com>
13818 David Sherwood <david.sherwood@arm.com>
13820 * coretypes.h (pad_direction): New enum.
13821 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
13822 (FUNCTION_ARG_PADDING): Likewise.
13823 * target.def (function_arg_padding): New hook.
13824 * targhooks.h (default_function_arg_padding): Declare.
13825 * targhooks.c (default_function_arg_padding): New function.
13826 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
13827 (TARGET_FUNCTION_ARG_PADDING): ...this.
13828 * doc/tm.texi: Regenerate.
13829 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
13830 instead of direction.
13831 (compute_argument_addresses): Likewise.
13832 (load_register_parameters): Likewise.
13833 (emit_library_call_value_1): Likewise.
13834 (store_one_arg): Use targetm.calls.function_arg_padding instead
13835 of FUNCTION_ARG_PADDING.
13836 (must_pass_in_stack_var_size_or_pad): Likewise.
13837 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
13838 (emit_group_store): Likewise.
13839 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
13840 instead of FUNCTION_ARG_PADDING.
13841 (emit_push_insn): Likewise, and propagate enum change throughout
13843 * function.h (direction): Delete.
13844 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
13846 * function.c (assign_parm_find_stack_rtl): Likewise.
13847 (assign_parm_setup_block_p): Likewise.
13848 (assign_parm_setup_block): Likewise.
13849 (gimplify_parameters): Likewise.
13850 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
13851 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
13853 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
13854 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13855 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
13856 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
13857 (aarch64_function_arg_padding): ...this new function.
13858 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
13859 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13860 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
13861 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13862 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
13863 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13864 (arm_pad_arg_upward): Replace with...
13865 (arm_function_arg_padding): ...this new function.
13866 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
13868 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
13869 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
13870 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13871 (ia64_hpux_function_arg_padding): Replace with...
13872 (ia64_function_arg_padding): ...this new function. Use pad_direction
13873 instead of direction. Check for TARGET_HPUX.
13874 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
13875 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13876 (iq2000_function_arg_padding): New function.
13877 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
13878 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
13879 (mips_function_arg_padding): ...this new function.
13880 (mips_pad_reg_upward): Update accordingly.
13881 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13882 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
13883 targetm.calls.function_arg_padding.
13884 (FUNCTION_ARG_PADDING): Delete.
13885 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13886 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13887 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13888 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13889 (nios2_block_reg_padding): Return pad_direction instead of direction.
13890 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13891 instead of direction.
13892 (nios2_function_arg_padding): Likewise. Make static.
13893 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13894 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13895 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13896 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13897 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13898 (pa_function_arg_padding): Make static. Return pad_direction instead
13900 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13901 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13902 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13903 instead of direction. Use targetm.calls.function_arg_padding.
13904 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13905 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13906 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13907 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13908 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13910 (function_arg_padding): Rename to...
13911 (rs6000_function_arg_padding): ...this. Make static. Return
13912 pad_direction instead of direction.
13913 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13914 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13915 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13916 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13917 instead of direction. Use targetm.calls.function_arg_padding.
13918 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13919 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13920 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13921 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13922 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13923 (function_arg_padding): Rename to...
13924 (rs6000_function_arg_padding): ...this. Make static. Return
13925 pad_direction instead of direction.
13926 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13927 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13928 * config/s390/s390.c (s390_function_arg_padding): New function.
13929 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13930 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13931 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13932 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13933 (function_arg_padding): Rename to...
13934 (sparc_function_arg_padding): ...this. Make static. Return
13935 pad_direction instead of direction.
13936 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13937 * config/spu/spu.c (spu_function_arg_padding): New function.
13938 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13939 * system.h (FUNCTION_ARG_PADDING): Poison.
13941 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13942 Alan Hayward <alan.hayward@arm.com>
13943 David Sherwood <david.sherwood@arm.com>
13945 * target.def (modes_tieable_p): New hook.
13946 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13947 (TARGET_MODES_TIEABLE_P): ...this.
13948 * doc/tm.texi.in: Regenerate.
13949 * hooks.h (hook_bool_mode_mode_true): Declare.
13950 * hooks.c (hook_bool_mode_mode_true): New function.
13951 * combine.c (subst): Use targetm.modes_tieable_p instead of
13953 * dse.c (find_shift_sequence): Likewise.
13954 * expmed.c (extract_low_bits): Likewise.
13955 * lower-subreg.c: Include target.h.
13956 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13958 * rtlanal.c (rtx_cost): Likewise.
13959 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13960 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13961 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13962 (TARGET_MODES_TIEABLE_P): Redefine.
13963 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13964 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13965 (TARGET_MODES_TIEABLE_P): Redefine.
13966 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13967 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13968 (arc_modes_tieable_p): New function.
13969 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13970 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13971 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13972 (arm_modes_tieable_p): Make static.
13973 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13974 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13975 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13976 (TARGET_MODES_TIEABLE_P): Redefine.
13977 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13978 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13979 (TARGET_MODES_TIEABLE_P): Redefine.
13980 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13981 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13982 (cr16_modes_tieable_p): New function.
13983 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13984 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13985 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13986 (TRULY_NOOP_TRUNCATION): Update comment.
13987 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13988 (TRULY_NOOP_TRUNCATION): Update comment.
13989 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13990 (frv_modes_tieable_p): New function.
13991 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13992 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13993 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13994 (TARGET_MODES_TIEABLE_P): Redefine.
13995 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13996 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13997 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13998 (TARGET_MODES_TIEABLE_P): Redefine.
13999 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14000 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14001 (ia64_modes_tieable_p): New function.
14002 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14003 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14004 (iq2000_modes_tieable_p): New function.
14005 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14006 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14007 (lm32_modes_tieable_p): New function.
14008 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14009 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14010 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14011 (TARGET_MODES_TIEABLE_P): Redefine.
14012 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14013 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14014 (m32r_modes_tieable_p): New function.
14015 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14016 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14017 (m68k_modes_tieable_p): New function.
14018 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14019 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14020 (mcore_modes_tieable_p): New function.
14021 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14022 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14024 (TARGET_MODES_TIEABLE_P): Redefine.
14025 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14026 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14027 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14028 (TARGET_MODES_TIEABLE_P): Redefine.
14029 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14030 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14031 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14032 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14033 (mn10300_modes_tieable_p): ...this and make static.
14034 (TARGET_MODES_TIEABLE_P): Redefine.
14035 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14036 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14037 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14038 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14039 (msp430_modes_tieable_p): Make static.
14040 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14041 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14042 (TARGET_MODES_TIEABLE_P): Redefine.
14043 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14044 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14045 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14046 (TARGET_MODES_TIEABLE_P): Redefine.
14047 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14048 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14049 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14050 (TARGET_MODES_TIEABLE_P): Redefine.
14051 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14052 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14053 (pdp11_modes_tieable_p): New function.
14054 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14055 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14056 (rs6000_modes_tieable_p): New function.
14057 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14058 * config/powerpcspe/powerpcspe.md: Update comment.
14059 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14060 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14061 (TARGET_MODES_TIEABLE_P): Redefine.
14062 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14063 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14064 (rl78_modes_tieable_p): New function.
14065 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14066 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14067 (rs6000_modes_tieable_p): New function.
14068 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14069 * config/rs6000/rs6000.md: Update comment.
14070 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14071 * config/rx/rx.c (rx_modes_tieable_p): New function.
14072 (TARGET_MODES_TIEABLE_P): Redefine.
14073 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14074 * config/s390/s390.c (s390_modes_tieable_p): New function.
14075 (TARGET_MODES_TIEABLE_P): Redefine.
14076 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14077 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14078 (sh_modes_tieable_p): New function.
14079 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14080 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14081 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14082 (sparc_modes_tieable_p): Make static.
14083 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14084 * config/spu/spu.c (spu_modes_tieable_p): New function.
14085 (TARGET_MODES_TIEABLE_P): Redefine.
14086 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14087 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14088 (TARGET_MODES_TIEABLE_P): Redefine.
14089 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14090 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14091 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14092 * config/v850/v850.c (v850_modes_tieable_p): New function.
14093 (TARGET_MODES_TIEABLE_P): Redefine.
14094 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14095 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14096 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14097 (visium_modes_tieable_p): New function.
14098 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14099 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14100 (xtensa_modes_tieable_p): New function.
14101 * system.h (MODES_TIEABLE_P): Poison.
14103 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14104 Alan Hayward <alan.hayward@arm.com>
14105 David Sherwood <david.sherwood@arm.com>
14107 * target.def (hard_regno_mode_ok): New hook.
14108 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14109 (TARGET_HARD_REGNO_MODE_OK): ...this.
14110 * doc/tm.texi.in: Regenerate.
14111 * hooks.h (hook_bool_uint_mode_true): Declare.
14112 * hooks.c (hook_bool_uint_mode_true): New function.
14113 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14114 HARD_REGNO_MODE_OK.
14115 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14116 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14117 instead of HARD_REGNO_MODE_OK.
14118 * caller-save.c: Include target.h.
14119 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14120 HARD_REGNO_MODE_OK.
14121 * combine.c (can_combine_p): Likewise.
14122 (combinable_i3pat): Likewise.
14123 (can_change_dest_mode): Likewise.
14124 * expr.c (init_expr_target): Likewise.
14125 (convert_move): Likewise.
14126 (convert_modes): Likewise.
14127 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14128 (setup_prohibited_mode_move_regs): Likewise.
14129 * ira.h (target_ira): Likewise.
14130 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14131 * lra-constraints.c (process_alt_operands): Likewise.
14132 (split_reg): Likewise.
14133 * recog.c (peep2_find_free_register): Likewise.
14134 * ree.c (combine_reaching_defs): Likewise.
14135 * regcprop.c (maybe_mode_change): Likewise.
14136 * reginfo.c (init_reg_sets_1): Likewise.
14137 (choose_hard_reg_mode): Likewise.
14138 (simplifiable_subregs): Likewise.
14139 * regrename.c (check_new_reg_p): Likewise.
14140 * reload.c (find_valid_class): Likewise.
14141 (find_valid_class_1): Likewise.
14142 (reload_inner_reg_of_subreg): Likewise.
14143 (push_reload): Likewise.
14144 (combine_reloads): Likewise.
14145 (find_dummy_reload): Likewise.
14146 (find_reloads): Likewise.
14147 * reload1.c (find_reg): Likewise.
14148 (set_reload_reg): Likewise.
14149 (allocate_reload_reg): Likewise.
14150 (choose_reload_regs): Likewise.
14151 (reload_adjust_reg_for_temp): Likewise.
14152 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14153 (simplify_subreg_regno): Likewise.
14154 * sel-sched.c (init_regs_for_mode): Likewise.
14155 * varasm.c (make_decl_rtl): Likewise.
14156 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14157 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14158 HARD_REGNO_MODE_OK.
14159 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14160 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14161 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14162 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14163 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14164 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14165 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14166 (arc_mode_class): Delete.
14167 (HARD_REGNO_MODE_OK): Delete.
14168 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14169 (arc_hard_regno_mode_ok): Rename old array to...
14170 (arc_hard_regno_mode_ok_modes): ...this.
14171 (arc_conditional_register_usage): Update accordingly.
14172 (arc_mode_class): Make static.
14173 (arc_hard_regno_mode_ok): New function.
14174 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14175 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14176 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14177 (arm_hard_regno_mode_ok): Make static.
14178 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14179 HARD_REGNO_MODE_OK.
14180 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14181 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14182 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14184 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14185 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14186 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14187 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14188 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14190 * config/bfin/predicates.md (valid_reg_operand): Use
14191 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14192 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14193 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14194 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14195 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14196 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14197 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14198 (cr16_hard_regno_mode_ok): Make static and return a bool.
14199 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14200 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14201 (cris_hard_regno_mode_ok): New function.
14202 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14203 (epiphany_mode_class): Delete.
14204 (HARD_REGNO_MODE_OK): Delete.
14205 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14206 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14207 (hard_regno_mode_ok): Rename to...
14208 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14209 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14210 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14211 HARD_REGNO_MODE_OK.
14212 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14213 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14214 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14215 (frv_hard_regno_mode_ok): Make static and return a bool.
14216 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14217 HARD_REGNO_MODE_OK.
14218 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14219 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14220 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14221 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14223 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14224 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14225 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14226 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14228 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14229 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14230 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14231 (ia64_hard_regno_mode_ok): New function.
14232 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14233 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14234 (iq2000_hard_regno_mode_ok): New function.
14235 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14236 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14237 (lm32_hard_regno_mode_ok): New function.
14238 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14239 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14240 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14241 instead of HARD_REGNO_MODE_OK.
14242 (m32c_hard_regno_ok): Rename to...
14243 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14244 (m32c_cannot_change_mode_class): Update accordingly.
14245 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14246 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14247 (m32r_mode_class): Delete.
14248 (HARD_REGNO_MODE_OK): Delete.
14249 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14250 (m32r_hard_regno_mode_ok): Rename to...
14251 (m32r_hard_regno_modes): ...this.
14252 (m32r_mode_class): Make static.
14253 (m32r_hard_regno_mode_ok): New function.
14254 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
14255 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
14256 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14257 (m68k_hard_regno_mode_ok): Make static.
14258 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
14259 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14260 (mcore_hard_regno_mode_ok): New function.
14261 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
14262 (HARD_REGNO_MODE_OK): Delete.
14263 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
14265 (microblaze_hard_regno_mode_ok_p): ...this and make static.
14266 (microblaze_hard_regno_mode_ok): New function.
14267 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14268 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
14269 (mips_hard_regno_mode_ok): Delete.
14270 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
14271 (mips_hard_regno_mode_ok_p): ...this and make static.
14272 (mips_hard_regno_mode_ok_p): Rename to...
14273 (mips_hard_regno_mode_ok_uncached): ...this.
14274 (mips_hard_regno_mode_ok): New function.
14275 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
14276 of HARD_REGNO_MODE_OK.
14277 (mips_option_override): Update after above name changes.
14278 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14279 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
14280 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
14281 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
14282 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
14283 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14284 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
14285 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
14286 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
14287 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14288 (msp430_hard_regno_mode_ok): Make static and return a bool.
14289 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
14290 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
14291 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
14293 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14294 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
14295 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
14296 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
14297 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
14298 (PA_HARD_REGNO_MODE_OK): ...this
14299 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
14300 (PA_HARD_REGNO_MODE_OK): ...this.
14301 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14302 (pa_hard_regno_mode_ok): New function.
14303 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
14304 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14305 (pdp11_hard_regno_mode_ok): New function.
14306 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
14307 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
14309 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
14311 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14312 (rs6000_hard_regno_mode_ok): Rename to...
14313 (rs6000_hard_regno_mode_ok_uncached): ...this.
14314 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14315 (rs6000_hard_regno_mode_ok): New function.
14316 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
14317 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
14318 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
14319 (riscv_hard_regno_mode_ok): ...this and make static.
14320 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14321 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
14322 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
14323 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14324 (rl78_hard_regno_mode_ok): Make static and return bool.
14325 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
14326 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
14328 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
14329 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14330 (rs6000_hard_regno_mode_ok): Rename to...
14331 (rs6000_hard_regno_mode_ok_uncached): ...this.
14332 (rs6000_init_hard_regno_mode_ok): Update accordingly.
14333 (rs6000_hard_regno_mode_ok): New function.
14334 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
14335 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
14336 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14337 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
14338 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
14339 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
14340 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14341 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
14342 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
14343 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14344 (sh_hard_regno_mode_ok): Make static.
14345 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
14346 instead of HARD_REGNO_MODE_OK.
14347 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
14348 (sparc_mode_class): Delete.
14349 (HARD_REGNO_MODE_OK): Delete.
14350 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14351 (hard_regno_mode_classes): Make static.
14352 (sparc_mode_class): Likewise.
14353 (sparc_hard_regno_mode_ok): New function.
14354 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
14355 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
14356 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
14358 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14359 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
14360 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
14361 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
14362 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
14363 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14364 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
14365 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
14366 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14367 (visium_hard_regno_mode_ok): New function.
14368 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
14369 instead of HARD_REGNO_MODE_OK.
14370 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
14371 (HARD_REGNO_MODE_OK): Delete.
14372 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
14373 (xtensa_hard_regno_mode_ok_p): ...this and make static.
14374 (xtensa_option_override): Update accordingly.
14375 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14376 (xtensa_hard_regno_mode_ok): New function.
14377 * system.h (HARD_REGNO_MODE_OK): Poison.
14379 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14380 Alan Hayward <alan.hayward@arm.com>
14381 David Sherwood <david.sherwood@arm.com>
14383 * target.def (hard_regno_call_part_clobbered): New hook.
14384 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
14385 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
14386 * doc/tm.texi: Regenerate.
14387 * hooks.h (hook_bool_uint_mode_false): Declare.
14388 * hooks.c (hook_bool_uint_mode_false): New function.
14389 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14390 * cselib.c (cselib_process_insn): Use
14391 targetm.hard_regno_call_part_clobbered instead of
14392 HARD_REGNO_CALL_PART_CLOBBERED.
14393 * ira-conflicts.c (ira_build_conflicts): Likewise.
14394 * ira-costs.c (ira_tune_allocno_costs): Likewise.
14395 * lra-constraints.c (need_for_call_save_p): Likewise.
14396 * lra-lives.c: Include target.h.
14397 (check_pseudos_live_through_calls): Use
14398 targetm.hard_regno_call_part_clobbered instead of
14399 HARD_REGNO_CALL_PART_CLOBBERED.
14400 * regcprop.c: Include target.h.
14401 (copyprop_hardreg_forward_1): Use
14402 targetm.hard_regno_call_part_clobbered instead of
14403 HARD_REGNO_CALL_PART_CLOBBERED.
14404 * reginfo.c (choose_hard_reg_mode): Likewise.
14405 * regrename.c (check_new_reg_p): Likewise.
14406 * reload.c (find_equiv_reg): Likewise.
14407 * reload1.c (emit_reload_insns): Likewise.
14408 * sched-deps.c (deps_analyze_insn): Likewise.
14409 * sel-sched.c (init_regs_for_mode): Likewise.
14410 (mark_unavailable_hard_regs): Likewise.
14411 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
14412 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14413 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
14415 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14416 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14417 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
14419 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
14421 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14422 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14423 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
14425 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14426 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14427 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
14429 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14430 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
14432 * config/powerpcspe/powerpcspe.c
14433 (rs6000_hard_regno_call_part_clobbered): New function.
14434 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14435 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14436 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
14438 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14439 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14440 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
14442 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
14443 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
14444 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
14446 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14447 Alan Hayward <alan.hayward@arm.com>
14448 David Sherwood <david.sherwood@arm.com>
14450 * rtl.h (subreg_memory_offset): Declare.
14451 * emit-rtl.c (subreg_memory_offset): New function.
14452 * expmed.c (store_bit_field_1): Use it.
14453 * expr.c (undefined_operand_subword_p): Likewise.
14454 * simplify-rtx.c (simplify_subreg): Likewise.
14456 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
14458 PR rtl-optimization/57448
14461 * optabs.c (expand_atomic_load): Place compiler memory barriers if
14462 using atomic_load pattern.
14463 (expand_atomic_store): Likewise.
14465 2017-09-04 Jakub Jelinek <jakub@redhat.com>
14468 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
14469 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
14470 handling. Use replace_call_with_value with NULL instead of
14471 gsi_replace, unlink_stmt_vdef and release_defs.
14473 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
14476 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
14478 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14481 * rtl.h (emit_library_call_value_1): Declare.
14482 (emit_library_call): Replace declaration with a series of overloads.
14483 Remove the parameter count argument.
14484 (emit_library_call_value): Likewise.
14485 * calls.c (emit_library_call_value_1): Make global. Replace varargs
14486 with an "rtx_mode_t *".
14487 (emit_library_call_value): Delete.
14488 (emit_library_call): Likewise.
14489 * asan.c (asan_emit_stack_protection): Update calls accordingly.
14490 (asan_emit_allocas_unpoison): Likewise.
14491 * builtins.c (expand_builtin_powi): Likewise.
14492 (expand_asan_emit_allocas_unpoison): Likewise.
14493 * cfgexpand.c (expand_main_function): Likewise.
14494 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
14495 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
14496 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
14497 * config/arm/arm.c (arm_trampoline_init): Likewise.
14498 (arm_call_tls_get_addr): Likewise.
14499 (arm_expand_divmod_libfunc): Likewise.
14500 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
14501 (smulsi3_highpart): Likewise.
14502 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
14503 (c6x_expand_compare): Likewise.
14504 (c6x_expand_movmem): Likewise.
14505 * config/frv/frv.c (frv_trampoline_init): Likewise.
14506 * config/i386/i386.c (ix86_trampoline_init): Likewise.
14507 (ix86_expand_divmod_libfunc): Likewise.
14508 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
14509 (ia64_expand_compare): Likewise.
14510 (ia64_profile_hook): Likewise.
14511 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
14512 (nonlocal_goto): Likewise.
14513 (restore_stack_nonlocal): Likewise.
14514 * config/m32r/m32r.c (block_move_call): Likewise.
14515 (m32r_trampoline_init): Likewise.
14516 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
14517 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
14518 (m68k_call_m68k_read_tp): Likewise.
14519 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
14520 (microblaze_expand_divide): Likewise.
14521 * config/mips/mips.h (mips_args): Likewise.
14522 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
14523 (MIPS_ICACHE_SYNC): Likewise.
14524 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
14525 (nios2_trampoline_init): Likewise.
14526 * config/pa/pa.c (hppa_tls_call): Likewise.
14527 (pa_trampoline_init): Likewise.
14528 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
14529 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
14530 (expand_strn_compare): Likewise.
14531 (rs6000_generate_compare): Likewise.
14532 (rs6000_expand_float128_convert): Likewise.
14533 (output_profile_hook): Likewise.
14534 (rs6000_trampoline_init): Likewise.
14535 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
14536 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
14537 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
14538 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
14539 (rs6000_generate_compare): Likewise.
14540 (rs6000_expand_float128_convert): Likewise.
14541 (output_profile_hook): Likewise.
14542 (rs6000_trampoline_init): Likewise.
14543 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
14544 * config/sh/sh.c (sh_trampoline_init): Likewise.
14545 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
14546 (sparc_emit_float_lib_cmp): Likewise.
14547 (sparc32_initialize_trampoline): Likewise.
14548 (sparc64_initialize_trampoline): Likewise.
14549 (sparc_profile_hook): Likewise.
14550 * config/spu/spu.c (ea_load_store): Likewise.
14551 * config/spu/spu.md (floatunssidf2): Likewise.
14552 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
14553 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
14554 * config/visium/visium.c (expand_block_move_4): Likewise.
14555 (expand_block_move_2): Likewise.
14556 (expand_block_move_1): Likewise.
14557 (expand_block_set_4): Likewise.
14558 (expand_block_set_2): Likewise.
14559 (expand_block_set_1): Likewise.
14560 (visium_trampoline_init): Likewise.
14561 (visium_profile_hook): Likewise.
14562 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
14563 (xtensa_setup_frame_addresses): Likewise.
14564 (xtensa_trampoline_init): Likewise.
14565 * except.c (sjlj_emit_function_enter): Likewise.
14566 (sjlj_emit_function_exit): Likewise.
14567 * explow.c (allocate_dynamic_stack_space): Likewise.
14568 (probe_stack_range): Likewise.
14569 * expr.c (convert_mode_scalar): Likewise.
14570 * optabs.c (expand_binop): Likewise.
14571 (expand_twoval_binop_libfunc): Likewise.
14572 (expand_unop): Likewise.
14573 (prepare_cmp_insn): Likewise.
14574 (prepare_float_lib_cmp): Likewise.
14575 (expand_float): Likewise.
14576 (expand_fix): Likewise.
14577 (expand_fixed_convert): Likewise.
14578 (maybe_emit_sync_lock_test_and_set): Likewise.
14579 (expand_atomic_compare_and_swap): Likewise.
14580 (expand_mem_thread_fence): Likewise.
14581 (expand_atomic_fetch_op): Likewise.
14583 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
14585 * doc/generic.texi (OpenACC): Adjust URL.
14586 * doc/invoke.texi (C Dialect Options): Ditto.
14588 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
14590 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
14591 predicate for operand 1. Add (m,<S>) constraint.
14592 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
14593 Prevent memory operand 1 with register operand 2.
14595 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
14597 PR rtl-optimization/82024
14598 * combine.c (try_combine): If the combination result is a PARALLEL,
14599 and we only need to retain the SET in there that would be placed
14600 at I2, check that we can place that at I3 instead, before doing so.
14602 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14605 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
14607 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
14608 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
14611 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
14612 Jeff Law <law@redhat.com>
14614 * varasm.c (bss_initializer_p): Do not put constants into .bss
14615 (categorize_decl_for_section): Handle bss_initializer_p returning
14616 false when DECL_INITIAL is NULL.
14618 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14621 * config/s390/s390.c (s390_can_inline_p): New function.
14623 2017-09-01 Jeff Law <law@redhat.com>
14625 PR tree-optimization/82052
14626 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14627 Always initialize the returned slot after a hash table miss
14628 when INSERT is true.
14630 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
14632 * config/s390/s390.md (mem_signal_fence): Remove.
14633 * doc/md.texi (mem_signal_fence): Remove.
14634 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
14636 * target-insns.def (mem_signal_fence): Remove.
14638 2017-09-01 Jakub Jelinek <jakub@redhat.com>
14641 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
14644 * asan.c (create_odr_indicator): Strip name encoding from assembler
14645 name before appending it after __odr_asan_.
14647 2017-09-01 Martin Liska <mliska@suse.cz>
14649 PR tree-optimization/82059
14650 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
14651 frequency only when an edge is redirected.
14653 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14655 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
14656 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
14657 (arc_conditional_register_usage): Remove ARC600 lp_count
14659 (arc_file_start): Emit Tag_ARC_CPU_variation.
14660 (arc_can_use_doloop_p): New conditions to use ZOLs.
14661 (hwloop_fail): New function.
14662 (hwloop_optimize): Likewise.
14663 (hwloop_pattern_reg): Likewise.
14664 (arc_doloop_hooks): New struct, to be used with reorg_loops.
14665 (arc_reorg_loops): New function, calls reorg_loops.
14666 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
14667 (arc600_corereg_hazard): Remove ZOL checking, case handled by
14669 (arc_loop_hazard): Remove function, functionality moved into
14671 (arc_hazard): Remove arc_loop_hazard call.
14672 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
14673 into hwloop_optimize.
14674 (arc_label_align): Remove ZOL handling.
14675 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
14676 * config/arc/arc.md (doloop_begin): Remove pattern.
14677 (doloop_begin_i): Likewise.
14678 (doloop_end_i): Likewise.
14679 (doloop_fallback): Likewise.
14680 (doloop_fallback_m): Likewise.
14681 (doloop_end): Reimplement expand.
14682 (arc_lp): New pattern for LP instruction.
14683 (loop_end): New pattern.
14684 (loop_fail): Likewise.
14685 (decrement_and_branch_until_zero): Likewise.
14686 * config/arc/arc.opt (mlpc-width): New option.
14687 * doc/invoke.texi (mlpc-width): Document option.
14689 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14691 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
14692 (arc_ccfsm_advance): Fix checking for delay slots.
14693 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
14695 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14697 * config/arc/arc.md (movqi_insn): Add stores to save constant long
14699 (movhi_insn): Update store instruction constraint which are saving
14700 6-bit short immediates.
14701 (movsi_insn): Consider also short scaled load operations.
14702 (zero_extendhisi2_i): Use Usd constraint instead of T.
14703 (extendhisi2_i): Add q constraint.
14704 (arc_clzsi2): Add type and length attributes.
14705 (arc_ctzsi2): Likewise.
14706 * config/arc/constraints.md (Usc): Update constraint, the
14707 assembler can parse two relocations for a single instruction.
14709 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
14711 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
14712 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
14714 2017-08-31 Olivier Hainque <hainque@adacore.com>
14716 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
14717 match as powerpc-wrs-vxworks*.
14719 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
14721 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
14722 register constraint for by-element operand.
14723 (aarch64_mls_elt_merge<mode>): Likewise.
14725 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14727 * config/arc/arc.c (arc_can_follow_jump): Check for short
14730 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14732 * config.gcc: Use g.opt for arc.
14733 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
14734 functionality moved to ...
14735 (legitimate_scaled_address_p): New function, ...here.
14736 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
14737 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
14738 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
14740 (arc_override_options): Handle G option.
14741 (arc_output_pic_addr_const): Correct function definition.
14742 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
14743 (arc_decl_anon_ns_mem_p): Delete.
14744 (arc_in_small_data_p): Overhaul this function to take into
14745 consideration the value given via G option.
14746 (arc_rewrite_small_data_1): Renamed and corrected old
14747 arc_rewrite_small_data function.
14748 (arc_rewrite_small_data): New function.
14749 (small_data_pattern): Don't use pic_offset_table_rtx.
14750 * config/arc/arc.h (CC1_SPEC): Recognize G option.
14751 * config/arc/simdext.md (movmisalignv2hi): Use
14752 prepare_move_operands function.
14754 (movmisalign*): Likewise.
14755 * doc/invoke.texi (ARC options): Document -G option.
14757 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
14759 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
14761 * config/arc/arc.c (arc_print_operand): Output scalled address for
14762 sdata whenever is possible.
14763 (arc_in_small_data_p): Allow sdata for 64bit datum when double
14764 load/stores are available.
14765 (compact_sda_memory_operand): Check for the alignment required by
14766 code density instructions.
14767 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
14769 * config/arc/constraints.md (Usd): Update constraint.
14770 (Us0): New constraint.
14771 (Usc): Update constraint.
14773 2017-08-31 Richard Biener <rguenther@suse.de>
14775 PR middle-end/82054
14776 * dwarf2out.c (dwarf2out_early_global_decl): Process each
14777 function only once.
14779 2017-08-31 Tamar Christina <tamar.christina@arm.com>
14781 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
14782 Resize type_signature.
14784 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14785 Alan Hayward <alan.hayward@arm.com>
14786 David Sherwood <david.sherwood@arm.com>
14788 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
14789 subregs whose inner modes can be stored in GPRs.
14790 (aarch64_classify_index): Likewise.
14792 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14793 Alan Hayward <alan.hayward@arm.com>
14794 David Sherwood <david.sherwood@arm.com>
14796 * config/aarch64/iterators.md (V_cmp_result): Rename to...
14797 (V_INT_EQUIV): ...this.
14798 (v_cmp_result): Rename to...
14799 (v_int_equiv): ...this.
14800 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
14801 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
14802 (copysign<mode>3): Likewise.
14803 (aarch64_simd_bsl<mode>_internal): Likewise.
14804 (aarch64_simd_bsl<mode>): Likewise.
14805 (vec_cmp<mode><mode>): Likewise.
14806 (vcond<mode><mode>): Likewise.
14807 (vcond<v_cmp_mixed><mode>): Likewise.
14808 (vcondu<mode><v_cmp_mixed>): Likewise.
14809 (aarch64_cm<optab><mode>): Likewise.
14810 (aarch64_cmtst<mode>): Likewise.
14811 (aarch64_fac<optab><mode>): Likewise.
14812 (vec_perm_const<mode>): Likewise.
14813 (vcond_mask_<mode><v_cmp_result>): Rename to...
14814 (vcond_mask_<mode><v_int_equiv>): ...this.
14815 (vec_cmp<mode><v_cmp_result>): Rename to...
14816 (vec_cmp<mode><v_int_equiv>): ...this.
14818 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
14819 Alan Hayward <alan.hayward@arm.com>
14820 David Sherwood <david.sherwood@arm.com>
14822 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
14824 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
14825 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
14826 (UNSPEC_LD4_DREG): New unspecs.
14827 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
14828 (aarch64_ld2<mode>_dreg_be): Replace with...
14829 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
14831 (aarch64_ld3<mode>_dreg_le)
14832 (aarch64_ld3<mode>_dreg_be): Replace with...
14833 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
14835 (aarch64_ld4<mode>_dreg_le)
14836 (aarch64_ld4<mode>_dreg_be): Replace with...
14837 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
14840 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14842 PR tree-optimization/81987
14843 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
14844 insert an initializer in a location not dominated by the stride
14847 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14849 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
14850 on the entire header of the finally block in the fallthru case.
14852 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
14854 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
14856 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
14858 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
14859 rs6000_emit_move_from_cr and call renamed function.
14860 (rs6000_emit_prologue): Call renamed functions.
14861 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
14862 movesi_from_cr, remove volatile CRs.
14864 2017-08-30 Jon Beniston <jon@beniston.com>
14865 Richard Biener <rguenther@suse.de>
14867 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
14868 of VECTOR_MODE_P check.
14869 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
14870 element vector types.
14872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14874 * df.h (df_read_modify_subreg_p): Remove in favor of...
14875 * rtl.h (read_modify_subreg_p): ...this new function. Take a
14876 const_rtx instead of an rtx.
14877 * cprop.c (local_cprop_find_used_regs): Update accordingly.
14878 * df-problems.c (df_word_lr_mark_ref): Likewise.
14879 * ira-lives.c (mark_pseudo_reg_live): Likewise.
14880 (mark_pseudo_reg_dead): Likewise.
14881 (mark_ref_dead): Likewise.
14882 * reginfo.c (init_subregs_of_mode): Likewise.
14883 * sched-deps.c (sched_analyze_1): Likewise.
14884 * df-scan.c (df_def_record_1): Likewise.
14885 (df_uses_record): Likewise.
14886 (df_read_modify_subreg_p): Remove in favor of...
14887 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14888 const_rtx instead of an rtx.
14890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14891 Alan Hayward <alan.hayward@arm.com>
14892 David Sherwood <david.sherwood@arm.com>
14894 * rtl.h (partial_subreg_p): New function.
14895 * caller-save.c (save_call_clobbered_regs): Use it.
14896 * calls.c (expand_call): Likewise.
14897 * combine.c (combinable_i3pat): Likewise.
14898 (simplify_set): Likewise.
14899 (make_extraction): Likewise.
14900 (make_compound_operation_int): Likewise.
14901 (gen_lowpart_or_truncate): Likewise.
14902 (force_to_mode): Likewise.
14903 (make_field_assignment): Likewise.
14904 (reg_truncated_to_mode): Likewise.
14905 (record_truncated_value): Likewise.
14906 (move_deaths): Likewise.
14907 * cse.c (record_jump_cond): Likewise.
14908 (cse_insn): Likewise.
14909 * cselib.c (cselib_lookup_1): Likewise.
14910 * expmed.c (extract_bit_field_using_extv): Likewise.
14911 * function.c (assign_parm_setup_reg): Likewise.
14912 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14913 * ira-build.c (create_insn_allocnos): Likewise.
14914 * lra-coalesce.c (merge_pseudos): Likewise.
14915 * lra-constraints.c (match_reload): Likewise.
14916 (simplify_operand_subreg): Likewise.
14917 (curr_insn_transform): Likewise.
14918 * lra-lives.c (process_bb_lives): Likewise.
14919 * lra.c (new_insn_reg): Likewise.
14920 (lra_substitute_pseudo): Likewise.
14921 * regcprop.c (mode_change_ok): Likewise.
14922 (maybe_mode_change): Likewise.
14923 (copyprop_hardreg_forward_1): Likewise.
14924 * reload.c (push_reload): Likewise.
14925 (find_reloads): Likewise.
14926 (find_reloads_subreg_address): Likewise.
14927 * reload1.c (alter_reg): Likewise.
14928 (eliminate_regs_1): Likewise.
14929 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14931 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14933 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14934 back to if statements, including unpack.
14936 2017-08-30 Martin Liska <mliska@suse.cz>
14938 PR inline-asm/82001
14939 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14941 (func_checker::compare_asm_inputs_outputs): ... this function.
14942 (func_checker::compare_gimple_asm): Use the function to compare
14943 also ASM constrains.
14944 * ipa-icf-gimple.h: Rename the function.
14946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14947 Alan Hayward <alan.hayward@arm.com>
14948 David Sherwood <david.sherwood@arm.com>
14950 * coretypes.h (complex_mode): New type.
14951 * gdbhooks.py (build_pretty_printer): Handle it.
14952 * machmode.h (complex_mode): New class.
14953 (complex_mode::includes_p): New function.
14954 (is_complex_int_mode): Likewise.
14955 (is_complex_float_mode): Likewise.
14956 * genmodes.c (get_mode_class): Handle complex mode classes.
14957 * function.c (expand_function_end): Use is_complex_int_mode.
14959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14960 Alan Hayward <alan.hayward@arm.com>
14961 David Sherwood <david.sherwood@arm.com>
14963 * coretypes.h (scalar_mode_pod): New typedef.
14964 * gdbhooks.py (build_pretty_printer): Handle it.
14965 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14966 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14967 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14968 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14969 as_a <scalar_mode>.
14971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14972 Alan Hayward <alan.hayward@arm.com>
14973 David Sherwood <david.sherwood@arm.com>
14975 * machmode.h (mode_for_vector): Take a scalar_mode instead
14977 * stor-layout.c (mode_for_vector): Likewise.
14978 * explow.c (promote_mode): Use as_a <scalar_mode>.
14979 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14982 Alan Hayward <alan.hayward@arm.com>
14983 David Sherwood <david.sherwood@arm.com>
14985 * target.def (preferred_simd_mode): Take a scalar_mode
14986 instead of a machine_mode.
14987 * targhooks.h (default_preferred_simd_mode): Likewise.
14988 * targhooks.c (default_preferred_simd_mode): Likewise.
14989 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14990 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14991 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14992 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14993 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14994 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14995 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14996 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14998 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14999 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15000 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15001 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15002 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15003 * doc/tm.texi: Regenerate.
15004 * optabs-query.c (can_vec_mask_load_store_p): Return false for
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 * target.def (scalar_mode_supported_p): Take a scalar_mode
15012 instead of a machine_mode.
15013 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15014 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15015 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15016 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15017 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15018 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15019 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15020 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15021 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15022 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15023 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15024 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15025 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15026 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15028 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15029 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15030 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15031 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15032 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15034 * doc/tm.texi: Regenerate.
15036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15037 Alan Hayward <alan.hayward@arm.com>
15038 David Sherwood <david.sherwood@arm.com>
15040 * coretypes.h (opt_scalar_mode): New typedef.
15041 * gdbhooks.py (build_pretty_printers): Handle it.
15042 * machmode.h (mode_iterator::get_2xwider): Add overload for
15044 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15046 * expr.c (convert_mode_scalar): Likewise.
15047 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15048 * optabs.c (expand_float): Likewise.
15049 (expand_fix): Likewise.
15050 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15053 Alan Hayward <alan.hayward@arm.com>
15054 David Sherwood <david.sherwood@arm.com>
15056 * optabs.c (expand_float): Explicitly check for scalars before
15057 using a branching expansion.
15058 (expand_fix): Likewise.
15060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15061 Alan Hayward <alan.hayward@arm.com>
15062 David Sherwood <david.sherwood@arm.com>
15064 * expr.c (convert_mode): Split scalar handling out into...
15065 (convert_mode_scalar): ...this new function. Treat the modes
15068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15069 Alan Hayward <alan.hayward@arm.com>
15070 David Sherwood <david.sherwood@arm.com>
15072 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15074 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15077 Alan Hayward <alan.hayward@arm.com>
15078 David Sherwood <david.sherwood@arm.com>
15080 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15081 rather than a machine_mode.
15082 (fixed_from_string): Likewise.
15083 (fixed_convert): Likewise.
15084 (fixed_convert_from_int): Likewise.
15085 (fixed_convert_from_real): Likewise.
15086 (real_convert_from_fixed): Likewise.
15087 * fixed-value.c (fixed_from_double_int): Likewise.
15088 (fixed_from_string): Likewise.
15089 (fixed_convert): Likewise.
15090 (fixed_convert_from_int): Likewise.
15091 (fixed_convert_from_real): Likewise.
15092 (real_convert_from_fixed): Likewise.
15093 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15096 Alan Hayward <alan.hayward@arm.com>
15097 David Sherwood <david.sherwood@arm.com>
15099 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15100 of separate mode class checks. Do not allow vector modes here.
15101 (immed_wide_int_const): Use as_a <scalar_mode>.
15102 * explow.c (trunc_int_for_mode): Likewise.
15103 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15104 (wi::shwi): Likewise.
15105 (wi::min_value): Likewise.
15106 (wi::max_value): Likewise.
15107 * dwarf2out.c (loc_descriptor): Likewise.
15108 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15109 for CONST_WIDE_INT.
15111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15112 Alan Hayward <alan.hayward@arm.com>
15113 David Sherwood <david.sherwood@arm.com>
15115 * tree.h (SCALAR_TYPE_MODE): New macro.
15116 * expr.c (expand_expr_addr_expr_1): Use it.
15117 (expand_expr_real_2): Likewise.
15118 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15119 (fold_convert_const_fixed_from_int): Likewise.
15120 (fold_convert_const_fixed_from_real): Likewise.
15121 (native_encode_fixed): Likewise
15122 (native_encode_complex): Likewise
15123 (native_encode_vector): Likewise.
15124 (native_interpret_fixed): Likewise.
15125 (native_interpret_real): Likewise.
15126 (native_interpret_complex): Likewise.
15127 (native_interpret_vector): Likewise.
15128 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15129 (simd_clone_adjust_argument_types): Likewise.
15130 (simd_clone_init_simd_arrays): Likewise.
15131 (simd_clone_adjust): Likewise.
15132 * stor-layout.c (layout_type): Likewise.
15133 * tree.c (build_minus_one_cst): Likewise.
15134 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15135 * tree-inline.c (estimate_move_cost): Likewise.
15136 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15137 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15138 (vectorizable_reduction): Likewise.
15139 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15140 (vect_recog_mixed_size_cond_pattern): Likewise.
15141 (check_bool_pattern): Likewise.
15142 (adjust_bool_pattern): Likewise.
15143 (search_type_for_mask_1): Likewise.
15144 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15145 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15146 (vectorizable_load): Likewise.
15147 (vectorizable_store): Likewise.
15148 * ubsan.c (ubsan_encode_value): Likewise.
15149 * varasm.c (output_constant): Likewise.
15151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15152 Alan Hayward <alan.hayward@arm.com>
15153 David Sherwood <david.sherwood@arm.com>
15155 * coretypes.h (scalar_mode): New class.
15156 * machmode.h (scalar_mode): Likewise.
15157 (scalar_mode::includes_p): New function.
15158 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15159 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15160 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15161 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15162 * expmed.c (store_bit_field_1): Likewise.
15163 (extract_bit_field_1): Likewise.
15164 * expr.c (write_complex_part): Likewise.
15165 (read_complex_part): Likewise.
15166 (emit_move_complex_push): Likewise.
15167 (expand_expr_real_2): Likewise.
15168 * function.c (assign_parm_setup_reg): Likewise.
15169 (assign_parms_unsplit_complex): Likewise.
15170 * optabs.c (expand_binop): Likewise.
15171 * rtlanal.c (subreg_get_info): Likewise.
15172 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15173 * varasm.c (output_constant_pool_2): Likewise.
15175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15176 Alan Hayward <alan.hayward@arm.com>
15177 David Sherwood <david.sherwood@arm.com>
15179 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15181 (expmed_mult_highpart_optab): Likewise.
15182 (expmed_mult_highpart): Likewise.
15184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15185 Alan Hayward <alan.hayward@arm.com>
15186 David Sherwood <david.sherwood@arm.com>
15188 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15189 instead of a machine_mode.
15190 (builtin_memset_read_str): Likewise.
15191 * builtins.c (c_readstr): Likewise.
15192 (builtin_memcpy_read_str): Likewise.
15193 (builtin_strncpy_read_str): Likewise.
15194 (builtin_memset_read_str): Likewise.
15195 (builtin_memset_gen_str): Likewise.
15196 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15197 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15198 instead of a machine_mode.
15199 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15201 (make_extraction): Likewise.
15202 (try_widen_shift_mode): Take and return scalar_int_modes instead
15204 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15205 a scalar_int_mode instead of a machine_mode.
15206 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15207 (avr_addr_space_pointer_mode): Likewise.
15208 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15209 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15210 (msp430_unwind_word_mode): Likewise.
15211 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15212 (spu_addr_space_pointer_mode): Likewise.
15213 (spu_addr_space_address_mode): Likewise.
15214 (spu_libgcc_cmp_return_mode): Likewise.
15215 (spu_libgcc_shift_count_mode): Likewise.
15216 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15217 (rl78_addr_space_pointer_mode): Likewise.
15218 (fl78_unwind_word_mode): Likewise.
15219 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15221 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15222 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15223 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15224 (mips_valid_pointer_mode): Likewise.
15225 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15226 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15227 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15229 (ft32_addr_space_address_mode): Likewise.
15230 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15231 scalar_int_mode instead of a machine_mode.
15232 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15234 (m32c_addr_space_address_mode): Likewise.
15235 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15236 (rs6000_eh_return_filter_mode): Likewise.
15237 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15238 (rs6000_eh_return_filter_mode): Likewise.
15239 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15240 (s390_libgcc_shift_count_mode): Likewise.
15241 (s390_unwind_word_mode): Likewise.
15242 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15244 * target.def (mode_rep_extended): Likewise.
15245 (valid_pointer_mode): Likewise.
15246 (addr_space.valid_pointer_mode): Likewise.
15247 (eh_return_filter_mode): Return a scalar_int_mode rather than
15249 (libgcc_cmp_return_mode): Likewise.
15250 (libgcc_shift_count_mode): Likewise.
15251 (unwind_word_mode): Likewise.
15252 (addr_space.pointer_mode): Likewise.
15253 (addr_space.address_mode): Likewise.
15254 * doc/tm.texi: Regenerate.
15255 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
15257 (do_jump): Use scalar_int_mode for local variables.
15258 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
15259 rather than a machine_mode.
15260 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
15261 (scompare_loc_descriptor_wide): Likewise.
15262 (scompare_loc_descriptor_narrow): Likewise.
15263 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
15265 * except.c (sjlj_emit_dispatch_table): Likewise.
15266 (expand_builtin_eh_copy_values): Likewise.
15267 * explow.c (convert_memory_address_addr_space_1): Likewise.
15268 Take a scalar_int_mode rather than a machine_mode.
15269 (convert_memory_address_addr_space): Take a scalar_int_mode rather
15270 than a machine_mode.
15271 (memory_address_addr_space): Use scalar_int_mode for local variables.
15272 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
15273 rather than a machine_mode.
15274 * expmed.c (mask_rtx): Likewise.
15275 (init_expmed_one_conv): Likewise.
15276 (expand_mult_highpart_adjust): Likewise.
15277 (extract_high_half): Likewise.
15278 (expmed_mult_highpart_optab): Likewise.
15279 (expmed_mult_highpart): Likewise.
15280 (expand_smod_pow2): Likewise.
15281 (expand_sdiv_pow2): Likewise.
15282 (emit_store_flag_int): Likewise.
15283 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
15285 (extract_low_bits): Likewise.
15286 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
15288 * expr.c (pieces_addr::adjust): Likewise.
15289 (can_store_by_pieces): Likewise.
15290 (store_by_pieces): Likewise.
15291 (clear_by_pieces_1): Likewise.
15292 (expand_expr_addr_expr_1): Likewise.
15293 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
15294 (expand_expr_real_1): Likewise.
15295 (try_casesi): Likewise.
15296 * final.c (shorten_branches): Likewise.
15297 * fold-const.c (fold_convert_const_int_from_fixed): Change the
15298 type of "mode" to machine_mode.
15299 * internal-fn.c (expand_arith_overflow_result_store): Take a
15300 scalar_int_mode rather than a machine_mode.
15301 (expand_mul_overflow): Use scalar_int_mode for local variables.
15302 * loop-doloop.c (doloop_modify): Likewise.
15303 (doloop_optimize): Likewise.
15304 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
15305 than a machine_mode.
15306 (expand_doubleword_shift_condmove): Likewise.
15307 (expand_doubleword_shift): Likewise.
15308 (expand_doubleword_clz): Likewise.
15309 (expand_doubleword_popcount): Likewise.
15310 (expand_doubleword_parity): Likewise.
15311 (expand_absneg_bit): Use scalar_int_mode for local variables.
15312 (prepare_float_lib_cmp): Likewise.
15313 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
15314 rather than a machine_mode.
15315 (convert_memory_address_addr_space): Likewise.
15316 (get_mode_bounds): Likewise.
15317 (get_address_mode): Return a scalar_int_mode rather than a
15319 * rtlanal.c (get_address_mode): Likewise.
15320 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
15321 than a machine_mode.
15322 * targhooks.c (default_mode_rep_extended): Likewise.
15323 (default_valid_pointer_mode): Likewise.
15324 (default_addr_space_valid_pointer_mode): Likewise.
15325 (default_eh_return_filter_mode): Return a scalar_int_mode rather
15326 than a machine_mode.
15327 (default_libgcc_cmp_return_mode): Likewise.
15328 (default_libgcc_shift_count_mode): Likewise.
15329 (default_unwind_word_mode): Likewise.
15330 (default_addr_space_pointer_mode): Likewise.
15331 (default_addr_space_address_mode): Likewise.
15332 * targhooks.h (default_eh_return_filter_mode): Likewise.
15333 (default_libgcc_cmp_return_mode): Likewise.
15334 (default_libgcc_shift_count_mode): Likewise.
15335 (default_unwind_word_mode): Likewise.
15336 (default_addr_space_pointer_mode): Likewise.
15337 (default_addr_space_address_mode): Likewise.
15338 (default_mode_rep_extended): Take a scalar_int_mode rather than
15340 (default_valid_pointer_mode): Likewise.
15341 (default_addr_space_valid_pointer_mode): Likewise.
15342 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
15344 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
15345 rather than a machine_mode.
15346 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
15347 for local variables.
15348 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15349 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
15350 than a machine_mode.
15352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15353 Alan Hayward <alan.hayward@arm.com>
15354 David Sherwood <david.sherwood@arm.com>
15356 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
15357 the mode argument to scalar_int_mode.
15358 (do_jump_by_parts_zero_rtx): Likewise.
15359 (do_jump_by_parts_equality_rtx): Likewise.
15360 (do_jump_by_parts_greater): Take a mode argument.
15361 (do_jump_by_parts_equality): Likewise.
15362 (do_jump_1): Update calls accordingly.
15364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15365 Alan Hayward <alan.hayward@arm.com>
15366 David Sherwood <david.sherwood@arm.com>
15368 * is-a.h (safe_dyn_cast): New function.
15369 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
15370 (jump_table_for_label): Likewise.
15371 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
15372 instead of an rtx_insn *.
15373 (shorten_branches): Use dyn_cast instead of LABEL_P and
15374 JUMP_TABLE_DATA_P. Use jump_table_for_label and
15375 rtx_jump_table_data::get_data_mode.
15376 (final_scan_insn): Likewise.
15378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15379 Alan Hayward <alan.hayward@arm.com>
15380 David Sherwood <david.sherwood@arm.com>
15382 * combine.c (try_combine): Use is_a <scalar_int_mode> when
15383 trying to combine a full-register integer set with a subreg
15386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15387 Alan Hayward <alan.hayward@arm.com>
15388 David Sherwood <david.sherwood@arm.com>
15390 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
15391 that is always either address_mode or pointer_mode.
15393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15394 Alan Hayward <alan.hayward@arm.com>
15395 David Sherwood <david.sherwood@arm.com>
15397 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
15398 when the two are known to be equal.
15400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15401 Alan Hayward <alan.hayward@arm.com>
15402 David Sherwood <david.sherwood@arm.com>
15404 * simplify-rtx.c (simplify_const_unary_operation): Use
15405 is_a <scalar_int_mode> instead of checking for a nonzero
15406 precision. Forcibly convert op_mode to a scalar_int_mode
15407 in that case. More clearly differentiate the operand and
15408 result modes and use the former when deciding what the value
15409 of a count-bits operation should be. Use is_int_mode instead
15410 of checking for a MODE_INT. Remove redundant check for whether
15411 this mode has a zero precision.
15413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15414 Alan Hayward <alan.hayward@arm.com>
15415 David Sherwood <david.sherwood@arm.com>
15417 * optabs.c (widen_leading): Change the type of the mode argument
15418 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
15419 (widen_bswap): Likewise.
15420 (expand_parity): Likewise.
15421 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
15422 (expand_ffs): Likewise.
15423 (epand_unop): Check for scalar integer modes before calling the
15426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15427 Alan Hayward <alan.hayward@arm.com>
15428 David Sherwood <david.sherwood@arm.com>
15430 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
15432 (expand_expr_real_1): Update call accordingly.
15434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15435 Alan Hayward <alan.hayward@arm.com>
15436 David Sherwood <david.sherwood@arm.com>
15438 * expmed.c (store_bit_field_using_insv): Add op0_mode and
15439 value_mode arguments. Use scalar_int_mode internally.
15440 (store_bit_field_1): Rename the new integer mode from imode
15441 to op0_mode and use it instead of GET_MODE (op0). Update calls
15442 to store_split_bit_field, store_bit_field_using_insv and
15443 store_fixed_bit_field.
15444 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
15445 Use scalar_int_mode internally. Use a bit count rather than a mode
15446 when calculating the largest bit size for get_best_mode.
15447 Update calls to store_split_bit_field and store_fixed_bit_field_1.
15448 (store_fixed_bit_field_1): Add mode and value_mode arguments.
15449 Remove assertion that OP0 has a scalar integer mode.
15450 (store_split_bit_field): Add op0_mode and value_mode arguments.
15451 Update calls to extract_fixed_bit_field.
15452 (extract_bit_field_using_extv): Add an op0_mode argument.
15453 Use scalar_int_mode internally.
15454 (extract_bit_field_1): Rename the new integer mode from imode to
15455 op0_mode and use it instead of GET_MODE (op0). Update calls to
15456 extract_split_bit_field, extract_bit_field_using_extv and
15457 extract_fixed_bit_field.
15458 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
15459 to extract_split_bit_field and extract_fixed_bit_field_1.
15460 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
15461 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
15462 on the target mode.
15463 (extract_split_bit_field): Add an op0_mode argument. Update call
15464 to extract_fixed_bit_field.
15466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15467 Alan Hayward <alan.hayward@arm.com>
15468 David Sherwood <david.sherwood@arm.com>
15470 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
15471 * explow.c (hard_function_value): Likewise.
15472 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
15473 convert_to_mode call outside the loop.
15474 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
15475 for the mode iterator. Require the mode specified by max_pieces
15477 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
15479 (copy_blkmode_to_reg): Likewise.
15480 (set_storage_via_setmem): Likewise.
15481 * optabs.c (prepare_cmp_insn): Likewise.
15482 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15483 * stor-layout.c (finish_bitfield_representative): Likewise.
15485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15486 Alan Hayward <alan.hayward@arm.com>
15487 David Sherwood <david.sherwood@arm.com>
15489 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
15490 * expr.c (convert_move): Use them.
15491 (convert_modes): Likewise.
15492 (store_expr_with_bounds): Likewise.
15494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15495 Alan Hayward <alan.hayward@arm.com>
15496 David Sherwood <david.sherwood@arm.com>
15498 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
15499 parameter for the mode of "x". Remove the "known_x", "known_mode"
15500 and "known_ret" arguments. Change the type of the mode argument
15501 to scalar_int_mode.
15502 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
15503 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
15504 (reg_num_sign_bit_copies_for_combine): Likewise.
15505 * rtlanal.c (nonzero_bits1): Likewise.
15506 (num_sign_bit_copies1): Likewise.
15507 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
15508 (reg_num_sign_bit_copies_general): Likewise.
15509 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
15510 (reg_nonzero_bits_general): Likewise.
15512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15513 Alan Hayward <alan.hayward@arm.com>
15514 David Sherwood <david.sherwood@arm.com>
15516 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
15517 than in subroutines. Return 1 for non-integer modes.
15518 (cached_num_sign_bit_copies): Change the type of the mode parameter
15519 to scalar_int_mode.
15520 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
15521 classes. Handle CONST_INT_P first and then check whether X also
15522 has a scalar integer mode. Check the same thing for inner registers
15523 of a SUBREG and for values that are being extended or truncated.
15525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15526 Alan Hayward <alan.hayward@arm.com>
15527 David Sherwood <david.sherwood@arm.com>
15529 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
15530 in subroutines. Return the mode mask for non-integer modes.
15531 (cached_nonzero_bits): Change the type of the mode parameter
15532 to scalar_int_mode.
15533 (nonzero_bits1): Likewise. Remove early exit for other mode
15534 classes. Handle CONST_INT_P first and then check whether X
15535 also has a scalar integer mode.
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 * expr.c (widest_int_mode_for_size): Make the comment match the code.
15542 Return a scalar_int_mode and assert that the size is greater than
15544 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
15545 (op_by_pieces_d::op_by_pieces_d): Likewise.
15546 (op_by_pieces_d::run): Likewise.
15547 (can_store_by_pieces): Likewise.
15549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15550 Alan Hayward <alan.hayward@arm.com>
15551 David Sherwood <david.sherwood@arm.com>
15553 * combine.c (extract_left_shift): Add a mode argument and update
15555 (make_compound_operation_int): Change the type of the mode parameter
15556 to scalar_int_mode and update the call to extract_left_shift.
15558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15559 Alan Hayward <alan.hayward@arm.com>
15560 David Sherwood <david.sherwood@arm.com>
15562 * combine.c (simplify_and_const_int): Change the type of the mode
15563 parameter to scalar_int_mode.
15564 (simplify_and_const_int_1): Likewise. Update recursive call.
15566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15567 Alan Hayward <alan.hayward@arm.com>
15568 David Sherwood <david.sherwood@arm.com>
15570 * combine.c (simplify_compare_const): Check that the mode is a
15571 scalar_int_mode (rather than VOIDmode) before testing its
15573 (simplify_comparison): Move COMPARISON_P handling out of the
15574 loop and restrict the latter part of the loop to scalar_int_modes.
15575 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
15576 and when considering SUBREG_REGs. Use is_int_mode instead of
15577 checking GET_MODE_CLASS against MODE_INT.
15579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15580 Alan Hayward <alan.hayward@arm.com>
15581 David Sherwood <david.sherwood@arm.com>
15583 * combine.c (try_widen_shift_mode): Move check for equal modes to...
15584 (simplify_shift_const_1): ...here. Use scalar_int_mode for
15585 shift_unit_mode and for modes involved in scalar shifts.
15587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15588 Alan Hayward <alan.hayward@arm.com>
15589 David Sherwood <david.sherwood@arm.com>
15591 * combine.c (force_int_to_mode): New function, split out from...
15592 (force_to_mode): ...here. Keep xmode up-to-date and use it
15593 instead of GET_MODE (x).
15595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15596 Alan Hayward <alan.hayward@arm.com>
15597 David Sherwood <david.sherwood@arm.com>
15599 * optabs-query.h (extraction_insn::struct_mode): Change type to
15600 opt_scalar_int_mode and update comment.
15601 (extraction_insn::field_mode): Change type to scalar_int_mode.
15602 (extraction_insn::pos_mode): Likewise.
15603 * combine.c (make_extraction): Update accordingly.
15604 * optabs-query.c (get_traditional_extraction_insn): Likewise.
15605 (get_optab_extraction_insn): Likewise.
15606 * recog.c (simplify_while_replacing): Likewise.
15607 * expmed.c (narrow_bit_field_mem): Change the type of the mode
15608 parameter to opt_scalar_int_mode.
15610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15611 Alan Hayward <alan.hayward@arm.com>
15612 David Sherwood <david.sherwood@arm.com>
15614 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
15615 to a scalar_int_mode instead of a machine_mode.
15616 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
15617 (get_best_mode): Return a boolean and use a pointer argument to store
15618 the selected mode. Replace the limit mode parameter with a bit limit.
15619 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
15620 for the values returned by bit_field_mode_iterator::next_mode.
15621 (store_bit_field): Update call to get_best_mode.
15622 (store_fixed_bit_field): Likewise.
15623 (extract_fixed_bit_field): Likewise.
15624 * expr.c (optimize_bitfield_assignment_op): Likewise.
15625 * fold-const.c (optimize_bit_field_compare): Likewise.
15626 (fold_truth_andor_1): Likewise.
15627 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
15628 Update for new type of m_mode.
15629 (get_best_mode): As above.
15631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15632 Alan Hayward <alan.hayward@arm.com>
15633 David Sherwood <david.sherwood@arm.com>
15635 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
15636 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
15637 (store_bit_field): Check is_a <scalar_int_mode> before calling
15638 strict_volatile_bitfield_p.
15639 (extract_bit_field): Likewise.
15641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15642 Alan Hayward <alan.hayward@arm.com>
15643 David Sherwood <david.sherwood@arm.com>
15645 * target.def (cstore_mode): Return a scalar_int_mode.
15646 * doc/tm.texi: Regenerate.
15647 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
15648 * targhooks.h (default_cstore_mode): Likewise.
15649 * targhooks.c (default_cstore_mode): Likewise, using a forced
15651 * expmed.c (emit_cstore): Expect the target of the cstore to be
15654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15655 Alan Hayward <alan.hayward@arm.com>
15656 David Sherwood <david.sherwood@arm.com>
15658 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
15660 (niter_desc): Likewise mode.
15661 (iv_analyze): Add a mode parameter.
15663 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
15664 and change its type to scalar_int_mode.
15665 * loop-iv.c: Update commentary at head of file.
15666 (iv_constant): Pass the mode paraeter before the rtx it describes
15667 and change its type to scalar_int_mode. Remove VOIDmode handling.
15668 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
15669 (iv_extend): Likewise.
15670 (shorten_into_mode): Likewise.
15671 (iv_add): Use scalar_int_mode.
15672 (iv_mult): Likewise.
15673 (iv_shift): Likewise.
15674 (canonicalize_iv_subregs): Likewise.
15675 (get_biv_step_1): Pass the outer_mode parameter before the rtx
15676 it describes and change its mode to scalar_int_mode. Also change
15677 the type of the returned inner_mode to scalar_int_mode.
15678 (get_biv_step): Likewise, turning outer_mode from a pointer
15679 into a direct parameter. Update call to get_biv_step_1.
15680 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
15681 iv_constant and get_biv_step.
15682 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
15683 and change its type to scalar_int_mode. Don't initialise iv->mode
15684 to VOIDmode and remove later checks for its still being VOIDmode.
15685 Update calls to iv_analyze_op and iv_analyze_expr. Check
15686 is_a <scalar_int_mode> when changing the mode under consideration.
15687 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
15688 Update call to iv_analyze_expr.
15689 (iv_analyze_op): Add a mode parameter. Reject subregs whose
15690 inner register is not also a scalar_int_mode. Update call to
15692 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
15693 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
15694 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
15695 separate mode class checks. Update calls to iv_analyze. Remove
15696 fix-up of VOIDmodes after iv_analyze_biv.
15697 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
15698 don't have a scalar_int_mode. Update call to biv_p.
15700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15701 Alan Hayward <alan.hayward@arm.com>
15702 David Sherwood <david.sherwood@arm.com>
15704 * cfgexpand.c (convert_debug_memory_address): Use
15705 as_a <scalar_int_mode>.
15706 * combine.c (expand_compound_operation): Likewise.
15707 (make_extraction): Likewise.
15708 (change_zero_ext): Likewise.
15709 (simplify_comparison): Likewise.
15710 * cse.c (cse_insn): Likewise.
15711 * dwarf2out.c (minmax_loc_descriptor): Likewise.
15712 (mem_loc_descriptor): Likewise.
15713 (loc_descriptor): Likewise.
15714 * expmed.c (init_expmed_one_mode): Likewise.
15715 (synth_mult): Likewise.
15716 (emit_store_flag_1): Likewise.
15717 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
15718 of a comparison with size.
15719 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
15720 (reduce_to_bit_field_precision): Likewise.
15721 * function.c (expand_function_end): Likewise.
15722 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
15723 * loop-doloop.c (doloop_modify): Likewise.
15724 * optabs.c (expand_binop): Likewise.
15725 (expand_unop): Likewise.
15726 (expand_copysign_absneg): Likewise.
15727 (prepare_cmp_insn): Likewise.
15728 (maybe_legitimize_operand): Likewise.
15729 * recog.c (const_scalar_int_operand): Likewise.
15730 * rtlanal.c (get_address_mode): Likewise.
15731 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15732 (simplify_cond_clz_ctz): Likewise.
15733 * tree-nested.c (get_nl_goto_field): Likewise.
15734 * tree.c (build_vector_type_for_mode): Likewise.
15735 * var-tracking.c (use_narrower_mode): 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 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
15742 * builtins.c (expand_builtin_signbit): Use it.
15743 * cfgexpand.c (expand_debug_expr): Likewise.
15744 * dojump.c (do_jump): Likewise.
15745 (do_compare_and_jump): Likewise.
15746 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
15747 * expmed.c (make_tree): Likewise.
15748 * expr.c (expand_expr_real_2): Likewise.
15749 (expand_expr_real_1): Likewise.
15750 (try_casesi): Likewise.
15751 * fold-const-call.c (fold_const_call_ss): Likewise.
15752 * fold-const.c (unextend): Likewise.
15753 (extract_muldiv_1): Likewise.
15754 (fold_single_bit_test): Likewise.
15755 (native_encode_int): Likewise.
15756 (native_encode_string): Likewise.
15757 (native_interpret_int): Likewise.
15758 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
15759 * internal-fn.c (expand_addsub_overflow): Likewise.
15760 (expand_neg_overflow): Likewise.
15761 (expand_mul_overflow): Likewise.
15762 (expand_arith_overflow): Likewise.
15763 * match.pd: Likewise.
15764 * stor-layout.c (layout_type): Likewise.
15765 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15766 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
15767 * tree-ssanames.c (get_range_info): Likewise.
15768 * tree-switch-conversion.c (array_value_type) Likewise.
15769 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
15770 (vect_recog_divmod_pattern): Likewise.
15771 (vect_recog_mixed_size_cond_pattern): Likewise.
15772 * tree-vrp.c (extract_range_basic): Likewise.
15773 (simplify_float_conversion_using_ranges): Likewise.
15774 * tree.c (int_fits_type_p): Likewise.
15775 * ubsan.c (instrument_bool_enum_load): Likewise.
15776 * varasm.c (mergeable_string_section): Likewise.
15777 (narrowing_initializer_constant_valid_p): Likewise.
15778 (output_constant): Likewise.
15780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15781 Alan Hayward <alan.hayward@arm.com>
15782 David Sherwood <david.sherwood@arm.com>
15784 * machmode.h (NARROWEST_INT_MODE): New macro.
15785 * expr.c (alignment_for_piecewise_move): Use it instead of
15786 GET_CLASS_NARROWEST_MODE (MODE_INT).
15787 (push_block): Likewise.
15788 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
15790 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15793 Alan Hayward <alan.hayward@arm.com>
15794 David Sherwood <david.sherwood@arm.com>
15796 * postreload.c (move2add_valid_value_p): Change the type of the
15797 mode parameter to scalar_int_mode.
15798 (move2add_use_add2_insn): Add a mode parameter and use it instead
15800 (move2add_use_add3_insn): Likewise.
15801 (reload_cse_move2add): Update accordingly.
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 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
15809 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
15810 * optabs.c (expand_unop): Likewise.
15812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15813 Alan Hayward <alan.hayward@arm.com>
15814 David Sherwood <david.sherwood@arm.com>
15816 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
15817 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
15818 (popcount_loc_descriptor): Likewise.
15819 (bswap_loc_descriptor): Likewise.
15820 (rotate_loc_descriptor): Likewise.
15821 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
15822 calling the functions above.
15824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15825 Alan Hayward <alan.hayward@arm.com>
15826 David Sherwood <david.sherwood@arm.com>
15828 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
15830 (try_combine): Likewise.
15831 (simplify_if_then_else): Likewise.
15832 * cse.c (cse_insn): Likewise.
15833 * dwarf2out.c (mem_loc_descriptor): Likewise.
15834 * emit-rtl.c (gen_lowpart_common): Likewise.
15835 * simplify-rtx.c (simplify_truncation): Likewise.
15836 (simplify_binary_operation_1): Likewise.
15837 (simplify_const_relational_operation): Likewise.
15838 (simplify_ternary_operation): Likewise.
15839 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
15841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15842 Alan Hayward <alan.hayward@arm.com>
15843 David Sherwood <david.sherwood@arm.com>
15845 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
15846 * reload.c (push_reload): Likewise.
15847 (find_reloads): Likewise.
15849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15850 Alan Hayward <alan.hayward@arm.com>
15851 David Sherwood <david.sherwood@arm.com>
15853 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
15854 (make_compound_operation_int): Likewise.
15855 (change_zero_ext): Likewise.
15856 * expr.c (convert_move): Likewise.
15857 (convert_modes): Likewise.
15858 * fwprop.c (forward_propagate_subreg): Likewise.
15859 * loop-iv.c (get_biv_step_1): Likewise.
15860 * optabs.c (widen_operand): Likewise.
15861 * postreload.c (move2add_valid_value_p): Likewise.
15862 * recog.c (simplify_while_replacing): Likewise.
15863 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15864 (simplify_binary_operation_1): Likewise. Remove redundant
15865 mode equality check.
15867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15868 Alan Hayward <alan.hayward@arm.com>
15869 David Sherwood <david.sherwood@arm.com>
15871 * combine.c (combine_simplify_rtx): Add checks for
15872 is_a <scalar_int_mode>.
15873 (simplify_if_then_else): Likewise.
15874 (make_field_assignment): Likewise.
15875 (simplify_comparison): Likewise.
15876 * ifcvt.c (noce_try_bitop): Likewise.
15877 * loop-invariant.c (canonicalize_address_mult): Likewise.
15878 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15881 Alan Hayward <alan.hayward@arm.com>
15882 David Sherwood <david.sherwood@arm.com>
15884 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
15885 is_a <scalar_int_mode> instead of != BLKmode.
15887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15888 Alan Hayward <alan.hayward@arm.com>
15889 David Sherwood <david.sherwood@arm.com>
15891 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15892 instead of != VOIDmode.
15893 * combine.c (if_then_else_cond): Likewise.
15894 (change_zero_ext): Likewise.
15895 * dwarf2out.c (mem_loc_descriptor): Likewise.
15896 (loc_descriptor): Likewise.
15897 * rtlanal.c (canonicalize_condition): Likewise.
15898 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15901 Alan Hayward <alan.hayward@arm.com>
15902 David Sherwood <david.sherwood@arm.com>
15904 * simplify-rtx.c (simplify_binary_operation_1): Use
15905 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15908 Alan Hayward <alan.hayward@arm.com>
15909 David Sherwood <david.sherwood@arm.com>
15911 * wide-int.h (int_traits<unsigned char>) New class.
15912 (int_traits<unsigned short>) Likewise.
15913 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15914 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15916 * combine.c (set_nonzero_bits_and_sign_copies): Use
15917 is_a <scalar_int_mode>.
15918 (find_split_point): Likewise.
15919 (combine_simplify_rtx): Likewise.
15920 (simplify_logical): Likewise.
15921 (expand_compound_operation): Likewise.
15922 (expand_field_assignment): Likewise.
15923 (make_compound_operation): Likewise.
15924 (extended_count): Likewise.
15925 (change_zero_ext): Likewise.
15926 (simplify_comparison): Likewise.
15927 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15928 (ucompare_loc_descriptor): Likewise.
15929 (minmax_loc_descriptor): Likewise.
15930 (mem_loc_descriptor): Likewise.
15931 (loc_descriptor): Likewise.
15932 * expmed.c (init_expmed_one_mode): Likewise.
15933 * lra-constraints.c (lra_constraint_offset): Likewise.
15934 * optabs.c (prepare_libcall_arg): Likewise.
15935 * postreload.c (move2add_note_store): Likewise.
15936 * reload.c (operands_match_p): Likewise.
15937 * rtl.h (load_extend_op): Likewise.
15938 * rtlhooks.c (gen_lowpart_general): Likewise.
15939 * simplify-rtx.c (simplify_truncation): Likewise.
15940 (simplify_unary_operation_1): Likewise.
15941 (simplify_binary_operation_1): Likewise.
15942 (simplify_const_binary_operation): Likewise.
15943 (simplify_const_relational_operation): Likewise.
15944 (simplify_subreg): Likewise.
15945 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15946 * var-tracking.c (adjust_mems): Likewise.
15947 (prepare_call_arguments): Likewise.
15949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15950 Alan Hayward <alan.hayward@arm.com>
15951 David Sherwood <david.sherwood@arm.com>
15953 * machmode.h (is_int_mode): New fuction.
15954 * combine.c (find_split_point): Use it.
15955 (combine_simplify_rtx): Likewise.
15956 (simplify_if_then_else): Likewise.
15957 (simplify_set): Likewise.
15958 (simplify_shift_const_1): Likewise.
15959 (simplify_comparison): Likewise.
15960 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15961 * cse.c (notreg_cost): Likewise.
15962 (cse_insn): Likewise.
15963 * cselib.c (cselib_lookup_1): Likewise.
15964 * dojump.c (do_jump_1): Likewise.
15965 (do_compare_rtx_and_jump): Likewise.
15966 * dse.c (get_call_args): Likewise.
15967 * dwarf2out.c (rtl_for_decl_init): Likewise.
15968 (native_encode_initializer): Likewise.
15969 * expmed.c (emit_store_flag_1): Likewise.
15970 (emit_store_flag): Likewise.
15971 * expr.c (convert_modes): Likewise.
15972 (store_field): Likewise.
15973 (expand_expr_real_1): Likewise.
15974 * fold-const.c (fold_read_from_constant_string): Likewise.
15975 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15976 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15977 * optabs.c (expand_binop): Likewise.
15978 (expand_unop): Likewise.
15979 (expand_abs_nojump): Likewise.
15980 (expand_one_cmpl_abs_nojump): Likewise.
15981 * simplify-rtx.c (mode_signbit_p): Likewise.
15982 (val_signbit_p): Likewise.
15983 (val_signbit_known_set_p): Likewise.
15984 (val_signbit_known_clear_p): Likewise.
15985 (simplify_relational_operation_1): Likewise.
15986 * tree.c (vector_type_mode): Likewise.
15988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15989 Alan Hayward <alan.hayward@arm.com>
15990 David Sherwood <david.sherwood@arm.com>
15992 * machmode.h (smallest_mode_for_size): Fix formatting.
15993 (smallest_int_mode_for_size): New function.
15994 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15995 instead of smallest_mode_for_size.
15996 * combine.c (make_extraction): Likewise.
15997 * config/arc/arc.c (arc_expand_movmem): Likewise.
15998 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15999 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16000 * config/s390/s390.c (s390_expand_insv): Likewise.
16001 * config/sparc/sparc.c (assign_int_registers): Likewise.
16002 * config/spu/spu.c (spu_function_value): Likewise.
16003 (spu_function_arg): Likewise.
16004 * coverage.c (get_gcov_type): Likewise.
16005 (get_gcov_unsigned_t): Likewise.
16006 * dse.c (find_shift_sequence): Likewise.
16007 * expmed.c (store_bit_field_1): Likewise.
16008 * expr.c (convert_move): Likewise.
16009 (store_field): Likewise.
16010 * internal-fn.c (expand_arith_overflow): Likewise.
16011 * optabs-query.c (get_best_extraction_insn): Likewise.
16012 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16013 * stor-layout.c (layout_type): Likewise.
16014 (initialize_sizetypes): Likewise.
16015 * targhooks.c (default_get_mask_mode): Likewise.
16016 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16019 Alan Hayward <alan.hayward@arm.com>
16020 David Sherwood <david.sherwood@arm.com>
16022 * machmode.h (opt_mode::else_blk): New function.
16023 (int_mode_for_mode): Declare.
16024 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16025 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16027 * cfgexpand.c (expand_debug_expr): Likewise.
16028 * combine.c (gen_lowpart_or_truncate): Likewise.
16029 (gen_lowpart_for_combine): Likewise.
16030 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16031 * config/avr/avr.c (avr_to_int_mode): Likewise.
16032 (avr_out_plus_1): Likewise.
16033 (avr_out_plus): Likewise.
16034 (avr_out_round): Likewise.
16035 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16036 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16037 (s390_expand_vcond): Likewise.
16038 * config/spu/spu.c (spu_split_immediate): Likewise.
16039 (spu_expand_mov): Likewise.
16040 * dse.c (get_stored_val): Likewise.
16041 * expmed.c (store_bit_field_1): Likewise.
16042 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16044 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16045 (extract_low_bits): Likewise.
16046 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16047 handling rather than repeating the check.
16048 (emit_group_store): Likewise.
16049 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16050 * optabs.c (expand_absneg_bit): Likewise.
16051 (expand_copysign_absneg): Likewise.
16052 (expand_copysign_bit): Likewise.
16053 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16054 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16055 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16056 * var-tracking.c (prepare_call_arguments): Likewise.
16057 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16058 int_mode_for_mode instead of mode_for_size.
16059 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
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 (int_mode_for_size): New function.
16066 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16067 instead of mode_for_size.
16068 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16070 * combine.c (expand_field_assignment): Use int_mode_for_size
16071 instead of mode_for_size.
16072 (make_extraction): Likewise.
16073 (simplify_shift_const_1): Likewise.
16074 (simplify_comparison): Likewise.
16075 * dojump.c (do_jump): Likewise.
16076 * dwarf2out.c (mem_loc_descriptor): Likewise.
16077 * emit-rtl.c (init_derived_machine_modes): Likewise.
16078 * expmed.c (flip_storage_order): Likewise.
16079 (convert_extracted_bit_field): Likewise.
16080 * expr.c (copy_blkmode_from_reg): Likewise.
16081 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16082 * internal-fn.c (expand_mul_overflow): Likewise.
16083 * lower-subreg.c (simple_move): Likewise.
16084 * optabs-libfuncs.c (init_optabs): Likewise.
16085 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16086 * tree.c (vector_type_mode): Likewise.
16087 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16088 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16089 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16090 * tree-vect-stmts.c (vectorizable_load): Likewise.
16091 (vectorizable_store): Likewise.
16093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16094 Alan Hayward <alan.hayward@arm.com>
16095 David Sherwood <david.sherwood@arm.com>
16097 * coretypes.h (pod_mode): New type.
16098 (scalar_int_mode_pod): New typedef.
16099 * machmode.h (pod_mode): New class.
16100 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16101 * genmodes.c (emit_mode_int_n): Update accordingly.
16102 * lower-subreg.h (target_lower_subreg): Change type to
16103 scalar_int_mode_pod.
16104 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16105 scalar_int_mode_pod.
16107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16108 Alan Hayward <alan.hayward@arm.com>
16109 David Sherwood <david.sherwood@arm.com>
16111 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16112 machine_mode to scalar_int_mode.
16113 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16114 (rs6000_option_override_internal): Remove cast to int.
16115 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16116 machine_mode to scalar_int_mode.
16117 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16118 (rs6000_option_override_internal): Remove cast to int.
16119 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16120 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16122 * config/s390/s390.c (s390_expand_builtin): Likewise.
16123 * coretypes.h (scalar_int_mode): New type.
16124 (opt_scalar_int_mode): New typedef.
16125 * machmode.h (scalar_int_mode): New class.
16126 (scalar_int_mode::includes_p): New function.
16127 (byte_mode): Change type to scalar_int_mode.
16128 (word_mode): Likewise.
16129 (ptr_mode): Likewise.
16130 * emit-rtl.c (byte_mode): Likewise.
16131 (word_mode): Likewise.
16132 (ptr_mode): Likewise.
16133 (init_derived_machine_modes): Update accordingly.
16134 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16135 and MODE_PARTIAL_INT.
16136 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16137 opt_scalar_int_mode.
16139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16140 Alan Hayward <alan.hayward@arm.com>
16141 David Sherwood <david.sherwood@arm.com>
16143 * target.def (libgcc_floating_mode_supported_p): Take a
16145 * doc/tm.texi: Regenerate.
16146 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16148 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16149 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16153 Alan Hayward <alan.hayward@arm.com>
16154 David Sherwood <david.sherwood@arm.com>
16156 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16157 * doc/tm.texi: Regenerate.
16158 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16159 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16160 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16161 * targhooks.h (default_floatn_mode): Likewise.
16162 * targhooks.c (default_floatn_mode): Likewise.
16163 * tree.c (build_common_tree_nodes): Update accordingly.
16165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16166 Alan Hayward <alan.hayward@arm.com>
16167 David Sherwood <david.sherwood@arm.com>
16169 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16170 (mode_iterator::iterate_p): Likewise.
16171 (mode_iterator::get_wider): Likewise.
16172 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16175 Alan Hayward <alan.hayward@arm.com>
16176 David Sherwood <david.sherwood@arm.com>
16178 * coretypes.h (opt_scalar_float_mode): New typedef.
16179 * machmode.h (float_mode_for_size): New function.
16180 * emit-rtl.c (double_mode): Delete.
16181 (init_emit_once): Use float_mode_for_size.
16182 * stor-layout.c (layout_type): Likewise.
16183 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16186 Alan Hayward <alan.hayward@arm.com>
16187 David Sherwood <david.sherwood@arm.com>
16189 * output.h (assemble_real): Take a scalar_float_mode.
16190 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16191 * config/arm/arm.md (consttable_4): Likewise.
16192 (consttable_8): Likewise.
16193 (consttable_16): Likewise.
16194 * config/mips/mips.md (consttable_float): Likewise.
16195 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16196 * varasm.c (assemble_real): Take a scalar_float_mode.
16197 (output_constant_pool_2): Update accordingly.
16198 (output_constant): Likewise.
16200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16201 Alan Hayward <alan.hayward@arm.com>
16202 David Sherwood <david.sherwood@arm.com>
16204 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16205 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16206 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16207 (native_encode_real): Likewise.
16208 (native_interpret_real): Likewise.
16209 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16210 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16213 Alan Hayward <alan.hayward@arm.com>
16214 David Sherwood <david.sherwood@arm.com>
16216 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16217 <scalar_float_mode>. Simplify.
16218 (gen_extend_conv_libfunc): Likewise.
16220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16221 Alan Hayward <alan.hayward@arm.com>
16222 David Sherwood <david.sherwood@arm.com>
16224 * coretypes.h (scalar_float_mode): New type.
16225 * machmode.h (mode_traits::from_int): Use machine_mode if
16226 USE_ENUM_MODES is defined.
16227 (is_a): New function.
16229 (dyn_cast): Likewise.
16230 (scalar_float_mode): New class.
16231 (scalar_float_mode::includes_p): New function.
16232 (is_float_mode): Likewise.
16233 * gdbhooks.py (MachineModePrinter): New class.
16234 (build_pretty_printer): Use it for scalar_float_mode.
16235 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16236 (format_helper::format_helper): Turn into a template.
16237 * genmodes.c (get_mode_class): New function.
16238 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16239 or machine_mode if none.
16240 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16241 as_a <scalar_float_mode>.
16242 * dwarf2out.c (mem_loc_descriptor): Likewise.
16243 (insert_float): Likewise.
16244 (add_const_value_attribute): Likewise.
16245 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16246 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16247 (expand_unop): Update accordingly.
16248 (expand_abs_nojump): Likewise.
16249 (expand_copysign_absneg): Take a scalar_float_mode.
16250 (expand_copysign_bit): Likewise.
16251 (expand_copysign): Update accordingly.
16253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16254 Alan Hayward <alan.hayward@arm.com>
16255 David Sherwood <david.sherwood@arm.com>
16257 * coretypes.h (opt_mode): New class.
16258 * machmode.h (opt_mode): Likewise.
16259 (opt_mode::else_void): New function.
16260 (opt_mode::require): Likewise.
16261 (opt_mode::exists): Likewise.
16262 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
16263 (GET_MODE_2XWIDER_MODE): Likewise.
16264 (mode_iterator::get_wider): Update accordingly.
16265 (mode_iterator::get_2xwider): Likewise.
16266 (mode_iterator::get_known_wider): Likewise, turning into a template.
16267 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
16268 forcing a wider mode to exist.
16269 * config/cr16/cr16.h (LONG_REG_P): Likewise.
16270 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16271 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
16272 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
16273 * lower-subreg.c (init_lower_subreg): Likewise.
16274 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
16275 on the final iteration.
16276 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
16277 a wider mode exists before asking for a move pattern.
16278 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
16279 forcing a wider mode to exist.
16280 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
16281 returning false if no such mode exists.
16282 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
16283 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
16284 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
16285 Avoid checking for a MODE_INT if we already know the mode is not a
16287 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
16288 forcing a wider mode to exist.
16289 (expmed_mult_highpart_optab): Likewise.
16290 (expmed_mult_highpart): Likewise.
16291 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
16293 * lto-streamer-in.c (lto_input_mode_table): Likewise.
16294 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
16295 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16296 * internal-fn.c (expand_mul_overflow): Update use of
16297 GET_MODE_2XWIDER_MODE.
16298 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16299 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
16300 GET_MODE_WIDER_MODE.
16301 (convert_plusminus_to_widen): Likewise.
16302 * tree-switch-conversion.c (array_value_type): Likewise.
16303 * var-tracking.c (emit_note_insn_var_location): Likewise.
16304 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16305 Return false inside rather than outside the loop if no wider mode
16307 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
16308 and GET_MODE_2XWIDER_MODE
16309 (can_compare_p): Use else_void.
16310 * gdbhooks.py (OptMachineModePrinter): New class.
16311 (build_pretty_printer): Use it for opt_mode.
16313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16314 Alan Hayward <alan.hayward@arm.com>
16315 David Sherwood <david.sherwood@arm.com>
16317 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
16318 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
16320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16321 Alan Hayward <alan.hayward@arm.com>
16322 David Sherwood <david.sherwood@arm.com>
16324 * machmode.h (mode_traits): New structure.
16325 (get_narrowest_mode): New function.
16326 (mode_iterator::start): Likewise.
16327 (mode_iterator::iterate_p): Likewise.
16328 (mode_iterator::get_wider): Likewise.
16329 (mode_iterator::get_known_wider): Likewise.
16330 (mode_iterator::get_2xwider): Likewise.
16331 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
16332 (FOR_EACH_MODE): Likewise.
16333 (FOR_EACH_MODE_FROM): Likewise.
16334 (FOR_EACH_MODE_UNTIL): Likewise.
16335 (FOR_EACH_WIDER_MODE): Likewise.
16336 (FOR_EACH_2XWIDER_MODE): Likewise.
16337 * builtins.c (expand_builtin_strlen): Use new mode iterators.
16338 * combine.c (simplify_comparison): Likewise
16339 * config/i386/i386.c (type_natural_mode): Likewise.
16340 * cse.c (cse_insn): Likewise.
16341 * dse.c (find_shift_sequence): Likewise.
16342 * emit-rtl.c (init_derived_machine_modes): Likewise.
16343 (init_emit_once): Likewise.
16344 * explow.c (hard_function_value): Likewise.
16345 * expmed.c (extract_fixed_bit_field_1): Likewise.
16346 (extract_bit_field_1): Likewise.
16347 (expand_divmod): Likewise.
16348 (emit_store_flag_1): Likewise.
16349 * expr.c (init_expr_target): Likewise.
16350 (convert_move): Likewise.
16351 (alignment_for_piecewise_move): Likewise.
16352 (widest_int_mode_for_size): Likewise.
16353 (emit_block_move_via_movmem): Likewise.
16354 (copy_blkmode_to_reg): Likewise.
16355 (set_storage_via_setmem): Likewise.
16356 (compress_float_constant): Likewise.
16357 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16358 * optabs-query.c (get_best_extraction_insn): Likewise.
16359 * optabs.c (expand_binop): Likewise.
16360 (expand_twoval_unop): Likewise.
16361 (expand_twoval_binop): Likewise.
16362 (widen_leading): Likewise.
16363 (widen_bswap): Likewise.
16364 (expand_parity): Likewise.
16365 (expand_unop): Likewise.
16366 (prepare_cmp_insn): Likewise.
16367 (prepare_float_lib_cmp): Likewise.
16368 (expand_float): Likewise.
16369 (expand_fix): Likewise.
16370 (expand_sfix_optab): Likewise.
16371 * postreload.c (move2add_use_add2_insn): Likewise.
16372 * reg-stack.c (reg_to_stack): Likewise.
16373 * reginfo.c (choose_hard_reg_mode): Likewise.
16374 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16375 * stor-layout.c (mode_for_size): Likewise.
16376 (smallest_mode_for_size): Likewise.
16377 (mode_for_vector): Likewise.
16378 (finish_bitfield_representative): Likewise.
16379 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
16380 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
16381 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16382 * var-tracking.c (prepare_call_arguments): Likewise.
16384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16385 Alan Hayward <alan.hayward@arm.com>
16386 David Sherwood <david.sherwood@arm.com>
16388 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
16389 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
16390 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
16391 * machmode.h (mode_size): Move earlier in file.
16392 (mode_precision): Likewise.
16393 (mode_inner): Likewise.
16394 (mode_nunits): Likewise.
16395 (mode_unit_size): Likewise.
16396 (unit_unit_precision): Likewise.
16397 (mode_wider): Likewise.
16398 (mode_2xwider): Likewise.
16399 (machine_mode): New class.
16400 (mode_to_bytes): New function.
16401 (mode_to_bits): Likewise.
16402 (mode_to_precision): Likewise.
16403 (mode_to_inner): Likewise.
16404 (mode_to_unit_size): Likewise.
16405 (mode_to_unit_precision): Likewise.
16406 (mode_to_nunits): Likewise.
16407 (GET_MODE_SIZE): Use mode_to_bytes.
16408 (GET_MODE_BITSIZE): Use mode_to_bits.
16409 (GET_MODE_PRECISION): Use mode_to_precision.
16410 (GET_MODE_INNER): Use mode_to_inner.
16411 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
16412 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
16413 (GET_MODE_NUNITS): Use mode_to_nunits.
16414 * system.h (ALWAYS_INLINE): New macro.
16415 * config/powerpcspe/powerpcspe-c.c
16416 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
16417 int for arg1_mode and arg2_mode.
16419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16420 Alan Hayward <alan.hayward@arm.com>
16421 David Sherwood <david.sherwood@arm.com>
16423 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16424 Prefix mode names with E_ in case statements.
16425 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16426 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
16427 (aarch64_split_simd_move): Likewise.
16428 (aarch64_gen_storewb_pair): Likewise.
16429 (aarch64_gen_loadwb_pair): Likewise.
16430 (aarch64_gen_store_pair): Likewise.
16431 (aarch64_gen_load_pair): Likewise.
16432 (aarch64_get_condition_code_1): Likewise.
16433 (aarch64_constant_pool_reload_icode): Likewise.
16434 (get_rsqrte_type): Likewise.
16435 (get_rsqrts_type): Likewise.
16436 (get_recpe_type): Likewise.
16437 (get_recps_type): Likewise.
16438 (aarch64_gimplify_va_arg_expr): Likewise.
16439 (aarch64_simd_container_mode): Likewise.
16440 (aarch64_emit_load_exclusive): Likewise.
16441 (aarch64_emit_store_exclusive): Likewise.
16442 (aarch64_expand_compare_and_swap): Likewise.
16443 (aarch64_gen_atomic_cas): Likewise.
16444 (aarch64_emit_bic): Likewise.
16445 (aarch64_emit_atomic_swap): Likewise.
16446 (aarch64_emit_atomic_load_op): Likewise.
16447 (aarch64_evpc_trn): Likewise.
16448 (aarch64_evpc_uzp): Likewise.
16449 (aarch64_evpc_zip): Likewise.
16450 (aarch64_evpc_ext): Likewise.
16451 (aarch64_evpc_rev): Likewise.
16452 (aarch64_evpc_dup): Likewise.
16453 (aarch64_gen_ccmp_first): Likewise.
16454 (aarch64_gen_ccmp_next): Likewise.
16455 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16456 (alpha_emit_xfloating_libcall): Likewise.
16457 (emit_insxl): Likewise.
16458 (alpha_arg_type): Likewise.
16459 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
16460 (arc_preferred_simd_mode): Likewise.
16461 (arc_secondary_reload): Likewise.
16462 (get_arc_condition_code): Likewise.
16463 (arc_print_operand): Likewise.
16464 (arc_legitimate_constant_p): Likewise.
16465 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16466 * config/arc/arc.md (casesi_load): Likewise.
16467 (casesi_compact_jump): Likewise.
16468 * config/arc/predicates.md (proper_comparison_operator): Likewise.
16469 (cc_use_register): Likewise.
16470 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16471 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16472 (arm_init_iwmmxt_builtins): Likewise.
16473 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
16474 (neon_expand_vector_init): Likewise.
16475 (arm_attr_length_move_neon): Likewise.
16476 (maybe_get_arm_condition_code): Likewise.
16477 (arm_emit_vector_const): Likewise.
16478 (arm_preferred_simd_mode): Likewise.
16479 (arm_output_iwmmxt_tinsr): Likewise.
16480 (thumb1_output_casesi): Likewise.
16481 (thumb2_output_casesi): Likewise.
16482 (arm_emit_load_exclusive): Likewise.
16483 (arm_emit_store_exclusive): Likewise.
16484 (arm_expand_compare_and_swap): Likewise.
16485 (arm_evpc_neon_vuzp): Likewise.
16486 (arm_evpc_neon_vzip): Likewise.
16487 (arm_evpc_neon_vrev): Likewise.
16488 (arm_evpc_neon_vtrn): Likewise.
16489 (arm_evpc_neon_vext): Likewise.
16490 (arm_validize_comparison): Likewise.
16491 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
16492 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
16493 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
16494 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
16495 (c6x_preferred_simd_mode): Likewise.
16496 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
16497 (epiphany_rtx_costs): Likewise.
16498 * config/epiphany/predicates.md (proper_comparison_operator):
16500 * config/frv/frv.c (condexec_memory_operand): Likewise.
16501 (frv_emit_move): Likewise.
16502 (output_move_single): Likewise.
16503 (output_condmove_single): Likewise.
16504 (frv_hard_regno_mode_ok): Likewise.
16505 (frv_matching_accg_mode): Likewise.
16506 * config/h8300/h8300.c (split_adds_subs): Likewise.
16507 (h8300_rtx_costs): Likewise.
16508 (h8300_print_operand): Likewise.
16509 (compute_mov_length): Likewise.
16510 (output_logical_op): Likewise.
16511 (compute_logical_op_length): Likewise.
16512 (compute_logical_op_cc): Likewise.
16513 (h8300_shift_needs_scratch_p): Likewise.
16514 (output_a_shift): Likewise.
16515 (compute_a_shift_length): Likewise.
16516 (compute_a_shift_cc): Likewise.
16517 (expand_a_rotate): Likewise.
16518 (output_a_rotate): Likewise.
16519 * config/i386/i386.c (classify_argument): Likewise.
16520 (function_arg_advance_32): Likewise.
16521 (function_arg_32): Likewise.
16522 (function_arg_64): Likewise.
16523 (function_value_64): Likewise.
16524 (ix86_gimplify_va_arg): Likewise.
16525 (ix86_legitimate_constant_p): Likewise.
16526 (put_condition_code): Likewise.
16527 (split_double_mode): Likewise.
16528 (ix86_avx256_split_vector_move_misalign): Likewise.
16529 (ix86_expand_vector_logical_operator): Likewise.
16530 (ix86_split_idivmod): Likewise.
16531 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
16532 (ix86_build_const_vector): Likewise.
16533 (ix86_build_signbit_mask): Likewise.
16534 (ix86_match_ccmode): Likewise.
16535 (ix86_cc_modes_compatible): Likewise.
16536 (ix86_expand_branch): Likewise.
16537 (ix86_expand_sse_cmp): Likewise.
16538 (ix86_expand_sse_movcc): Likewise.
16539 (ix86_expand_int_sse_cmp): Likewise.
16540 (ix86_expand_vec_perm_vpermi2): Likewise.
16541 (ix86_expand_vec_perm): Likewise.
16542 (ix86_expand_sse_unpack): Likewise.
16543 (ix86_expand_int_addcc): Likewise.
16544 (ix86_split_to_parts): Likewise.
16545 (ix86_vectorize_builtin_gather): Likewise.
16546 (ix86_vectorize_builtin_scatter): Likewise.
16547 (avx_vpermilp_parallel): Likewise.
16548 (inline_memory_move_cost): Likewise.
16549 (ix86_tieable_integer_mode_p): Likewise.
16550 (x86_maybe_negate_const_int): Likewise.
16551 (ix86_expand_vector_init_duplicate): Likewise.
16552 (ix86_expand_vector_init_one_nonzero): Likewise.
16553 (ix86_expand_vector_init_one_var): Likewise.
16554 (ix86_expand_vector_init_concat): Likewise.
16555 (ix86_expand_vector_init_interleave): Likewise.
16556 (ix86_expand_vector_init_general): Likewise.
16557 (ix86_expand_vector_set): Likewise.
16558 (ix86_expand_vector_extract): Likewise.
16559 (emit_reduc_half): Likewise.
16560 (ix86_emit_i387_round): Likewise.
16561 (ix86_mangle_type): Likewise.
16562 (ix86_expand_round_sse4): Likewise.
16563 (expand_vec_perm_blend): Likewise.
16564 (canonicalize_vector_int_perm): Likewise.
16565 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
16566 (expand_vec_perm_1): Likewise.
16567 (expand_vec_perm_interleave3): Likewise.
16568 (expand_vec_perm_even_odd_pack): Likewise.
16569 (expand_vec_perm_even_odd_1): Likewise.
16570 (expand_vec_perm_broadcast_1): Likewise.
16571 (ix86_vectorize_vec_perm_const_ok): Likewise.
16572 (ix86_expand_vecop_qihi): Likewise.
16573 (ix86_expand_mul_widen_hilo): Likewise.
16574 (ix86_expand_sse2_abs): Likewise.
16575 (ix86_expand_pextr): Likewise.
16576 (ix86_expand_pinsr): Likewise.
16577 (ix86_preferred_simd_mode): Likewise.
16578 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
16579 * config/i386/sse.md (*andnot<mode>3): Likewise.
16580 (<mask_codefor><code><mode>3<mask_name>): Likewise.
16581 (*<code><mode>3): Likewise.
16582 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
16583 (ia64_expand_atomic_op): Likewise.
16584 (ia64_arg_type): Likewise.
16585 (ia64_mode_to_int): Likewise.
16586 (ia64_scalar_mode_supported_p): Likewise.
16587 (ia64_vector_mode_supported_p): Likewise.
16588 (expand_vec_perm_broadcast): Likewise.
16589 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
16590 (iq2000_function_arg_advance): Likewise.
16591 (iq2000_function_arg): Likewise.
16592 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
16593 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
16594 (m68k_libcall_value): Likewise.
16595 (m68k_function_value): Likewise.
16596 (sched_attr_op_type): Likewise.
16597 * config/mcore/mcore.c (mcore_output_move): Likewise.
16598 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
16600 (microblaze_function_arg): Likewise.
16601 * config/mips/mips.c (mips16_build_call_stub): Likewise.
16602 (mips_print_operand): Likewise.
16603 (mips_mode_ok_for_mov_fmt_p): Likewise.
16604 (mips_vector_mode_supported_p): Likewise.
16605 (mips_preferred_simd_mode): Likewise.
16606 (mips_expand_vpc_loongson_even_odd): Likewise.
16607 (mips_expand_vec_unpack): Likewise.
16608 (mips_expand_vi_broadcast): Likewise.
16609 (mips_expand_vector_init): Likewise.
16610 (mips_expand_vec_reduc): Likewise.
16611 (mips_expand_msa_cmp): Likewise.
16612 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
16613 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
16614 (cc_flags_for_mode): Likewise.
16615 * config/msp430/msp430.c (msp430_print_operand): Likewise.
16616 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
16617 (nds32_output_casesi_pc_relative): Likewise.
16618 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16619 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
16620 (nvptx_gen_unpack): Likewise.
16621 (nvptx_gen_pack): Likewise.
16622 (nvptx_gen_shuffle): Likewise.
16623 (nvptx_gen_wcast): Likewise.
16624 (nvptx_preferred_simd_mode): Likewise.
16625 * config/pa/pa.c (pa_secondary_reload): Likewise.
16626 * config/pa/predicates.md (base14_operand): Likewise.
16627 * config/powerpcspe/powerpcspe-c.c
16628 (altivec_resolve_overloaded_builtin): Likewise.
16629 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
16631 (rs6000_preferred_simd_mode): Likewise.
16632 (output_vec_const_move): Likewise.
16633 (rs6000_expand_vector_extract): Likewise.
16634 (rs6000_split_vec_extract_var): Likewise.
16635 (reg_offset_addressing_ok_p): Likewise.
16636 (rs6000_legitimate_offset_address_p): Likewise.
16637 (rs6000_legitimize_address): Likewise.
16638 (rs6000_emit_set_const): Likewise.
16639 (rs6000_const_vec): Likewise.
16640 (rs6000_emit_move): Likewise.
16641 (spe_build_register_parallel): Likewise.
16642 (rs6000_darwin64_record_arg_recurse): Likewise.
16643 (swap_selector_for_mode): Likewise.
16644 (spe_init_builtins): Likewise.
16645 (paired_init_builtins): Likewise.
16646 (altivec_init_builtins): Likewise.
16647 (do_load_for_compare): Likewise.
16648 (rs6000_generate_compare): Likewise.
16649 (rs6000_expand_float128_convert): Likewise.
16650 (emit_load_locked): Likewise.
16651 (emit_store_conditional): Likewise.
16652 (rs6000_output_function_epilogue): Likewise.
16653 (rs6000_handle_altivec_attribute): Likewise.
16654 (rs6000_function_value): Likewise.
16655 (emit_fusion_gpr_load): Likewise.
16656 (emit_fusion_p9_load): Likewise.
16657 (emit_fusion_p9_store): Likewise.
16658 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
16659 (fusion_gpr_mem_load): Likewise.
16660 (fusion_addis_mem_combo_load): Likewise.
16661 (fusion_addis_mem_combo_store): Likewise.
16662 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
16663 (fusion_gpr_mem_load): Likewise.
16664 (fusion_addis_mem_combo_load): Likewise.
16665 (fusion_addis_mem_combo_store): Likewise.
16666 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16668 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
16669 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16670 (rs6000_preferred_simd_mode): Likewise.
16671 (output_vec_const_move): Likewise.
16672 (rs6000_expand_vector_extract): Likewise.
16673 (rs6000_split_vec_extract_var): Likewise.
16674 (reg_offset_addressing_ok_p): Likewise.
16675 (rs6000_legitimate_offset_address_p): Likewise.
16676 (rs6000_legitimize_address): Likewise.
16677 (rs6000_emit_set_const): Likewise.
16678 (rs6000_const_vec): Likewise.
16679 (rs6000_emit_move): Likewise.
16680 (rs6000_darwin64_record_arg_recurse): Likewise.
16681 (swap_selector_for_mode): Likewise.
16682 (paired_init_builtins): Likewise.
16683 (altivec_init_builtins): Likewise.
16684 (rs6000_expand_float128_convert): Likewise.
16685 (emit_load_locked): Likewise.
16686 (emit_store_conditional): Likewise.
16687 (rs6000_output_function_epilogue): Likewise.
16688 (rs6000_handle_altivec_attribute): Likewise.
16689 (rs6000_function_value): Likewise.
16690 (emit_fusion_gpr_load): Likewise.
16691 (emit_fusion_p9_load): Likewise.
16692 (emit_fusion_p9_store): Likewise.
16693 * config/rx/rx.c (rx_gen_move_template): Likewise.
16694 (flags_from_mode): Likewise.
16695 * config/s390/predicates.md (s390_alc_comparison): Likewise.
16696 (s390_slb_comparison): Likewise.
16697 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
16698 (s390_vector_mode_supported_p): Likewise.
16699 (s390_cc_modes_compatible): Likewise.
16700 (s390_match_ccmode_set): Likewise.
16701 (s390_canonicalize_comparison): Likewise.
16702 (s390_emit_compare_and_swap): Likewise.
16703 (s390_branch_condition_mask): Likewise.
16704 (s390_rtx_costs): Likewise.
16705 (s390_secondary_reload): Likewise.
16706 (__SECONDARY_RELOAD_CASE): Likewise.
16707 (s390_expand_cs): Likewise.
16708 (s390_preferred_simd_mode): Likewise.
16709 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
16710 * config/sh/sh.c (sh_print_operand): Likewise.
16711 (dump_table): Likewise.
16712 (sh_secondary_reload): Likewise.
16713 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16714 * config/sh/sh.md (casesi_worker_1): Likewise.
16715 (casesi_worker_2): Likewise.
16716 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
16717 (fcc_comparison_operator): Likewise.
16718 * config/sparc/sparc.c (sparc_expand_move): Likewise.
16719 (emit_soft_tfmode_cvt): Likewise.
16720 (sparc_preferred_simd_mode): Likewise.
16721 (output_cbranch): Likewise.
16722 (sparc_print_operand): Likewise.
16723 (sparc_expand_vec_perm_bmask): Likewise.
16724 (vector_init_bshuffle): Likewise.
16725 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16726 (spu_vector_mode_supported_p): Likewise.
16727 (spu_expand_insv): Likewise.
16728 (spu_emit_branch_or_set): Likewise.
16729 (spu_handle_vector_attribute): Likewise.
16730 (spu_builtin_splats): Likewise.
16731 (spu_builtin_extract): Likewise.
16732 (spu_builtin_promote): Likewise.
16733 (spu_expand_sign_extend): Likewise.
16734 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16735 (tilegx_simd_int): Likewise.
16736 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
16737 (tilepro_simd_int): Likewise.
16738 * config/v850/v850.c (const_double_split): Likewise.
16739 (v850_print_operand): Likewise.
16740 (ep_memory_offset): Likewise.
16741 * config/vax/vax.c (vax_rtx_costs): Likewise.
16742 (vax_output_int_move): Likewise.
16743 (vax_output_int_add): Likewise.
16744 (vax_output_int_subtract): Likewise.
16745 * config/visium/predicates.md (visium_branch_operator): Likewise.
16746 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
16747 (visium_print_operand_address): Likewise.
16748 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16749 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
16750 (xtensa_expand_conditional_branch): Likewise.
16751 (xtensa_copy_incoming_a7): Likewise.
16752 (xtensa_output_literal): Likewise.
16753 * dfp.c (decimal_real_maxval): Likewise.
16754 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16757 Alan Hayward <alan.hayward@arm.com>
16758 David Sherwood <david.sherwood@arm.com>
16760 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
16761 (mode_nunits_inline): Likewise.
16762 (mode_inner_inline): Likewise.
16763 (mode_unit_size_inline): Likewise.
16764 (mode_unit_precision_inline): Likewise.
16765 (emit_insn_modes_h): Likewise. Also emit a #define of the
16767 (emit_mode_wider): Add an E_ prefix to mode names.
16768 (emit_mode_complex): Likewise.
16769 (emit_mode_inner): Likewise.
16770 (emit_mode_adjustments): Likewise.
16771 (emit_mode_int_n): Likewise.
16772 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
16773 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
16774 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
16775 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
16776 (CRC32_BUILTIN, ENTRY): Likewise.
16777 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
16778 (aarch64_pop_regs): Likewise.
16779 (aarch64_process_components): Likewise.
16780 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
16781 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
16782 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
16783 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
16784 * config/arm/arm.c (arm_init_libfuncs): Likewise.
16785 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
16787 * config/i386/i386-builtin.def (pcmpestr): Likewise.
16788 (pcmpistr): Likewise.
16789 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
16790 * config/mmix/mmix.c (mmix_output_condition): Likewise.
16791 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16793 * config/rl78/rl78.c (mduc_regs): Likewise.
16794 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16795 (htm_expand_builtin): Likewise.
16796 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
16797 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
16798 * config/xtensa/xtensa.c (print_operand): Likewise.
16799 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
16800 (NUM_MODE_VECTOR_INT): Likewise.
16801 * genoutput.c (null_operand): Likewise.
16802 (output_operand_data): Likewise.
16803 * genrecog.c (print_parameter_value): Likewise.
16804 * lra.c (debug_operand_data): Likewise.
16806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16808 * dwarf2out.c (scompare_loc_descriptor_wide)
16809 (scompare_loc_descriptor_narrow): New functions, split out from...
16810 (scompare_loc_descriptor): ...here.
16811 * expmed.c (emit_store_flag_int): New function, split out from...
16812 (emit_store_flag): ...here.
16814 2017-08-30 Richard Biener <rguenther@suse.de>
16816 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
16817 (dwarf2out_early_finish): Move setting of AT_pubnames from
16818 early debug output to early finish.
16820 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16822 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
16823 and -mdata-region to the assembler.
16825 2017-08-30 Richard Biener <rguenther@suse.de>
16827 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
16829 (gen_subprogram_die): Add DW_AT_object_pointer only early.
16830 (dwarf2out_early_global_decl): Only generate a DIE for the
16831 abstract origin if it doesn't already exist or is a declaration DIE.
16832 (resolve_addr): Do not add the linkage name twice when
16833 generating a stub DIE for the DW_TAG_GNU_call_site target.
16835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16837 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16838 Use machine_mode rather than int for arg1_mode.
16840 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
16843 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
16844 that the second argument of the built-in functions to unpack
16845 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
16846 switch statement instead a lot of if statements.
16847 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
16848 Allow 64-bit values to be in Altivec registers as well as
16849 traditional floating point registers.
16850 (pack<mode>, FMOVE128_VSX iterator): Likewise.
16852 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
16854 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
16855 MAX_REGS_PER_ADDRESS == 1.
16857 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
16859 * config/i386/i386.opt (flag_fentry): Do not init to -1.
16860 * config/i386/i386.c (ix86_option_override_internal): Simplify
16861 setting of opts->x_flag_entry.
16863 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16864 Jakub Jelinek <jakub@redhat.com>
16865 Richard Biener <rguenther@suse.de>
16867 PR tree-optimization/81503
16868 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
16869 folded constant fits in the target type; reorder tests for clarity.
16871 2017-08-29 Martin Liska <mliska@suse.cz>
16873 * passes.def: Include pass_lower_switch.
16874 * stmt.c (dump_case_nodes): Remove and move to
16875 tree-switch-conversion.
16876 (case_values_threshold): Likewise.
16877 (expand_switch_as_decision_tree_p): Likewise.
16878 (emit_case_decision_tree): Likewise.
16879 (expand_case): Likewise.
16880 (balance_case_nodes): Likewise.
16881 (node_has_low_bound): Likewise.
16882 (node_has_high_bound): Likewise.
16883 (node_is_bounded): Likewise.
16884 (emit_case_nodes): Likewise.
16885 (struct simple_case_node): New struct.
16886 (add_case_node): Remove.
16887 (emit_case_dispatch_table): Use vector instead of case_list.
16888 (reset_out_edges_aux): Remove.
16889 (compute_cases_per_edge): Likewise.
16890 (expand_case): Build list of simple_case_node.
16891 (expand_sjlj_dispatch_table): Use it.
16892 * tree-switch-conversion.c (struct case_node): Moved from
16893 stmt.c and adjusted.
16894 (emit_case_nodes): Likewise.
16895 (node_has_low_bound): Likewise.
16896 (node_has_high_bound): Likewise.
16897 (node_is_bounded): Likewise.
16898 (case_values_threshold): Likewise.
16899 (reset_out_edges_aux): Likewise.
16900 (compute_cases_per_edge): Likewise.
16901 (add_case_node): Likewise.
16902 (dump_case_nodes): Likewise.
16903 (balance_case_nodes): Likewise.
16904 (expand_switch_as_decision_tree_p): Likewise.
16905 (emit_jump): Likewise.
16906 (emit_case_decision_tree): Likewise.
16907 (try_switch_expansion): Likewise.
16908 (do_jump_if_equal): Likewise.
16909 (emit_cmp_and_jump_insns): Likewise.
16910 (fix_phi_operands_for_edge): New function.
16911 (record_phi_operand_mapping): Likewise.
16912 (class pass_lower_switch): New pass.
16913 (pass_lower_switch::execute): New function.
16914 (make_pass_lower_switch): Likewise.
16915 (conditional_probability):
16916 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16917 * tree-pass.h: Add make_pass_lower_switch.
16919 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16922 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16925 2017-08-29 Richard Biener <rguenther@suse.de>
16927 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16928 we do not add a DW_AT_inline attribute twice.
16929 (gen_subprogram_die): Remove code setting DW_AT_inline on
16930 DECL_ABSTRACT_P nodes.
16932 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16934 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16935 calls to internal functions.
16936 (gimplify_modify_expr): Likewise.
16937 * tree-call-cdce.c (use_internal_fn): Likewise.
16938 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16939 (convert_to_divmod): Set the nothrow flag.
16940 * tree-if-conv.c (predicate_mem_writes): Likewise.
16941 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16942 (vectorizable_call): Likewise.
16943 (vectorizable_store): Likewise.
16944 (vectorizable_load): Likewise.
16945 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16946 (vect_recog_mask_conversion_pattern): Likewise.
16948 2017-08-29 Martin Liska <mliska@suse.cz>
16951 * gcc.c (driver_handle_option): Add new argument.
16952 * opts-common.c (handle_option): Pass
16953 target_option_override_hook.
16954 * opts-global.c (lang_handle_option): Add new option.
16955 (set_default_handlers): Add new argument.
16956 (decode_options): Likewise.
16957 * opts.c (target_handle_option): Likewise.
16958 (common_handle_option): Call target_option_override_hook.
16959 * opts.h (struct cl_option_handler_func): Add hook for
16960 target option override.
16961 (struct cl_option_handlers): Likewise.
16962 (set_default_handlers): Add new argument.
16963 (decode_options): Likewise.
16964 (common_handle_option): Likewise.
16965 (target_handle_option): Likewise.
16966 * toplev.c (toplev::main): Pass targetm.target_option.override
16969 2017-08-29 Richard Biener <rguenther@suse.de>
16970 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16972 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16973 life to the active subtree.
16975 2017-08-28 Jeff Law <law@redhat.com>
16977 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16978 derive_equivalences.
16979 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16980 Code moved into....
16981 (edge_info::derive_equivalences): New private member function
16983 * tree-ssa-dom.c (class edge_info): Changed from a struct
16984 to a class. Add ctor/dtor, methods and data members.
16985 (edge_info::edge_info): Renamed from allocate_edge_info.
16986 Initialize additional members.
16987 (edge_info::~edge_info): New.
16988 (free_dom_edge_info): Delete the edge info.
16989 (record_edge_info): Use new class & associated member functions.
16990 Tighten forms for testing for edge equivalences.
16991 (record_temporary_equivalences): Iterate over the simple
16992 equivalences rather than assuming there's only one per edge.
16993 (cprop_into_successor_phis): Iterate over the simple
16994 equivalences rather than assuming there's only one per edge.
16995 (optimize_stmt): Use operand_equal_p rather than pointer
16996 equality for mini-DSE code.
16998 2017-08-28 Nathan Sidwell <nathan@acm.org>
17000 * gcc.c (execute): Fold SIGPIPE handling into switch
17001 statement. Adjust internal error message.
17003 2017-08-28 Richard Biener <rguenther@suse.de>
17006 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17007 Do nothing for removed DIEs.
17009 2017-08-28 Richard Biener <rguenther@suse.de>
17011 PR tree-optimization/81977
17012 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17015 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17018 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17019 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17020 using targetm.gen_mem_thread_fence.
17022 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17025 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17026 predicate to register_operand. Reorder operands.
17027 (*btr<mode>): Ditto.
17028 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17029 (*btr<mode>_mask): Ditto.
17031 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17033 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17034 * config/rs6000/xmmintrin.h: New file.
17035 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17037 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17040 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17041 parameter and_insn and return it.
17042 (recombine_lvx_pattern): Insert a copy to ensure availability of
17043 the base register of the copied masking operation at the point of
17044 the instruction replacement.
17045 (recombine_stvx_pattern): Likewise.
17047 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17049 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17050 undocumented switches.
17051 (-mpower9-dform-vector): Likewise.
17052 (-mpower9-dform): Likewise.
17053 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17054 comments to delete references to -mpower9-dform* switches.
17055 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17056 Delete reference to -mpower9-dform* switches, test for
17057 -mpower9-vector instead.
17058 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17059 (OTHER_P9_VECTOR_MASKS): Likewise.
17060 (POWERPC_MASKS): Likewise.
17061 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17062 tests against -mpower9-dform* to -mpower9-vector. Delete code
17063 that checked for -mpower9-dform* consistancy with other options.
17064 Add test for -mpower9-misc to enable other power9 switches.
17065 (rs6000_init_hard_regno_mode_ok): Likewise.
17066 (rs6000_option_override_internal): Likewise.
17067 (rs6000_emit_prologue): Likewise.
17068 (rs6000_emit_epilogue): Likewise.
17069 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17070 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17071 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17073 (emit_fusion_p9_store): Likewise.
17074 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17075 resetting these macros if the assembler does not support ISA 3.0
17077 (TARGET_P9_DFORM_VECTOR): Likewise.
17078 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17079 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17081 2017-08-25 Alan Modra <amodra@gmail.com>
17083 PR rtl-optimization/81747
17084 * cse.c (cse_extended_basic_block): Don't attempt to record
17085 equivalences for degenerate conditional branches that branch
17086 to their fall-through.
17088 2017-08-24 Martin Sebor <msebor@redhat.com>
17090 PR middle-end/81908
17091 * gimple-fold.c (size_must_be_zero_p): New function.
17092 (gimple_fold_builtin_memory_op): Call it.
17094 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17096 * config/rs6000/mm_malloc.h: New file.
17098 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17100 PR tree-optimization/81913
17101 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17102 analysis when either IVs in condition can wrap.
17104 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17106 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17107 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17109 2017-08-24 Richard Biener <rguenther@suse.de>
17112 * targhooks.c (default_target_can_inline_p): Properly
17113 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17114 is present and always compare.
17115 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17116 infer -mfpmath=sse from TARGET_SSE_P.
17117 (ix86_can_inline_p): Properly use target_option_default_node when
17118 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17120 2017-08-24 Richard Biener <rguenther@suse.de>
17123 * dwarf2out.c (output_die): Handle flag_generate_offload like
17125 (output_comp_unit): Likewise.
17126 (gen_array_type_die): Likewise.
17127 (dwarf2out_early_finish): Likewise.
17128 (note_variable_value_in_expr): Likewise.
17129 (dwarf2out_finish): Likewise. Adjust assert.
17130 * cgraphunit.c (symbol_table::compile): Move setting of
17131 flag_generate_offload earlier ...
17132 (symbol_table::finalize_compilation_unit): ... here, before
17133 early debug finalization.
17135 2017-08-24 Richard Biener <rguenther@suse.de>
17137 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17138 and ipa-fnsummary.h.
17139 (ix86_can_inline_p): When ix86_fpmath flags do not match
17140 check whether the callee uses FP math at all.
17142 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17144 PR middle-end/81931
17145 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17146 instead of TYPE_PRECISION.
17148 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17149 Alan Hayward <alan.hayward@arm.com>
17150 David Sherwood <david.sherwood@arm.com>
17152 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17154 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17156 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17158 (ix86_frame::stack_realign_allocate): New field.
17159 (struct machine_frame_state): Modify comments.
17160 (machine_frame_state::sp_realigned_fp_end): New field.
17161 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17162 layout calculation.
17163 (sp_valid_at): Add assertion to assure no attempt to access invalid
17164 offset of a realigned stack.
17165 (fp_valid_at): Likewise.
17166 (choose_baseaddr): Modify comments.
17167 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17168 ix86_expand_prologue.
17169 (ix86_expand_prologue): Modify stack realignment and allocation.
17170 (ix86_expand_epilogue): Modify comments.
17171 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17172 avx2_runtime, avx512f, and avx512f_runtime.
17174 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17176 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17177 (mstackrealign): Do not init to -1.
17178 * config/i386/i386.c (ix86_option_override_internal):
17179 Check opts_set, not opts when setting default value of
17180 opts->x_ix86_force_align_arg_pointer.
17182 2017-08-23 Richard Biener <rguenther@suse.de>
17184 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17185 lang_hooks.decl_printable_name.
17186 * print-rtl-function.c (print_rtx_function): Likewise.
17187 * tree-pretty-print.c (dump_function_header): Likewise.
17189 2017-08-23 Richard Biener <rguenther@suse.de>
17192 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17193 -g0 at compile-time.
17195 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17197 PR middle-end/19706
17198 * doc/sourcebuild.texi (Other hardware attributes):
17201 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17203 PR middle-end/19706
17204 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17205 Fix single-use check.
17207 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17209 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17211 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17213 * doc/install.texi: Modify to add more details on running selected
17216 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17218 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17219 is combined with -mabi=ms.
17220 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17223 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17225 PR tree-optimization/81488
17226 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17227 and cached_basis fields.
17228 (MAX_SPREAD): New constant.
17229 (alloc_cand_and_find_basis): Initialize new fields.
17230 (clear_visited): New function.
17231 (create_phi_basis_1): Rename from create_phi_basis, set visited
17232 and cached_basis fields.
17233 (create_phi_basis): New wrapper function.
17234 (phi_add_costs_1): Rename from phi_add_costs, add spread
17235 parameter, set visited field, short-circuit when limits reached.
17236 (phi_add_costs): New wrapper function.
17237 (record_phi_increments_1): Rename from record_phi_increments, set
17239 (record_phi_increments): New wrapper function.
17240 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17241 (phi_incr_cost): New wrapper function.
17242 (all_phi_incrs_profitable_1): Rename from
17243 all_phi_incrs_profitable, set visited field.
17244 (all_phi_incrs_profitable): New wrapper function.
17246 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17247 Alan Hayward <alan.hayward@arm.com>
17248 David Sherwood <david.sherwood@arm.com>
17250 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
17251 that takes the outer and inner modes.
17252 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
17253 comparison as the canonical test for a paradoxical subreg.
17254 * combine.c (simplify_set): Use paradoxical_subreg_p.
17255 (make_extraction): Likewise.
17256 (force_to_mode): Likewise.
17257 (rtx_equal_for_field_assignment_p): Likewise.
17258 (gen_lowpart_for_combine): Likewise.
17259 (simplify_comparison): Likewise.
17260 * cse.c (equiv_constant): Likewise.
17261 * expmed.c (store_bit_field_1): Likewise.
17262 * final.c (alter_subreg): Likewise.
17263 * fwprop.c (propagate_rtx): Likewise.
17264 (forward_propagate_subreg): Likewise.
17265 * ira-conflicts.c (ira_build_conflicts): Likewise.
17266 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
17267 * lra-constraints.c (curr_insn_transform): Likewise.
17268 (split_reg): Likewise.
17269 * lra-eliminations.c (move_plus_up): Likewise.
17270 (lra_eliminate_regs_1): Likewise.
17271 * recog.c (general_operand): Likewise.
17272 * ree.c (combine_reaching_defs): Likewise.
17273 * reload.c (push_reload): Likewise.
17274 (find_reloads): Likewise.
17275 * reload1.c (elimination_effects): Likewise.
17276 (compute_reload_subreg_offset): Likewise.
17277 (choose_reload_regs): Likewise.
17278 * rtlanal.c (subreg_lsb_1): Likewise.
17279 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17280 (simplify_subreg): Likewise.
17281 * var-tracking.c (track_loc_p): Likewise.
17282 * emit-rtl.c (byte_lowpart_offset): Likewise.
17283 (paradoxical_subreg_p): Delete out-of-line definition.
17285 2017-08-22 Jeff Law <law@redhat.com>
17287 PR tree-optimization/81741
17288 PR tree-optimization/71947
17289 * tree-ssa-dom.c: Include tree-inline.h.
17290 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
17291 equivalences if one is more expensive to compute than the other.
17292 * tree-ssa-scopedtables.h (class const_or_copies): Make
17293 record_const_or_copy_raw method private.
17294 (class avail_exprs_stack): New method simplify_binary_operation.
17295 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
17296 avail_exprs_stack::simplify_binary_operation as needed.
17297 (avail_exprs_stack::simplify_binary_operation): New function.
17299 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
17301 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
17302 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
17303 (DOT_SYMBOLS): Likewise.
17304 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
17305 (RELOCATABLE_NEEDS_FIXUP): Likewise.
17306 (RS6000_ABI_NAME): Likewise.
17307 (TARGET_CMODEL): Likewise.
17308 (TOC_SECTION_ASM_OP): Likewise.
17309 (SET_CMODEL): New macro.
17310 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
17312 2017-08-22 Richard Biener <rguenther@suse.de>
17314 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
17315 to allow for free-lang-data replacements similar to verify_type_variant.
17317 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17318 Alan Hayward <alan.hayward@arm.com>
17319 David Sherwood <david.sherwood@arm.com>
17321 * config/aarch64/aarch64.md (casesi): Use DImode rather than
17322 VOIDmode for the LABEL_REF.
17324 2017-08-22 Richard Biener <rguenther@suse.de>
17326 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
17328 2017-08-22 Richard Biener <rguenther@suse.de>
17330 * common.opt (feliminate-dwarf2-dups): Ignore.
17331 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
17332 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
17333 same_die_p_wrap, compute_section_prefix,
17334 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
17335 (comdat_symbol_id, comdat_symbol_number): Likewise.
17336 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
17338 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
17339 (output_die): Mark unreachable path unreachable.
17340 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
17341 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
17342 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
17343 (dwarf2out_early_finish): Likewise.
17345 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
17347 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
17349 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
17352 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
17353 not VAR_P. Filter attribute warnings with OPT_Wattributes.
17354 (avr_attribute_table) <io, io_low, address>: Initialize
17355 .decl_required with true.
17357 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
17359 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
17360 undocumented debugging options.
17361 (-mvsx-scalar-double): Likewise.
17362 (-mallow-df-permute): Likewise.
17363 (-mvectorize-builtins): Likewise.
17364 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17365 (rs6000_builtin_vectorized_function): Likewise.
17366 (rs6000_builtin_md_vectorized_function): Likewise.
17367 (rs6000_opt_vars): Likewise.
17369 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
17372 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
17373 (*btrq_imm): Rename from *btrq.
17374 (*btcq_imm): Rename from *btcq.
17375 (btsc): New code attribute.
17376 (*<btsc><mode>): New insn pattern.
17377 (*btr<mode>): Ditto.
17378 (*<btsc><mode>_mask): New insn_and_split pattern.
17379 (*btr<mode>_mask): Ditto.
17381 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17382 Alan Hayward <alan.hayward@arm.com>
17383 David Sherwood <david.sherwood@arm.com>
17385 * function.c (pad_below): Simplify padding calculation.
17387 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17388 Alan Hayward <alan.hayward@arm.com>
17389 David Sherwood <david.sherwood@arm.com>
17391 * target.def (function_prologue): Remove frame size argument.
17392 (function_epilogue): Likewise.
17393 * doc/tm.texi: Regenerate.
17394 * final.c (final_start_function): Update call to function_prologue.
17395 (final_end_function): Update call to function_epilogue.
17396 (default_function_pro_epilogue): Remove frame size argument.
17397 * output.h (default_function_pro_epilogue): Likewise.
17398 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
17399 (arm_output_function_prologue): Likewise.
17400 * config/frv/frv.c (frv_function_prologue): Likewise.
17401 (frv_function_epilogue): Likewise.
17402 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
17403 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
17404 (ia64_output_function_epilogue): Likewise.
17405 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
17406 (m32r_output_function_epilogue): Likewise.
17407 * config/microblaze/microblaze.c (microblaze_function_prologue)
17408 (microblaze_function_epilogue): Likewise.
17409 * config/mips/mips.c (mips_output_function_prologue): Likewise.
17410 (mips_output_function_epilogue): Likewise.
17411 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
17412 (mmix_target_asm_function_epilogue): Likewise.
17413 * config/msp430/msp430.c (msp430_start_function): Likewise.
17414 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
17415 (nds32_asm_function_epilogue): Likewise.
17416 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
17417 * config/pa/pa.c (pa_output_function_prologue): Likewise.
17418 (pa_output_function_epilogue): Likewise.
17419 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
17420 (rs6000_output_function_epilogue): Likewise.
17421 * config/rl78/rl78.c (rl78_start_function): Likewise.
17422 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
17423 (rs6000_output_function_epilogue): Likewise.
17424 * config/rx/rx.c (rx_output_function_prologue): Likewise.
17425 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
17426 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
17427 (sparc_asm_function_epilogue): Likewise.
17429 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17431 * tree.h (type_has_mode_precision_p): New function.
17432 * convert.c (convert_to_integer_1): Use it.
17433 * expr.c (expand_expr_real_2): Likewise.
17434 (expand_expr_real_1): Likewise.
17435 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
17436 * match.pd: Likewise.
17437 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
17438 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
17439 * tree-tailcall.c (process_assignment): Likewise.
17440 * tree-vect-loop.c (vectorizable_reduction): Likewise.
17441 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
17442 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
17443 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17444 (vectorizable_assignment): Likewise.
17445 (vectorizable_shift): Likewise.
17446 (vectorizable_operation): Likewise.
17447 * tree-vrp.c (register_edge_assert_for_2): Likewise.
17449 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
17451 * match.pd: Add pow (C, x) simplification.
17453 2017-08-21 Richard Biener <rguenther@suse.de>
17455 PR tree-optimization/81900
17456 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
17457 for blocks with abnormal predecessors.
17458 (compute_antic): Do not set visited flag prematurely.
17460 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
17463 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
17465 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17467 * stor-layout.h (vector_type_mode): Move to...
17468 * tree.h (vector_type_mode): ...here.
17469 * stor-layout.c (vector_type_mode): Move to...
17470 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
17472 2017-08-21 Richard Biener <rguenther@suse.de>
17474 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
17475 register_external_die hooks.
17476 (debug_false_tree_charstarstar_uhwistar): Declare.
17477 (debug_nothing_tree_charstar_uhwi): Likewise.
17478 * debug.c (do_nothing_debug_hooks): Adjust.
17479 (debug_false_tree_charstarstar_uhwistar): New do nothing.
17480 (debug_nothing_tree_charstar_uhwi): Likewise.
17481 * dbxout.c (dbx_debug_hooks): Adjust.
17482 (xcoff_debug_hooks): Likewise.
17483 * sdbout.c (sdb_debug_hooks): Likewise.
17484 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
17485 * dwarf2out.c (macinfo_label_base): New global.
17486 (dwarf2out_register_external_die): New function for the
17487 register_external_die hook.
17488 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
17489 (dwarf2_debug_hooks): Use them.
17490 (dwarf2_lineno_debug_hooks): Adjust.
17491 (struct die_struct): Add with_offset flag.
17492 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
17493 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
17494 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
17495 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
17496 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
17497 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
17498 defining section names for the early LTO debug variants.
17499 (reset_indirect_string): New helper.
17500 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
17501 (print_dw_val): Add support for offsetted symbol references.
17502 (get_ultimate_context): Split out from is_cxx.
17503 (is_cxx): Use get_ultimate_context.
17504 (is_fortran): Add decl overload.
17505 (compute_comp_unit_symbol): Split out worker from
17506 compute_section_prefix.
17507 (compute_section_prefix): Call compute_comp_unit_symbol and
17508 set comdat_type_p here.
17509 (output_die): Skip DIE symbol output for the LTO added one.
17510 Handle DIE symbol references with offset.
17511 (output_comp_unit): Guard section name mangling properly.
17512 For LTO debug sections emit a symbol at the section beginning
17513 which we use to refer to its DIEs.
17514 (add_abstract_origin_attribute): For DIEs registered via
17515 dwarf2out_register_external_die directly refer to the early
17516 DIE rather than indirectly through the shadow one we created.
17517 Remove obsolete call to dwarf2out_abstract_function for
17518 non-function/block origins.
17519 (gen_array_type_die): When generating early LTO debug do
17520 not emit DW_AT_string_length.
17521 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
17522 late when in LTO. As suggested place a gcc_unreachable for
17523 the DECL_ABSTRACT_P case.
17524 (gen_subprogram_die): Avoid another specification DIE
17525 for early built declarations/definitions for the late LTO case.
17526 (gen_variable_die): Add type references for late duplicated VLA dies
17528 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
17529 we have the abstract instance already.
17530 (process_scope_var): Adjust decl DIE contexts in LTO which
17531 first puts them in limbo.
17532 (gen_decl_die): Do not generate type DIEs late apart from
17533 types for VLAs or for decls we do not yet have a DIE. Do not
17534 call dwarf2out_abstract_function late.
17535 (dwarf2out_early_global_decl): Make sure to create DIEs
17536 for abstract instances of a decl first.
17537 (dwarf2out_late_global_decl): Adjust comment.
17538 (output_macinfo_op): With multiple macro sections use
17539 macinfo_label_base to distinguish labels.
17540 (output_macinfo): Likewise. Update macinfo_label_base.
17541 Pass in the line info label.
17542 (note_variable_value_in_expr): When generating LTO resolve
17543 all variable values here by generating DIEs as needed.
17544 (init_sections_and_labels): Add early LTO debug flag parameter
17545 and generate different sections and names if set. Add generation
17546 counter for the labels so we can have multiple of them.
17547 (reset_dies): Helper to allow DIEs to be output multiple times.
17548 (dwarf2out_finish): When outputting DIEs to the fat part of an
17549 LTO object first reset DIEs.
17550 (dwarf2out_early_finish): Output early DIEs when generating LTO.
17551 (modified_type_die): Check for decl_ultimate_origin being self
17553 (gen_type_die_with_usage): Likewise.
17554 (gen_typedef_die): Allow decl_ultimate_origin being self.
17555 (set_decl_abstract_flags): Remove.
17556 (set_block_abstract_flags): Likewise.
17557 (dwarf2out_abstract_function): Treat the early generated DIEs
17558 as the abstract copy and only add DW_AT_inline and
17559 DW_AT_artificial here and call set_decl_origin_self.
17560 If the DIE has an abstract origin don't do anything.
17561 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
17562 if we have none yet (Go fails to build one, PR78628).
17563 (variably_modified_type_p): Prevent endless recursion for Ada
17564 cyclic pointer types.
17565 * lto-streamer-in.c: Include debug.h.
17566 (dref_queue): New global.
17567 (lto_read_tree_1): Stream in DIE references.
17568 (lto_input_tree): Register DIE references.
17569 (input_function): Stream DECL_DEBUG_ARGS.
17570 * lto-streamer-out.c: Include debug.h.
17571 (lto_write_tree_1): Output DIE references.
17572 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
17573 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
17574 (output_function): Stream DECL_DEBUG_ARGS.
17575 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17576 Stream DECL_ABSTRACT_ORIGIN.
17577 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17578 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
17579 DECL_CONTEXT for file-scope decls.
17580 * lto-streamer.h (struct dref_entry): Declare.
17581 (dref_queue): Likewise.
17582 * cfgexpand.c (pass_expand::execute): Do not call the
17583 outlining_inline_function hook here.
17584 * lto-wrapper.c (debug_obj): New global.
17585 (tool_cleanup): Unlink it if required.
17586 (debug_objcopy): New function.
17587 (run_gcc): Handle early debug sections in the IL files by
17588 extracting them to separate files, partially linkin them and
17589 feeding the result back as result to the linker.
17590 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
17591 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
17592 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
17593 sections into a separate segment.
17594 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
17596 (darwin_asm_dwarf_section): Likewise.
17597 (darwin_asm_output_dwarf_offset): Likewise.
17598 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
17600 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17601 Alan Hayward <alan.hayward@arm.com>
17602 David Sherwood <david.sherwood@arm.com>
17604 * read-md.h (md_reader::record_potential_iterator_use): Replace
17605 pointer argument with an rtx and an index.
17606 * read-rtl.c (iterator_group::apply_iterator): Likewise.
17607 (apply_mode_iterator): Likewise.
17608 (apply_code_iterator): Likewise.
17609 (apply_int_iterator): Likewise.
17610 (apply_subst_iterator): Likewise.
17611 (record_iterator_use): Likewise.
17612 (record_attribute_use): Likewise.
17613 (md_reader::record_potential_iterator_use): Likewise. Update calls
17614 to record_iterator_use and apply_iterator.
17615 (iterator_use): Replace ptr with x and index.
17616 (attribute_use): Likewise.
17617 (apply_attribute_uses): Update calls to apply_iterator.
17618 (apply_iterators): Likewise. Update initialization of iterator_use.
17619 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
17620 and record_potential_iterator_use.
17621 (rtx_reader::read_rtx_operand): Likewise.
17623 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
17624 Alan Hayward <alan.hayward@arm.com>
17625 David Sherwood <david.sherwood@arm.com>
17627 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
17630 2017-08-21 Richard Biener <rguenther@suse.de>
17632 PR middle-end/81884
17633 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
17634 at struct end conservatively when comparing common bases.
17636 2017-08-21 Richard Biener <rguenther@suse.de>
17638 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
17639 (mem_ref_in_stmt): Remove.
17640 (determine_max_movement): Use ref index to get at the reference.
17641 (invariantness_dom_walker::before_dom_children): Deal with
17642 lim data already initialized.
17643 (gather_mem_refs_stmt): Initialize lim data and record ref index.
17645 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17647 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
17648 (TARGET_ISA_ROUND): Ditto.
17649 (TARGET_ROUND): Ditto.
17650 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
17651 * config/i386/i386.md: Ditto.
17652 * config/i386/sse.md: Ditto.
17653 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
17654 with OPTION_MASK_ISA_SSE4_1.
17656 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
17659 * doc/extend.texi (x86 Built-in Functions): Correct the name of
17660 __builtin_ia32_lzcnt_u16.
17662 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
17665 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
17666 (rs6000_set_current_function): Rewrite function to use it.
17668 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
17671 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
17672 and TYPE_WARN_IF_NOT_ALIGN.
17673 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
17674 (handle_warn_if_not_align): New.
17675 (place_union_field): Call handle_warn_if_not_align.
17676 (place_field): Call handle_warn_if_not_align.
17677 Copy TYPE_WARN_IF_NOT_ALIGN.
17678 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
17679 (layout_type): Likewise.
17680 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
17682 (tree_decl_common): Add warn_if_not_align.
17683 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
17684 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
17685 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
17686 (DECL_WARN_IF_NOT_ALIGN): Likewise.
17687 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
17688 * doc/extend.texi: Document warn_if_not_aligned attribute.
17689 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
17691 2017-08-17 Martin Liska <mliska@suse.cz>
17694 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
17695 (get_data_dependence): Use it as pointer type.
17696 (distribute_loop): Likewise.
17698 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17700 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
17701 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
17702 (p8_vmrgow_<mode>_direct): New define_insn.
17703 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
17704 handle endianness for vmrgew and vmrgow permute patterns.
17706 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17708 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
17709 * config/rs6000/rs6000-cpus.def: Remove comment.
17710 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
17711 (POWERPC_MASKS): Likewise.
17712 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
17713 use of TARGET_VSX_TIMODE.
17714 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
17715 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
17716 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
17717 (rs6000_option_override_internal): Remove dead code.
17718 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
17719 (rs6000_legitimize_reload_address): Likewise.
17720 (rs6000_legitimate_address_p): Likewise.
17721 (rs6000_opt_masks): Delete "vsx-timode".
17722 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
17723 from function comment.
17724 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
17725 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
17726 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
17728 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
17729 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
17730 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
17731 (VSX_TI): Likewise.
17733 (define_peephole2): Likewise.
17735 2017-08-17 Martin Sebor <msebor@redhat.com>
17738 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
17739 past the end of an array.
17740 (test_pp_format): Add test cases.
17742 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
17744 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
17745 missing ECF_NOTHROW flags.
17747 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
17750 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
17751 operands residing in integer registers.
17752 (*vsx_le_perm_load_<mode>): Likewise.
17753 (*vsx_le_perm_store_<mode>): Likewise.
17754 (define_peephole2): Add peepholes to optimize the above.
17756 2017-08-17 Marek Polacek <polacek@redhat.com>
17758 PR middle-end/81814
17759 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
17760 to mimic what shorten_compare did. Change the return type to bool.
17761 (fold_cond_expr_with_comparison): Update call to
17762 operand_equal_for_comparison_p.
17763 (fold_ternary_loc): Likewise.
17765 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
17767 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
17769 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
17771 2017-08-17 Richard Biener <rguenther@suse.de>
17773 * tree-ssa-structalias.c (solve_graph): When propagating
17774 to successors update the graphs succ edges and avoid duplicate work.
17776 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
17779 * config/i386/i386.c (ix86_option_override_internal): Save target
17780 specific options after ix86_stack_protector_guard_reg was changed.
17782 2017-08-17 Richard Biener <rguenther@suse.de>
17784 PR tree-optimization/81827
17785 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
17786 (new_var_info): Initialize it conservatively.
17787 (get_call_vi): Mark register vars.
17788 (new_scalar_tmp_constraint_exp): Likewise.
17789 (handle_rhs_call): Likewise.
17790 (handle_const_call): Likewise.
17791 (create_function_info_for): Likewise.
17792 (solve_constraints): Sort varinfos to separate register from
17793 non-register vars to pack points-to solution bitmaps during
17796 2017-08-17 Marek Polacek <polacek@redhat.com>
17798 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
17800 2017-08-17 Richard Biener <rguenther@suse.de>
17802 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
17803 to true when overflow is undefined and we saturated the result.
17805 2017-08-17 Alan Modra <amodra@gmail.com>
17808 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
17809 Don't use store multiple if only one reg needs saving.
17810 (interesting_frame_related_regno): New function.
17811 (rs6000_frame_related): Don't emit frame info for regs that
17813 (rs6000_emit_epilogue): Likewise.
17815 2017-08-16 Nathan Sidwell <nathan@acm.org>
17817 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
17818 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
17819 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
17820 * tree.c (free_lang_data_in_type): Use else-if chain. Always
17821 clear TYPE_LANG_1. Remove obsolete member-function stripping.
17822 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
17823 (verify_type): Adjust for TYPE_BINFO move.
17824 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
17825 process TYPE_BINFO directly.
17826 (hash_tree): Likewise.
17827 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17829 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17832 2017-08-16 David Malcolm <dmalcolm@redhat.com>
17834 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
17836 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
17839 * config/i386/i386.md (*anddi_1_btr): Change predicates of
17840 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
17841 Add ix86_binary_operator_ok to insn constraint.
17842 (*iordi_1_bts): Ditto.
17843 (*xordi_1_btc): Ditto.
17844 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
17845 Update corresponding peephole2 pattern.
17849 2017-08-16 Bin Cheng <bin.cheng@arm.com>
17851 PR tree-optimization/81832
17852 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
17853 copy loop header which has IFN_LOOP_DIST_ALIAS call.
17855 2017-08-16 Marek Polacek <polacek@redhat.com>
17858 * fold-const.c (fold_indirect_ref_1): Restore original behavior
17859 regarding size_zero_node.
17861 2017-08-16 Martin Liska <mliska@suse.cz>
17864 * config.gcc: Respect previously set extra_objs in case
17867 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
17869 PR tree-optimization/81835
17870 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
17871 the phi SSA_NAME. Check that the condition in a COND_EXPR does
17872 not depend on the phi.
17874 2017-08-16 Alan Modra <amodra@gmail.com>
17876 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
17879 2017-08-16 Alan Modra <amodra@gmail.com>
17881 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
17882 (save_reg_p): ..into this. Update all callers.
17883 (first_reg_to_save): Simplify.
17885 2017-08-16 Alan Modra <amodra@gmail.com>
17887 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17890 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17894 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17895 constant count if that count is less than 32.
17897 2017-08-15 Nathan Sidwell <nathan@acm.org>
17899 * gcc.c (execute): Emit friendlier message if inferior is killed
17900 by an external cause.
17902 2017-08-15 Richard Biener <rguenther@suse.de>
17904 PR tree-optimization/81790
17905 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17906 CONSTRUCTORs from simplifying and VN.
17908 2017-08-14 Martin Sebor <msebor@redhat.com>
17910 * builtin-attrs.def: Add comments.
17912 2017-08-14 Martin Sebor <msebor@redhat.com>
17915 * doc/extend.texi (attribute nonstring): Document new attribute.
17917 2017-08-14 Martin Sebor <msebor@redhat.com>
17920 * tree-diagnostic.c (default_tree_printer): Handle %G.
17921 * gimple-pretty-print.h (percent_G_format): Declare new function.
17922 * gimple-pretty-print.c (percent_G_format): Define.
17923 * tree-pretty-print.c (percent_K_format): Add argument.
17925 2017-08-14 Martin Sebor <msebor@redhat.com>
17927 PR translation/79998
17928 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17929 Remove a stray space.
17931 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17934 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17935 (*iordi_1_bts): Ditto.
17936 (*xordi_1_btc): Ditto.
17938 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17941 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17942 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17944 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17945 quoted strings, and make more translator-friendly.
17946 (darwin_rs6000_override_options): Likewise.
17947 (rs6000_option_override_internal): Likewise.
17948 (rs6000_return_in_memory): Fix overlong line.
17949 (init_cmulative_args): Use quoted strings, and make more
17950 translator-friendly.
17951 (rs6000_pass_by_reference): Fix overlong line.
17952 (def_builtin): Use quoted strings.
17953 (altivec_expand_predicate_builtin): Use quoted strings, and make
17954 more translator-friendly.
17955 (htm_expand_builtin): Use quoted strings.
17956 (cpu_expand_builtin): Use quoted strings, and make more
17957 translator-friendly.
17958 (altivec_expand_builtin): Likewise.
17959 (paired_expand_predicate_builtin): Likewise.
17960 (rs6000_invalid_builtin): Likewise.
17961 (builtin_function_type): Use quoted strings.
17962 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17963 more translator-friendly.
17964 (rs6000_trampoline_init): Likewise.
17965 (rs6000_handle_altivec_attribute): Likewise.
17966 (rs6000_inner_target_options): Use quoted strings.
17967 (rs6000_disable_incompatible_switches): Likewise.
17968 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17969 strings, and make more translator-friendly.
17970 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17972 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17974 PR tree-optimization/81799
17975 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17976 cond_expr to simple gimple operand.
17978 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17980 PR middle-end/46932
17981 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17983 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17987 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17988 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17989 TARGET_GASISR_PROLOGUES.
17990 * config/avr/avr.c (avr_option_override): Same.
17991 (avr_pass_pre_proep::execute): Same.
17993 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17996 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17997 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17999 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18001 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18002 define_insn_and_split. Split to a memory load from 0 in
18003 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18004 using PTR mode iterator.
18005 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18006 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18007 (*add_tp_<mode>): Redefine as define_insn_and_split.
18008 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18009 address space. Merge with *add_tp_x32 using PTR mode iterator.
18010 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18011 Split to an add with a memory load from 0 in
18012 DEFAULT_TLS_SEG_REG address space.
18014 2017-08-12 Andrew Pinski <apinski@cavium.com>
18016 * config/aarch64/aarch64-option-extensions.def (rdma):
18017 Fix feature string to what Linux prints out in /proc/cpuinfo.
18019 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18022 * dwarf2out.c (modified_type_die): For C typedef types that have
18023 an ultimate origin, process the ultimate origin instead of the
18025 (gen_typedef_die): Assert that input DECLs have no ultimate
18027 (gen_type_die_with_usage): For typedef variants that have an
18028 ultimate origin, just call gen_decl_die on the original DECL.
18029 (process_scope_var): Avoid creating DIEs for local typedefs and
18030 concrete static variables.
18032 2017-08-12 Alan Modra <amodra@gmail.com>
18036 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18037 match gnu-user.h startfile.
18038 (ENDFILE_LINUX_SPEC): Similarly.
18040 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18043 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18045 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18047 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18048 * config/aarch64/aarch64.md (mov<mode>): Change.
18049 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18050 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18051 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18053 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18055 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18056 for storage order barriers.
18058 2017-08-11 Martin Liska <mliska@suse.cz>
18061 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18062 variables of void type.
18064 2017-08-11 Martin Liska <mliska@suse.cz>
18066 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18067 TARGET_SUPPORTS_ALIASES.
18068 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18069 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18070 (optimize_weakref): Likewise.
18071 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18072 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18073 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18075 2017-08-11 Martin Liska <mliska@suse.cz>
18078 * config/i386/i386.c (make_resolver_func): Do complete
18079 refactoring of the function.
18081 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18084 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18085 * config/i386/i386.c (ix86_stack_protect_guard): Use
18086 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18087 * doc/invoke.texi (x86 Options): Document
18088 -mstack-protector-guard-symbol= option.
18090 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18092 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18093 * config/i386/i386.c (ix86_split_stack_guard): New function.
18094 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18095 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18096 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18097 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18098 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18099 (split_stack_space_check): Call ix86_split_stack_guard.
18101 2017-08-10 Martin Sebor <msebor@redhat.com>
18103 * print-tree.c (print_node): Print location using the established
18105 Replace spaces with colons.
18106 (debug_raw, debug): Ditto.
18108 2017-08-10 Martin Sebor <msebor@redhat.com>
18111 * pretty-print.c (pp_format): Correct the handling of %s precision.
18113 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18116 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18118 (ix86_finalize_stack_frame_flags): This. Also clear
18119 frame_pointer_needed if -fno-omit-frame-pointer is used without
18121 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18122 with ix86_finalize_stack_frame_flags.
18123 (ix86_expand_epilogue): Likewise.
18124 (ix86_expand_split_stack_prologue): Likewise.
18125 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18127 2017-08-10 Martin Liska <mliska@suse.cz>
18130 * c-attribs.c (handle_target_attribute):
18131 Report warning for an empty string argument of target attribute.
18133 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18136 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18138 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18139 or DECL_NONLOCAL labels.
18140 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18141 or DECL_NONLOCAL labels here.
18143 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18145 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18146 to indicate when early gimple folding has been disabled.
18147 (rs6000_gimple_fold_builtin): Add debug content.
18148 (rs6000_invalid_builtin): Fix whitespace.
18149 (rs6000_expand_builtin): Fix whitespace.
18150 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18152 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18155 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18156 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18158 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18160 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18161 (qdf24xx): Likewise.
18162 * config/aarch64/aarch64.md: Include falkor.md.
18163 * config/aarch64/falkor.md: New.
18165 2017-08-09 Marek Polacek <polacek@redhat.com>
18168 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18169 * diagnostic.c (emit_diagnostic): Add a comment.
18170 (emit_diagnostic_valist): New function.
18172 2017-08-09 Marek Polacek <polacek@redhat.com>
18175 * input.c (make_location): New overload.
18176 * input.h (make_location): Declare.
18178 2017-08-08 Alan Modra <amodra@gmail.com>
18179 H.J. Lu <hongjiu.lu@intel.com>
18182 * gcc.c (NO_PIE_SPEC): Delete.
18183 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18185 (LINK_PIE_SPEC): ..to here.
18186 (LINK_COMMAND_SPEC): Support -no-pie.
18187 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18188 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18190 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18191 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18192 (ENDFILE_CRTEND_SPEC): Similarly.
18194 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18197 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18198 (mstack-protector-guard-offset=): Ditto.
18199 * config/i386/i386.c (ix86_option_override): Handle
18200 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18202 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18203 ix86_stack_protect_guard_offset variables.
18204 (TARGET_STACK_PROTECT_GUARD): Always define.
18205 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18206 and -mstack-protector-guard-offset= options.
18208 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18210 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18211 boundary case for the last candidate.
18213 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18215 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18216 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18218 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18220 PR middle-end/19706
18221 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18222 * config/aarch64/aarch64-builtins.c
18223 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18224 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18225 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18227 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18228 Andrew Pinski <pinskia@gmail.com>
18230 PR middle-end/19706
18231 * internal-fn.def (XORSIGN): New.
18232 * optabs.def (xorsign_optab): New.
18233 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18234 (convert_expand_mult_copysign): New.
18235 (pass_optimize_widening_mul::execute): Call
18236 convert_expand_mult_copysign.
18238 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18240 PR tree-optimization/81354
18241 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18242 Insert on edges rather than explicitly creating landing pads.
18243 (analyze_candidates_and_replace): Commit edge inserts.
18245 2017-08-08 Richard Biener <rguenther@suse.de>
18247 PR middle-end/81719
18248 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18249 (expand_simple_operations): Also look through ADDR_EXPRs with
18250 MEM_REF bases treating them as POINTER_PLUS_EXPR.
18252 2017-08-08 Richard Biener <rguenther@suse.de>
18254 PR tree-optimization/81723
18255 * tree-vect-slp.c (struct bst_traits): New hash traits.
18256 (bst_fail): New global.
18257 (vect_build_slp_tree_2): New worker, split out from ...
18258 (vect_build_slp_tree): ... this now wrapping it with using
18259 bst_fail set to cache SLP tree build fails. Properly handle
18261 (vect_analyze_slp_instance): Allocate and free bst_fail.
18263 2017-08-08 Martin Liska <mliska@suse.cz>
18266 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
18267 LABEL_DECLs that can be from a different function.
18269 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18271 PR tree-optimization/81744
18272 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
18273 loop's number of iterations.
18275 2017-08-08 Martin Liska <mliska@suse.cz>
18277 * asan.c: Include header files.
18278 * attribs.c (build_decl_attribute_variant): New function moved
18280 (build_type_attribute_qual_variant): Likewise.
18281 (cmp_attrib_identifiers): Likewise.
18282 (simple_cst_list_equal): Likewise.
18283 (omp_declare_simd_clauses_equal): Likewise.
18284 (attribute_value_equal): Likewise.
18285 (comp_type_attributes): Likewise.
18286 (build_type_attribute_variant): Likewise.
18287 (lookup_ident_attribute): Likewise.
18288 (remove_attribute): Likewise.
18289 (merge_attributes): Likewise.
18290 (merge_type_attributes): Likewise.
18291 (merge_decl_attributes): Likewise.
18292 (merge_dllimport_decl_attributes): Likewise.
18293 (handle_dll_attribute): Likewise.
18294 (attribute_list_equal): Likewise.
18295 (attribute_list_contained): Likewise.
18296 * attribs.h (lookup_attribute): New function moved from tree.[ch].
18297 (lookup_attribute_by_prefix): Likewise.
18298 * bb-reorder.c: Include header files.
18299 * builtins.c: Likewise.
18300 * calls.c: Likewise.
18301 * cfgexpand.c: Likewise.
18302 * cgraph.c: Likewise.
18303 * cgraphunit.c: Likewise.
18304 * convert.c: Likewise.
18305 * dwarf2out.c: Likewise.
18306 * final.c: Likewise.
18307 * fold-const.c: Likewise.
18308 * function.c: Likewise.
18309 * gimple-expr.c: Likewise.
18310 * gimple-fold.c: Likewise.
18311 * gimple-pretty-print.c: Likewise.
18312 * gimple.c: Likewise.
18313 * gimplify.c: Likewise.
18314 * hsa-common.c: Likewise.
18315 * hsa-gen.c: Likewise.
18316 * internal-fn.c: Likewise.
18317 * ipa-chkp.c: Likewise.
18318 * ipa-cp.c: Likewise.
18319 * ipa-devirt.c: Likewise.
18320 * ipa-fnsummary.c: Likewise.
18321 * ipa-inline.c: Likewise.
18322 * ipa-visibility.c: Likewise.
18324 * lto-cgraph.c: Likewise.
18325 * omp-expand.c: Likewise.
18326 * omp-general.c: Likewise.
18327 * omp-low.c: Likewise.
18328 * omp-offload.c: Likewise.
18329 * omp-simd-clone.c: Likewise.
18330 * opts-global.c: Likewise.
18331 * passes.c: Likewise.
18332 * predict.c: Likewise.
18333 * sancov.c: Likewise.
18334 * sanopt.c: Likewise.
18335 * symtab.c: Likewise.
18336 * toplev.c: Likewise.
18337 * trans-mem.c: Likewise.
18338 * tree-chkp.c: Likewise.
18339 * tree-eh.c: Likewise.
18340 * tree-into-ssa.c: Likewise.
18341 * tree-object-size.c: Likewise.
18342 * tree-parloops.c: Likewise.
18343 * tree-profile.c: Likewise.
18344 * tree-ssa-ccp.c: Likewise.
18345 * tree-ssa-live.c: Likewise.
18346 * tree-ssa-loop.c: Likewise.
18347 * tree-ssa-sccvn.c: Likewise.
18348 * tree-ssa-structalias.c: Likewise.
18349 * tree-ssa.c: Likewise.
18350 * tree-streamer-in.c: Likewise.
18351 * tree-vectorizer.c: Likewise.
18352 * tree-vrp.c: Likewise.
18353 * tsan.c: Likewise.
18354 * ubsan.c: Likewise.
18355 * varasm.c: Likewise.
18356 * varpool.c: Likewise.
18357 * tree.c: Remove functions moved to attribs.[ch].
18358 * tree.h: Likewise.
18359 * config/aarch64/aarch64.c: Add attrs.h header file.
18360 * config/alpha/alpha.c: Likewise.
18361 * config/arc/arc.c: Likewise.
18362 * config/arm/arm.c: Likewise.
18363 * config/avr/avr.c: Likewise.
18364 * config/bfin/bfin.c: Likewise.
18365 * config/c6x/c6x.c: Likewise.
18366 * config/cr16/cr16.c: Likewise.
18367 * config/cris/cris.c: Likewise.
18368 * config/darwin.c: Likewise.
18369 * config/epiphany/epiphany.c: Likewise.
18370 * config/fr30/fr30.c: Likewise.
18371 * config/frv/frv.c: Likewise.
18372 * config/ft32/ft32.c: Likewise.
18373 * config/h8300/h8300.c: Likewise.
18374 * config/i386/winnt.c: Likewise.
18375 * config/ia64/ia64.c: Likewise.
18376 * config/iq2000/iq2000.c: Likewise.
18377 * config/lm32/lm32.c: Likewise.
18378 * config/m32c/m32c.c: Likewise.
18379 * config/m32r/m32r.c: Likewise.
18380 * config/m68k/m68k.c: Likewise.
18381 * config/mcore/mcore.c: Likewise.
18382 * config/microblaze/microblaze.c: Likewise.
18383 * config/mips/mips.c: Likewise.
18384 * config/mmix/mmix.c: Likewise.
18385 * config/mn10300/mn10300.c: Likewise.
18386 * config/moxie/moxie.c: Likewise.
18387 * config/msp430/msp430.c: Likewise.
18388 * config/nds32/nds32-isr.c: Likewise.
18389 * config/nds32/nds32.c: Likewise.
18390 * config/nios2/nios2.c: Likewise.
18391 * config/nvptx/nvptx.c: Likewise.
18392 * config/pa/pa.c: Likewise.
18393 * config/pdp11/pdp11.c: Likewise.
18394 * config/powerpcspe/powerpcspe.c: Likewise.
18395 * config/riscv/riscv.c: Likewise.
18396 * config/rl78/rl78.c: Likewise.
18397 * config/rx/rx.c: Likewise.
18398 * config/s390/s390.c: Likewise.
18399 * config/sh/sh.c: Likewise.
18400 * config/sol2.c: Likewise.
18401 * config/sparc/sparc.c: Likewise.
18402 * config/spu/spu.c: Likewise.
18403 * config/stormy16/stormy16.c: Likewise.
18404 * config/tilegx/tilegx.c: Likewise.
18405 * config/tilepro/tilepro.c: Likewise.
18406 * config/v850/v850.c: Likewise.
18407 * config/vax/vax.c: Likewise.
18408 * config/visium/visium.c: Likewise.
18409 * config/xtensa/xtensa.c: Likewise.
18411 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18414 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
18415 constraints since the -mupper-regs-* switches have been
18417 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
18418 into a vector from a double word element that was extracted from
18419 another vector, and eliminate extra XXPERMDI instructions.
18420 (vsx_concat_<mode>_2): Likewise.
18421 (vsx_concat_<mode>_3): Likewise.
18422 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
18423 concat to allow optimizing inserts from previous extracts.
18425 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
18427 * config/i386/i386.c (ix86_stack_protect_guard): Generate
18428 memory reference to a SSP offset in TLS address space.
18429 (ix86_print_operand) <case '@'>: Remove.
18430 (ix86_print_operand_punct_valid_p): Remove '@' code.
18431 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
18432 UNSPEC_SP_TLS_TEST.
18433 (stack_tls_protect_set_<mode>): Remove.
18434 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
18435 (stack_tls_protect_test_<mode>): Remove.
18436 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
18438 2017-08-07 Olivier Hainque <hainque@adacore.com>
18441 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
18443 2017-08-07 Douglas Rupp <rupp@adacore.com>
18445 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
18446 variable was referenced as multidir in command.
18448 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18451 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
18454 2017-08-07 Martin Liska <mliska@suse.cz>
18456 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
18457 * config/rl78/rl78.c: Add include of attribs.h.
18458 * config/sh/sh.c: Likewise.
18459 * config/v850/v850.c: Likewise.
18461 2017-08-07 Tom de Vries <tom@codesourcery.com>
18463 PR middle-end/78266
18464 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
18466 2017-08-07 Martin Liska <mliska@suse.cz>
18468 * config/mips/mips.c: Include attribs.h.
18470 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18473 * doc/invoke.texi: Document change in behvaior for -Ofast for
18476 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
18478 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
18480 (aarch64_pop_regs): Likewise.
18481 (aarch64_gen_load_pair): Likewise.
18482 (aarch64_save_callee_saves): Likewise.
18483 (aarch64_restore_callee_saves): Likewise.
18485 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18487 * config/i386/i386.c: Revert the last change.
18489 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18492 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18494 (ix86_finalize_stack_frame_flags): This. Also clear
18495 frame_pointer_needed if -fno-omit-frame-pointer is used without
18497 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18498 with ix86_finalize_stack_frame_flags.
18499 (ix86_expand_epilogue): Likewise.
18500 (ix86_expand_split_stack_prologue): Likewise.
18502 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
18505 * config/i386/i386.c (get_builtin_code_for_version): Set priority
18506 to P_AES for Westmere.
18508 2017-08-07 Jonathan Yong <10walls@gmail.com>
18510 * config/i386/mingw.opt (fset-stack-executable): Removed.
18511 * config/i386/cygming.opt (fset-stack-executable): Moved
18513 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
18515 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
18517 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
18519 2017-08-07 Marek Polacek <polacek@redhat.com>
18521 PR middle-end/81737
18522 * fold-const.c (fold_indirect_ref_1): Check type_domain.
18524 2017-08-07 Martin Liska <mliska@suse.cz>
18526 * attribs.h (canonicalize_attr_name): New function.
18527 (cmp_attribs): Move from c-format.c and adjusted.
18528 (is_attribute_p): Moved from tree.h.
18529 * tree-inline.c: Add new includes.
18530 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
18531 (private_is_attribute_p): Remove.
18532 (private_lookup_attribute): Likewise.
18533 (private_lookup_attribute_by_prefix): Simplify.
18534 (remove_attribute): Use is_attribute_p.
18535 * tree.h: Remove removed declarations.
18537 2017-08-07 Jakub Jelinek <jakub@redhat.com>
18539 PR middle-end/81698
18540 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
18541 instead of computing it in the function. Formatting fix.
18542 (expand_case): Don't rely on default_edge being the first edge,
18543 clear it if removing it, pass default_edge to
18544 emit_case_dispatch_table.
18545 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
18548 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
18550 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
18551 insn in the function, emit NOP after the insn.
18553 2017-08-06 Tom de Vries <tom@codesourcery.com>
18555 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
18558 2017-08-06 Tom de Vries <tom@codesourcery.com>
18560 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
18563 2017-08-04 Yury Gribov <tetra2005@gmail.com>
18565 PR tree-optimization/57371
18566 * match.pd: New pattern.
18568 2017-08-04 Marek Polacek <polacek@redhat.com>
18570 PR middle-end/81695
18571 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
18572 perform the computation in offset_int.
18574 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18576 PR tree-optimization/81136
18577 * tree-vectorizer.h: Include tree-hash-traits.h.
18578 (vec_base_alignments): New typedef.
18579 (vec_info): Add a base_alignments field.
18580 (vect_record_base_alignments): Declare.
18581 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
18583 (DR_IS_CONDITIONAL_IN_STMT): New macro.
18584 (create_data_ref): Add an is_conditional_in_stmt argument.
18585 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
18586 the is_conditional_in_stmt field.
18587 (data_ref_loc): Add an is_conditional_in_stmt field.
18588 (get_references_in_stmt): Set the is_conditional_in_stmt field.
18589 (find_data_references_in_stmt): Update call to create_data_ref.
18590 (graphite_find_data_references_in_stmt): Likewise.
18591 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
18592 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
18593 (vect_record_base_alignment): New function.
18594 (vect_record_base_alignments): Likewise.
18595 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
18596 for nested statements even if we fail to compute a misalignment.
18597 Use pooled base alignments for unconditional references.
18598 (vect_find_same_alignment_drs): Compare base addresses instead
18600 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
18601 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
18603 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18605 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
18606 Add an explicit name for the enum. Use auto_vec for slp_instances
18607 and grouped_stores.
18608 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
18610 (_bb_vec_info): Add a constructor and destructor.
18611 (vinfo_for_stmt): Return NULL for uids of -1 as well.
18612 (destroy_loop_vec_info): Delete.
18613 (vect_destroy_datarefs): Likewise.
18614 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
18615 (vec_info::vec_info): New function.
18616 (vec_info::~vec_info): Likewise.
18617 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
18618 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
18619 destroy_loop_vec_info.
18620 * tree-vect-loop.c (new_loop_vec_info): Replace with...
18621 (_loop_vec_info::_loop_vec_info): ...this.
18622 (destroy_loop_vec_info): Replace with...
18623 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
18624 the stmt_vec_infos. Leave handling of vec_info information to its
18625 destructor. Remove explicit vector releases.
18626 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
18627 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
18628 * tree-vect-slp.c (new_bb_vec_info): Replace with...
18629 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
18630 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
18631 (destroy_bb_vec_info): Replace with...
18632 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
18633 information to its destructor.
18634 (vect_slp_analyze_bb_1): Use new and delete instead of
18635 new_bb_vec_info and destroy_bb_vec_info.
18636 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
18639 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
18641 * tree-data-ref.h (subscript): Add access_fn field.
18642 (data_dependence_relation): Add could_be_independent_p.
18643 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
18644 (same_access_functions): Move to tree-data-ref.c.
18645 * tree-data-ref.c (ref_contains_union_access_p): New function.
18646 (access_fn_component_p): Likewise.
18647 (access_fn_components_comparable_p): Likewise.
18648 (dr_analyze_indices): Add a reference to access_fn_component_p.
18649 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
18651 (constant_access_functions): Likewise.
18652 (add_other_self_distances): Likewise.
18653 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
18654 (initialize_data_dependence_relation): Use XCNEW and remove
18655 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
18656 of access functions that have the same type. Allow the
18657 subsequence to end with different bases in some circumstances.
18658 Record the chosen access functions in SUB_ACCESS_FN.
18659 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
18660 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
18661 (subscript_dependence_tester_1): Likewise dra and drb.
18662 (build_classic_dist_vector): Update calls accordingly.
18663 (subscript_dependence_tester): Likewise.
18664 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
18665 DDR_COULD_BE_INDEPENDENT_P.
18666 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
18667 comp_alias_ddrs instead of may_alias_ddrs.
18668 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
18670 (vect_analyze_data_ref_dependence): Use it if
18671 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
18672 distance vectors if that fails.
18673 (dependence_distance_ge_vf): New function.
18674 (vect_prune_runtime_alias_test_list): Use it. Don't clear
18675 LOOP_VINFO_MAY_ALIAS_DDRS.
18677 2017-08-04 Richard Biener <rguenther@suse.de>
18679 PR middle-end/81705
18680 * fold-const.c (fold_binary_loc): Properly restrict
18681 minus_var0 && minus_var1 case when associating undefined overflow
18684 2017-08-04 Tom de Vries <tom@codesourcery.com>
18686 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
18688 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18690 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18691 Don't start diagnostic messages with a capital letter.
18692 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18694 (rs6000_invalid_builtin): Likewise.
18695 (rs6000_trampoline_init): Likewise.
18697 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18700 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
18701 after setting changeable df flags.
18703 2017-08-03 Richard Biener <rguenther@suse.de>
18705 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
18706 up if the use is in USE - X.
18708 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
18710 * toplev.c (dumpfile.h): New include.
18711 (internal_error_reentered): New static function. Use it...
18712 (internal_error_function): ...here to handle reentered internal_error.
18714 2017-08-03 Richard Biener <rguenther@suse.de>
18716 PR middle-end/81148
18717 * fold-const.c (split_tree): Add minus_var and minus_con
18718 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
18719 here but always use minus_*.
18720 (associate_trees): Assert we never associate with MINUS_EXPR
18721 and NULL first operand. Do not recurse for PLUS_EXPR operands
18722 when associating as MINUS_EXPR either.
18723 (fold_binary_loc): Track minus_var and minus_con.
18725 2017-08-03 Tom de Vries <tom@codesourcery.com>
18728 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
18729 ACCEL_COMPILER, apply finish_options on
18730 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
18732 2017-08-03 Tom de Vries <tom@codesourcery.com>
18735 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
18736 function_entry_patch_area_size > 0.
18738 2017-08-03 Jakub Jelinek <jakub@redhat.com>
18741 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
18742 instead of 10??LU, perform unit multiplication in wide_int,
18743 don't change alloc_object_size_limit if the limit is larger
18746 PR tree-optimization/81655
18747 PR tree-optimization/81588
18748 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
18749 the case when ranges[i].low and high are 1 for unsigned type with
18752 PR middle-end/81052
18753 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
18754 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
18756 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18758 * tree-vrp.h: Add include guard.
18760 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
18763 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
18764 (ud2): New insn pattern.
18765 * config/i386/i386.c (ix86_expand_epilogue):
18766 For naked functions, generate ud2 instead of trap insn.
18768 2017-08-02 Marek Polacek <polacek@redhat.com>
18771 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
18773 2017-08-02 Tom de Vries <tom@codesourcery.com>
18774 Cesar Philippidis <cesar@codesourcery.com>
18776 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
18777 Add missing edge probabilities.
18779 2017-08-02 Tamar Christina <tamar.christina@arm.com>
18781 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18782 Correct endianness.
18784 2017-08-02 Jakub Jelinek <jakub@redhat.com>
18786 PR middle-end/79499
18787 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
18788 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
18789 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
18790 prologue_seq sequences - if any.
18792 2017-08-02 Richard Biener <rguenther@suse.de>
18794 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
18795 via vectors if supported, integer extracts via punning if supported
18796 or otherwise vector extracts.
18798 2017-08-02 Richard Biener <rguenther@suse.de>
18800 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
18802 (bitmap_insert_into_set): ... this.
18804 2017-08-02 Richard Biener <rguenther@suse.de>
18806 PR tree-optimization/81633
18808 2015-08-17 Alan Hayward <alan.hayward@arm.com>
18810 PR tree-optimization/71752
18811 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
18813 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
18815 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
18816 (machine_function::call_ms2sysv_pad_out): Remove field.
18817 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
18818 (ix86_compute_frame_layout): Likewise.
18820 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
18823 * config/i386/i386.c (ix86_set_func_type): Disallow naked
18824 attribute with interrupt attribute.
18826 2017-08-01 Andrew Pinski <apinski@cavium.com>
18828 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
18829 BIT_INSERT_EXPR's operand 1
18830 to see if the types precision matches.
18832 2017-08-01 Martin Liska <mliska@suse.cz>
18834 PR middle-end/70140
18835 * builtins.c (expand_builtin_memcpy_args): Remove.
18836 (expand_builtin_memcpy): Call newly added function
18837 expand_builtin_memory_copy_args.
18838 (expand_builtin_memcpy_with_bounds): Likewise.
18839 (expand_builtin_mempcpy): Remove last argument.
18840 (expand_builtin_mempcpy_with_bounds): Likewise.
18841 (expand_builtin_memory_copy_args): New function created from
18842 expand_builtin_mempcpy_args with small modifications.
18843 (expand_builtin_mempcpy_args): Remove.
18844 (expand_builtin_stpcpy): Remove unused argument.
18845 (expand_builtin): Likewise.
18846 (expand_builtin_with_bounds): Likewise.
18848 2017-08-01 Martin Liska <mliska@suse.cz>
18851 Make mempcpy more optimal (PR middle-end/70140).
18853 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18856 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
18857 __builtin_vec_cmpne verify both arguments are compatible vectors
18858 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
18859 verify arg1_type is a pointer or array type. For __builtin_vec_st,
18860 move computation of aligned to after checking the argument types.
18864 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18865 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
18868 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
18869 Doug Rupp <rupp@adacore.com>
18870 Olivier Hainque <hainque@adacore.com>
18872 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
18873 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
18875 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
18876 for TARGET_OS_CPP_BUILTIN.
18877 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
18878 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
18880 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
18881 passing required abi options to the assembler for EABI configurations.
18882 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
18883 of .text.hot and .text.unlikely sections for kernel modules when
18884 using ARM style exceptions.
18885 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18886 options. Add EXTRA_CC1_SPEC.
18887 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18889 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18891 (ARM_TARGET2_DWARF_FORMAT): Define.
18892 * config/arm/t-vxworks: Adjust multilib control to removal of the
18893 Diab command line options.
18895 2017-08-01 Martin Liska <mliska@suse.cz>
18897 PR gcov-profile/81561
18898 * gcov.c (unblock): Make unblocking safe as we need to preserve
18899 index correspondence of blocks and block_lists.
18901 2017-08-01 Richard Biener <rguenther@suse.de>
18903 PR tree-optimization/81181
18904 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18905 (compute_antic): ... end of iteration here.
18907 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18909 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18910 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18911 (ftree-slp-vectorize): Likewise.
18912 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18913 can no longer be set independent of flag_tree_loop_vectorize.
18914 * omp-general.c (emp_max_vf): Likewise.
18915 * opts.c (enable_fdo_optimizations): Remove references to
18916 flag_tree_vectorize, these are now implicit.
18917 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18918 and leave it for the options machinery.
18920 2017-08-01 Martin Liska <mliska@suse.cz>
18922 PR middle-end/70140
18923 * builtins.c (expand_builtin_memcpy_args): Remove.
18924 (expand_builtin_memcpy): Call newly added function
18925 expand_builtin_memory_copy_args.
18926 (expand_builtin_memcpy_with_bounds): Likewise.
18927 (expand_builtin_mempcpy): Remove last argument.
18928 (expand_builtin_mempcpy_with_bounds): Likewise.
18929 (expand_builtin_memory_copy_args): New function created from
18930 expand_builtin_mempcpy_args with small modifications.
18931 (expand_builtin_mempcpy_args): Remove.
18932 (expand_builtin_stpcpy): Remove unused argument.
18933 (expand_builtin): Likewise.
18934 (expand_builtin_with_bounds): Likewise.
18936 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18939 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18940 print "ds:" only for immediates in generic address space.
18942 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18945 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18946 (ix86_function_ok_for_sibcall): Return false for naked functions.
18948 2017-08-01 Richard Biener <rguenther@suse.de>
18950 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18951 (compute_antic): Seed worklist with exit block predecessors.
18952 * cfganal.c (dfs_find_deadend): For a cycle return the source
18953 of the edge closing it.
18955 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18957 * config/aarch64/aarch64.c
18958 (aarch64_can_const_movi_rtx_p): Move 0 check.
18960 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18962 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18963 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18965 * match.pd: Ditto in address comparison pattern.
18967 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18969 PR tree-optimization/81627
18970 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18971 closed ssa form for store-store chain.
18973 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18975 PR tree-optimization/81620
18976 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18977 for store-store chain.
18979 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18981 PR tree-optimization/81588
18982 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18983 ranges[i].in_p, invert comparison code ccode. For >/>=,
18984 swap rhs1 and rhs2 and comparison code unconditionally,
18985 for </<= don't do that. Don't swap rhs1/rhs2 again if
18986 ranges[i].in_p, instead invert comparison code ccode if
18987 opcode or oe->rank is BIT_IOR_EXPR.
18990 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18991 a direct optab to conversion optab.
18992 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18993 with GET_MODE_INNER as last argument instead of optab_handler.
18994 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18995 vector extraction if possible and optab is available.
18996 * expr.c (store_constructor): Use convert_optab_handler instead
18997 of optab_handler. Use vector initialization from smaller
18998 vectors if possible and optab is available.
18999 * tree-vect-stmts.c (vectorizable_load): Likewise.
19000 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19001 now have two modes.
19002 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19003 of vec_init from half-sized vectors with the same element mode.
19004 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19005 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19006 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19007 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19008 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19009 after mode in gen_vec_extract* calls.
19010 (vec_extract<mode>): Renamed to ...
19011 (vec_extract<mode><ssescalarmodelower>): ... this.
19012 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19013 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19014 element mode after mode in gen_vec_init* calls.
19015 (VEC_INIT_HALF_MODE): New mode iterator.
19016 (vec_init<mode>): Renamed to ...
19017 (vec_init<mode><ssescalarmodelower>): ... this.
19018 (vec_init<mode><ssehalfvecmodelower>): New expander.
19019 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19020 (vec_extractv2sfsf): ... this.
19021 (vec_initv2sf): Renamed to ...
19022 (vec_initv2sfsf): ... this.
19023 (vec_extractv2si): Renamed to ...
19024 (vec_extractv2sisi): ... this.
19025 (vec_initv2si): Renamed to ...
19026 (vec_initv2sisi): ... this.
19027 (vec_extractv4hi): Renamed to ...
19028 (vec_extractv4hihi): ... this.
19029 (vec_initv4hi): Renamed to ...
19030 (vec_initv4hihi): ... this.
19031 (vec_extractv8qi): Renamed to ...
19032 (vec_extractv8qiqi): ... this.
19033 (vec_initv8qi): Renamed to ...
19034 (vec_initv8qiqi): ... this.
19035 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19036 (vec_init<mode>): Renamed to ...
19037 (vec_init<mode><VEC_base_l>): ... this.
19038 (vec_extract<mode>): Renamed to ...
19039 (vec_extract<mode><VEC_base_l>): ... this.
19040 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19041 (vec_initv2sfsf): ... this.
19042 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19043 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19044 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19045 element mode after mode in gen_vec_init* calls.
19046 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19047 (vec_init<mode><Vel>): ... this.
19048 (vec_extract<mode>): Renamed to ...
19049 (vec_extract<mode><Vel>): ... this.
19050 * config/aarch64/iterators.md (Vel): New mode attribute.
19051 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19052 Add element mode after mode in gen_vec_extract* calls.
19053 * config/s390/vector.md (non_vec_l): New mode attribute.
19054 (vec_extract<mode>): Renamed to ...
19055 (vec_extract<mode><non_vec_l>): ... this.
19056 (vec_init<mode>): Renamed to ...
19057 (vec_init<mode><non_vec_l>): ... this.
19058 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19059 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19061 * config/arm/iterators.md (V_elem_l): New mode attribute.
19062 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19063 (vec_extract<mode><V_elem_l>): ... this.
19064 (vec_extractv2di): Renamed to ...
19065 (vec_extractv2didi): ... this.
19066 (vec_init<mode>): Renamed to ...
19067 (vec_init<mode><V_elem_l>): ... this.
19068 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19069 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19070 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19071 Add element mode after gen_vec_extract* calls.
19072 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19073 (vec_init<mode><unitmode>): ... this.
19074 (vec_extract<mode>): Renamed to ...
19075 (vec_extract<mode><unitmode>): ... this.
19076 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19077 (vec_init<mode><unitmode>): ... this.
19078 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19079 (vec_initv2sfsf): ... this.
19080 (vec_extractv2sf): Renamed to ...
19081 (vec_extractv2sfsf): ... this.
19082 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19083 Add element mode after gen_vec_extract* calls.
19084 * config/mips/mips.md (unitmode): New mode iterator.
19085 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19086 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19087 * config/spu/spu.md (inner_l): New mode attribute.
19088 (vec_init<mode>): Renamed to ...
19089 (vec_init<mode><inner_l>): ... this.
19090 (vec_extract<mode>): Renamed to ...
19091 (vec_extract<mode><inner_l>): ... this.
19092 * config/sparc/sparc.md (veltmode): New mode iterator.
19093 (vec_init<VMALL:mode>): Renamed to ...
19094 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19095 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19096 (vec_initv2sisi): ... this.
19097 (vec_initv2sf): Renamed to ...
19098 (vec_initv2sfsf): ... this.
19099 (vec_extractv2sf): Renamed to ...
19100 (vec_extractv2sfsf): ... this.
19101 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19102 (vec_init<mode>): Renamed to ...
19103 (vec_init<mode><VEC_base_l>): ... this.
19104 (vec_extract<mode>): Renamed to ...
19105 (vec_extract<mode><VEC_base_l>): ... this.
19106 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19107 (vec_initv2sfsf): ... this.
19108 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19109 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19110 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19111 gen_vec_init* calls.
19113 2017-08-01 Richard Biener <rguenther@suse.de>
19115 PR tree-optimization/81297
19116 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19117 TREE_OVERFLOW from INTEGER_CSTs.
19119 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19121 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19123 2017-07-31 Carl Love <cel@us.ibm.com>
19125 * config/rs6000/rs6000-c: Add support for built-in functions
19126 vector signed char vec_xl_be (signed long long, signed char *);
19127 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19128 vector signed int vec_xl_be (signed long long, signed int *);
19129 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19130 vector signed long long vec_xl_be (signed long long, signed long long *);
19131 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19132 vector signed short vec_xl_be (signed long long, signed short *);
19133 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19134 vector double vec_xl_be (signed long long, double *);
19135 vector float vec_xl_be (signed long long, float *);
19136 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19137 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19138 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19140 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19141 (altivec_expand_builtin): Add switch statement to call
19142 altivec_expand_xl_be for each builtin.
19143 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19144 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19145 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19146 __builtin_vsx_le_be_v16qi.
19147 * doc/extend.texi: Update the built-in documentation file for the
19148 new built-in functions.
19150 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19153 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19155 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19157 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19159 * config.gcc: Add z14.
19160 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19161 CPU model numbers for z13s and z14.
19162 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19164 * config/s390/s390-opts.h (enum processor_type): Rename
19165 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19166 * config/s390/s390.c (processor_table): Add field for CPU name to
19167 be passed to Binutils.
19168 (s390_asm_output_machine_for_arch): Use the new field in
19169 processor_table for Binutils.
19170 (s390_expand_builtin): Replace arch12 with z14.
19171 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19172 (s390_get_sched_attrmask): Likewise.
19173 (s390_get_unit_mask): Likewise.
19174 * config/s390/s390.opt: Add z14 to processor_type enum.
19176 2017-07-31 Martin Jambor <mjambor@suse.cz>
19179 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19180 regardless of optimization level.
19182 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19183 Martin Liska <mliska@suse.cz>
19185 * predict.def: Remove old comment and adjust probability.
19186 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19187 PREDICT statements.
19189 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19192 * config/i386/i386.c (ix86_function_naked): New function.
19193 (ix86_can_use_return_insn_p): Return false for naked functions.
19194 (ix86_expand_prologue): Skip prologue for naked functions.
19195 (ix86_expand_epilogue): Skip epilogue for naked functions
19196 and emit trap instruction.
19197 (ix86_warn_func_return): New function.
19198 (ix86_attribute_table): Add "naked" attribute specification.
19199 (TARGET_WARN_FUNC_RETURN): Define.
19200 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19202 2017-07-31 Martin Liska <mliska@suse.cz>
19204 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19205 (dump_gimple_bb_header): Always dump BB info.
19206 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19208 2017-07-31 Martin Liska <mliska@suse.cz>
19211 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19212 also with current_function_decl non-null equality.
19214 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19217 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19218 change type to the element type, instead add eltype variable and
19219 use it where we are interested in the element type.
19221 PR tree-optimization/81603
19222 * ipa-polymorphic-call.c
19223 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19224 offset arithmetic in offset_int, bail out if the resulting bit offset
19225 doesn't fit into shwi.
19227 2017-07-31 Martin Liska <mliska@suse.cz>
19229 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19230 (gimplify_save_expr): Fix comment.
19232 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19235 * config/i386/i386.c (ix86_function_arg): Update arguments for
19237 (ix86_compute_frame_layout): Set the initial stack offset to
19238 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19239 INCOMING_FRAME_SP_OFFSET.
19240 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19241 stack before exception handler returns.
19242 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19243 the 'ERROR_CODE' for exception handler.
19245 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19247 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19248 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19249 (ASM_OUTPUT_REG_POP): Ditto.
19250 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
19251 instead of asm_fprintf to output pure string.
19253 2017-07-29 Jakub Jelinek <jakub@redhat.com>
19255 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
19256 to imported_module_or_decl hook.
19257 (debug_nothing_tree_tree_tree_bool): Remove.
19258 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
19259 * debug.c (do_nothing_debug_hooks): Use
19260 debug_nothing_tree_tree_tree_bool_bool instead of
19261 debug_nothing_tree_tree_tree_bool.
19262 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19263 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
19264 * sdbout.c (sdb_debug_hooks): Likewise.
19265 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
19266 (gen_namespace_die): Add DW_AT_export_symbols attribute if
19268 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
19269 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
19270 attribute, don't add anything.
19272 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19274 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
19275 (fold_build2_stat_loc): Likewise.
19276 (fold_build3_stat_loc): Likewise.
19277 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
19278 (fold_build1_loc): Remove macro.
19279 (fold_build2_loc): Likewise.
19280 (fold_build3_loc): Likewise.
19282 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19284 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
19285 (gimple_build_debug_bind_source_stat): Likewise.
19286 * gimple.h (gimple_build_debug_bind): Remove macro.
19287 (gimple_build_debug_bind_source): Likewise.
19289 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19291 * bitmap.c (bitmap_alloc): Adjust.
19292 (bitmap_gc_alloc): Likewise.
19293 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
19295 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19297 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
19298 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
19299 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
19300 (bitmap_gc_alloc_stat): Likewise.
19301 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
19303 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19305 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
19306 * rtl.h (shallow_copy_rtx): Remove macro.
19308 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19310 * emit-rtl.c (gen_raw_REG): Adjust.
19311 * gengenrtl.c (gendef): Likewise.
19312 * rtl.c (rtx_alloc_stat): Remove _stat from name.
19313 * rtl.h (rtx_alloc): Remove macro.
19315 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19317 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
19318 (build_tree_list_stat): Likewise.
19319 * tree.h (build_tree_list): Remove macro.
19320 (build_tree_list_vec): Likewise.
19322 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19324 * tree.c (make_vector_stat): Remove _stat from name.
19325 (build_vector_stat): Likewise.
19326 * tree.h (make_vector_stat): Remove macro.
19327 (build_vector_stat): Likewise.
19329 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19331 * tree.h (build_var_debug_value): Remove prototype.
19333 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19335 * tree.c (tree_cons_stat): Remove _stat from name.
19336 * tree.h (tree_cons): Remove macro.
19338 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19340 * tree.c (build_vl_exp_stat): Remove _stat from name.
19341 * tree.h (build_vl_exp): Remove macro.
19343 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19345 * tree.c (build_decl_stat): Remove _stat from name.
19346 * tree.h (build_decl): Remove macro.
19348 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19350 * gimple.c (gimple_build_with_ops_stat): Adjust.
19351 (gimple_alloc_stat): Remove _stat from name.
19352 * gimple.h (gimple_alloc): Remove macro.
19354 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19356 * tree.c (make_tree_vec_stat): Remove _stat from name.
19357 (grow_tree_vec_stat): Likewise.
19358 * tree.h (make_tree_vec_stat): Adjust prototype.
19359 (grow_tree_vec_stat): Likewise.
19360 (make_tree_vec): Remove macro.
19361 (grow_tree_vec): Likewise.
19363 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19365 * fold-const.c (fold_build1_stat_loc): Adjust.
19366 (fold_build2_stat_loc): Likewise.
19367 (fold_build3_stat_loc): Likewise.
19368 * tree.c (build0_stat): Remove _stat from name.
19369 (build1_stat): Likewise.
19370 (build2_stat): Likewise.
19371 (build3_stat): Likewise.
19372 (build4_stat): Likewise.
19373 (build5_stat): Likewise.
19374 * tree.h (build1_loc): Remove macro, and rename _stat function
19376 (build2_loc): Likewise.
19377 (build3_loc): Likewise.
19378 (build4_loc): Likewise.
19379 (build5_loc): Likewise.
19381 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19383 * tree.c (make_int_cst_stat): Remove _stat from name.
19384 * tree.h (make_int_cst_stat): Adjust prototype.
19385 (make_int_cst): Remove macro.
19387 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19389 * tree.c (make_tre_binfo_stat): Remove _stat from name.
19390 * tree.h (make_tree_binfo_stat): Adjust prototype.
19391 (make_tree_binfo): Remove.
19393 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19395 * tree.c (copy_node_stat): Rename to copy_node.
19396 (build_distinct_type_copy): Adjust.
19397 * tree.h (copy_node_stat): Adjust prototype.
19398 (copy_node): Remove macro.
19400 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19402 * tree.c (make_node_stat): rename to make_node.
19403 (build_tree_list_stat): Adjust.
19404 (build0_stat): Likewise.
19405 (build2_stat): Likewise.
19406 (build3_stat): Likewise.
19407 (build4_stat): Likewise.
19408 (build5_stat): Likewise.
19409 (build_decl_stat): Likewise.
19410 * tree.h (make_node_stat): Adjust prototype.
19411 (make_node): remove macro.
19413 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
19415 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
19416 (PPC_FEATURE2_SCV): Likewise.
19417 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
19419 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19421 * config/aarch64/aarch64.c
19422 (aarch64_internal_mov_immediate): Add new special pattern.
19423 * config/aarch64/aarch64.md (*movdi_aarch64):
19424 Add reg/32bit const mov case.
19426 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19427 Richard Sandiford <richard.sandiford@linaro.org>
19429 * config/aarch64/aarch64.md (mov<mode>): Generalize.
19430 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19431 Add integer and movi cases.
19432 (movi-split-hf-df-sf split, fp16): New.
19433 (enabled): Added TARGET_FP_F16INST.
19434 * config/aarch64/iterators.md (GPF_HF): New.
19435 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
19437 2017-07-28 Tamar Christina <tamar.christina@arm.com>
19439 * config/aarch64/aarch64.c
19440 (aarch64_simd_container_mode): Add prototype.
19441 (aarch64_expand_mov_immediate): Add HI support.
19442 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
19443 (aarch64_can_const_movi_rtx_p): New.
19444 (aarch64_preferred_reload_class):
19445 Remove restrictions of using FP registers for certain SIMD operations.
19446 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
19447 (aarch64_valid_floating_const): Add integer move validation.
19448 (aarch64_simd_imm_scalar_p): Remove.
19449 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
19450 (aarch64_legitimate_constant_p): Expand list of supported cases.
19451 * config/aarch64/aarch64-protos.h
19452 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
19453 (aarch64_reinterpret_float_as_int): New.
19454 (aarch64_simd_imm_scalar_p): Remove.
19455 * config/aarch64/constraints.md (Uvi): New.
19456 (Dd): Split into Ds and new Dd.
19457 * config/aarch64/aarch64.md (*movsi_aarch64):
19459 (*movdi_aarch64): Add SIMD mov case.
19461 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19463 * tree-predcom.c: (struct chain): Handle store-store chain in which
19464 stores for elimination only store loop invariant values.
19465 (execute_pred_commoning_chain): Ditto.
19466 (prepare_initializers_chain_store_elim): Ditto.
19467 (prepare_finalizers): Ditto.
19468 (is_inv_store_elimination_chain): New function.
19469 (initialize_root_vars_store_elim_1): New function.
19471 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19473 * tree-predcom.c: Revise general description of the pass.
19474 (enum chain_type): New enum type for store elimination.
19475 (struct chain): New field supporting store elimination.
19476 (struct component): Ditto.
19477 (dump_chain): Dump store-stores chain.
19478 (release_chain): Release resources.
19479 (split_data_refs_to_components): Compute and create component
19480 contains only stores for elimination.
19481 (get_chain_last_ref_at): New function.
19482 (make_invariant_chain): Initialization.
19483 (make_rooted_chain): Specify chain type in parameter and record it.
19484 (add_looparound_copies): Skip for store-stores chain.
19485 (determine_roots_comp): Compute type of chain and pass it to
19487 (initialize_root_vars_store_elim_2): New function.
19488 (finalize_eliminated_stores): New function.
19489 (remove_stmt): Handle store for elimination.
19490 (execute_pred_commoning_chain): Execute predictive commoning on
19491 store-store chains.
19492 (determine_unroll_factor): Skip unroll for store-stores chain.
19493 (prepare_initializers_chain_store_elim): New function.
19494 (prepare_initializers_chain): Hanlde store-store chain.
19495 (prepare_finalizers_chain, prepare_finalizers): New function.
19496 (tree_predictive_commoning_loop): Return integer value indicating
19497 if loop is unrolled or lcssa form is corrupted.
19498 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
19500 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19502 * tree-predcom.c (initialize_root): Delete.
19503 (execute_pred_commoning_chain): Initialize root vars and replace
19504 reference of non-combined chain directly, rather than call above
19507 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19509 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
19510 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
19512 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19514 * tree-predcom.c (struct chain): New field init_seq.
19515 (release_chain): Release init_seq.
19516 (prepare_initializers_chain): Record intialization stmts in above
19518 (insert_init_seqs): New function.
19519 (tree_predictive_commoning_loop): Call insert_init_seqs.
19521 2017-07-28 Bin Cheng <bin.cheng@arm.com>
19523 * tree-predcom.c (determine_roots_comp): Skip trivial components.
19525 2017-07-28 Richard Biener <rguenther@suse.de>
19527 * match.pd: Remove superfluous :c.
19528 * genmatch.c (simplify::id): Add member.
19529 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
19531 (current_id): New global.
19532 (dt_node::parent): Move from ...
19533 (dt_operand::parent): ... here. Add for_id member.
19534 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
19535 (decision_tree::find_node): Relax order requirement when
19536 merging DT_TRUE nodes to ones inbetween the current simplify
19537 and the one we try to merge with. Add diagnostic whenever
19538 we need to enforce pattern order by not merging.
19539 (decision_tree::insert): Set current_id.
19540 (decision_tree::print_node): Dump parent node and for_id.
19541 (parser::last_id): Add member.
19542 (parser::push_simplify): Assign unique id.
19543 (parser::parser): Initialize last_id.
19545 2017-07-28 Martin Liska <mliska@suse.cz>
19548 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
19549 gimple_build_debug_bind.
19551 2017-07-28 Richard Biener <rguenther@suse.de>
19553 PR tree-optimization/81502
19554 * match.pd: Add pattern combining BIT_INSERT_EXPR with
19556 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
19558 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
19559 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
19560 for BIT_FIELD_REF args.
19561 * fold-const.c (make_bit_field_ref): Likewise.
19562 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19564 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19567 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
19568 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
19569 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
19570 Or it into SANITIZER_UNDEFINED.
19571 * ubsan.c: Include gimple-fold.h and varasm.h.
19572 (ubsan_expand_ptr_ifn): New function.
19573 (instrument_pointer_overflow): New function.
19574 (maybe_instrument_pointer_overflow): New function.
19575 (instrument_object_size): Formatting fix.
19576 (pass_ubsan::execute): Call instrument_pointer_overflow
19577 and maybe_instrument_pointer_overflow.
19578 * internal-fn.c (expand_UBSAN_PTR): New function.
19579 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
19580 * sanitizer.def (__ubsan_handle_pointer_overflow,
19581 __ubsan_handle_pointer_overflow_abort): New builtins.
19582 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
19583 * internal-fn.def (UBSAN_PTR): New internal function.
19584 * opts.c (sanitizer_opts): Add pointer-overflow.
19585 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
19586 * fold-const.c (build_range_check): Compute pointer range check in
19587 integral type if pointer arithmetics would be needed. Formatting
19590 2017-07-28 Martin Liska <mliska@suse.cz>
19593 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
19594 parameters that are of a variable-length.
19596 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
19598 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
19600 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
19601 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
19602 (CRT_CALL_STATIC_FUNCTION): Likewise.
19603 (ASM_DEFAULT_SPEC): New define.
19604 (ASM_SPEC32): Likewise.
19605 (ASM_SPEC64): Likewise.
19606 (ASM_SPEC_COMMON): Likewise.
19607 (ASM_SPEC): Likewise.
19608 (INVALID_64BIT): Likewise.
19609 (LINK_OS_DEFAULT_SPEC): Likewise.
19610 (LINK_OS_SPEC32): Likewise.
19611 (LINK_OS_SPEC64): Likewise.
19612 (POWERPC_LINUX): Likewise.
19613 (PTRDIFF_TYPE): Likewise.
19614 (RESTORE_FP_PREFIX): Likewise.
19615 (RESTORE_FP_SUFFIX): Likewise.
19616 (SAVE_FP_PREFIX): Likewise.
19617 (SAVE_FP_SUFFIX): Likewise.
19618 (SIZE_TYPE): Likewise.
19619 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
19620 (TARGET_64BIT): Likewise.
19621 (TARGET_64BIT): Likewise.
19622 (TARGET_AIX): Likewise.
19623 (WCHAR_TYPE_SIZE): Likewise.
19624 (WCHAR_TYPE): Undefine.
19625 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
19626 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
19627 (CPP_OS_RTEMS_SPEC): Delete.
19628 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
19629 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
19631 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
19633 2017-07-28 Jakub Jelinek <jakub@redhat.com>
19635 PR tree-optimization/81578
19636 * tree-parloops.c (build_new_reduction): Bail out if
19637 reduction_code isn't one of the standard OpenMP reductions.
19638 Move the details printing after that decision.
19640 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19642 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
19643 related to reload_in_progress.
19644 (splat_input_operand): Likewise.
19645 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
19647 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
19649 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
19650 (TARGET_INSTANTIATE_DECLS): Likewise.
19651 (legitimate_indexed_address_p): Delete reload_in_progress code.
19652 (rs6000_debug_legitimate_address_p): Likewise.
19653 (rs6000_eliminate_indexed_memrefs): Likewise.
19654 (rs6000_emit_le_vsx_store): Likewise.
19655 (rs6000_emit_move_si_sf_subreg): Likewise.
19656 (rs6000_emit_move): Likewise.
19657 (register_to_reg_type): Likewise.
19658 (rs6000_pre_atomic_barrier): Likewise.
19659 (rs6000_machopic_legitimize_pic_address): Likewise.
19660 (rs6000_allocate_stack_temp): Likewise.
19661 (rs6000_address_for_fpconvert): Likewise.
19662 (rs6000_address_for_altivec): Likewise.
19663 (rs6000_secondary_memory_needed_rtx): Delete function.
19664 (rs6000_check_sdmode): Likewise.
19665 (rs6000_alloc_sdmode_stack_slot): Likewise.
19666 (rs6000_instantiate_decls): Likewise.
19667 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
19668 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
19669 Delete reload_in_progress.
19670 (*vec_reload_and_plus_<mptrsize>): Likewise.
19671 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
19672 (vsx_div_v2di): Likewise.
19673 (vsx_udiv_v2di): Likewise.
19675 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19677 * config/rs6000/rs6000.opt (mlra): Replace with stub.
19678 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
19679 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
19680 (rs6000_debug_reg_global): Delete print of LRA status.
19681 (rs6000_option_override_internal): Delete dead LRA related code.
19682 (rs6000_lra_p): Delete function.
19683 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
19685 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
19687 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
19688 * config/riscv/rtems.h: New file.
19690 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19691 Sudakshina Das <sudi.das@arm.com>
19693 * config/aarch64/aarch64.md
19694 (define_split for and<mode>3nr_compare): Move
19695 non aarch64_logical_operand to a register.
19696 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
19697 register immediate operand to a register.
19698 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
19700 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
19702 PR middle-end/81564
19703 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
19705 2017-07-27 Richard Biener <rguenther@suse.de>
19707 PR tree-optimization/81573
19708 PR tree-optimization/81494
19709 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
19710 multi defuse cycle case.
19712 2017-07-27 Richard Biener <rguenther@suse.de>
19714 PR tree-optimization/81571
19715 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
19718 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
19720 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
19721 earlier and only if MASK_FPU is set. Adjust formatting.
19723 2017-07-27 Martin Liska <mliska@suse.cz>
19725 * opt-functions.awk: Add validation of value of Init.
19726 * optc-gen.awk: Pass new argument.
19728 2017-07-27 Martin Liska <mliska@suse.cz>
19730 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
19731 Fix wrong condition.
19733 2017-07-27 Martin Liska <mliska@suse.cz>
19735 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
19736 BBs and edges seen by autoFDO.
19738 2017-07-27 Richard Biener <rguenther@suse.de>
19740 PR tree-optimization/81502
19741 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
19742 with incompatible but same sized type.
19743 (execute_update_addresses_taken): Likewise.
19745 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
19747 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
19748 flag_tree_loop_vectorize rather than flag_tree_vectorize.
19750 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19753 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
19754 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
19755 Change s_operand to memory_operand.
19757 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
19759 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
19760 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
19761 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
19762 Emit instructions rather than returning an expression. Handle TFmode
19763 and KFmode by casting to TImode.
19764 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
19765 (rs6000_emit_le_vsx_store): Likewise.
19766 * config/rs6000/vsx.md (VSX_TI): New iterator.
19767 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
19768 (*vsx_le_undo_permute_<mode>): Likewise.
19769 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
19770 emit the split sequence.
19771 (*vsx_le_perm_store_<mode>): Likewise.
19773 2017-07-27 Jakub Jelinek <jakub@redhat.com>
19775 PR tree-optimization/81555
19776 PR tree-optimization/81556
19777 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
19778 if true, force CHANGED for the recursive invocation.
19779 (reassociate_bb): Remember original length of ops array, pass
19780 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
19782 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
19783 attributes for noipa attribute. For naked attribute use
19784 lookup_attribute first before lookup_attribute_spec.
19785 * final.c (rest_of_handle_final): Disable IPA RA for functions with
19787 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
19788 for functions with noipa attribute.
19789 (cgraph_externally_visible_p): Return true for functions with noipa
19791 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
19792 for functions with noipa attribute.
19793 * doc/extend.texi: Document noipa function attribute.
19794 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
19795 also for functions with noipa attribute.
19796 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
19798 2017-07-26 Andrew Pinski <apinski@cavium.com>
19800 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
19801 vec_unalign_load_cost and vec_unalign_store_cost.
19803 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
19805 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
19806 -mvsx-small-integer option.
19807 (ISA_3_0_MASKS_IEEE): Likewise.
19808 (OTHER_VSX_VECTOR_MASKS): Likewise.
19809 (POWERPC_MASKS): Likewise.
19810 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
19811 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
19812 code, only testing for DImode being allowed in non-VSX floating
19814 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
19815 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
19817 (rs6000_option_override_internal): Delete -mvsx-small-integer.
19818 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
19819 TARGET_P8_VECTOR test.
19820 (rs6000_secondary_reload_simple_move): Likewise.
19821 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
19822 since TARGET_P9_VECTOR was already tested.
19823 (rs6000_opt_masks): Remove -mvsx-small-integer.
19824 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
19825 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19827 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
19828 test for TARGET_VEXTRACTUB was used, and that uses
19830 (p9 extract splitter): Likewise.
19831 (vsx_extract_<mode>_di_p9): Likewise.
19832 (vsx_extract_<mode>_store_p9): Likewise.
19833 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
19834 for TARGET_P9_VECTOR was used. Delete code that is now dead with
19835 the elimination of TARGET_VSX_SMALL_INTEGER.
19836 (vsx_extract_<mode>_p8): Likewise.
19837 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
19838 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
19839 (vsx_set_<mode>_p9): Likewise.
19840 (vsx_set_v4sf_p9): Likewise.
19841 (vsx_set_v4sf_p9_zero): Likewise.
19842 (vsx_insert_extract_v4sf_p9): Likewise.
19843 (vsx_insert_extract_v4sf_p9_2): Likewise.
19844 * config/rs6000/rs6000.md (sign extend splitter): Change
19845 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
19846 (floatsi<mode>2_lfiwax_mem): Likewise.
19847 (floatunssi<mode>2_lfiwzx_mem): Likewise.
19848 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
19849 since a test for TARGET_P9_VECTOR was used.
19850 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19851 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
19852 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19853 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19854 TARGET_P8_VECTOR test.
19855 (fix_trunc<mode>si2_stfiwx): Likewise.
19856 (fix_trunc<mode>si2_internal): Likewise.
19857 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
19858 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19860 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19861 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
19862 TARGET_P8_VECTOR test.
19863 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19864 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
19865 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
19867 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19868 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
19869 since a test for TARGET_P9_VECTOR was used.
19870 (splitter for loading small constants): Likewise.
19872 2017-07-26 Andrew Pinski <apinski@cavium.com>
19874 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
19877 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
19880 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
19881 (fp_valid_at): Likewise.
19883 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19885 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19886 (qdf24xx_addrcost_table): Likewise.
19887 (cortexa57_tunings): Update to use generic_branch_cost.
19888 (cortexa72_tunings): Likewise.
19889 (cortexa73_tunings): Likewise.
19890 (qdf24xx_tunings): Likewise.
19892 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19894 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19895 (thunderx2t99_branch_cost): Likewise.
19896 (cortexa35_tunings): Update to use generic_branch_cost.
19897 (cortexa53_tunings): Likewise.
19898 (cortexa57_tunings): Likewise.
19899 (cortexa72_tunings): Likewise.
19900 (cortexa73_tunings): Likewise.
19901 (thunderx2t99_tunings): Likewise.
19903 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19905 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19906 (sparc_option_override): Honour MASK_FSMULD.
19907 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19908 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19909 * config/sparc/sparc.opt (mfsmuld): New option.
19910 * doc/invoke.texi (mfsmuld): Document option.
19912 2017-07-26 Marek Polacek <polacek@redhat.com>
19914 PR middle-end/70992
19915 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19917 2017-07-26 Richard Biener <rguenther@suse.de>
19919 * gimple-match-head.c (do_valueize): Return OP if valueize
19921 (get_def): New helper to get at the def stmt of a SSA name
19922 if valueize allows.
19923 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19924 do_valueize to get at the def stmt.
19925 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19927 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19929 PR middle-end/46932
19930 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19932 2017-07-26 Martin Liska <mliska@suse.cz>
19935 * function.c (expand_function_start): Make expansion of
19936 nonlocal_goto_save_area after parm_birth_insn.
19938 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19940 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19941 from all CPU target flags enable members.
19943 2017-07-26 Richard Biener <rguenther@suse.de>
19945 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19946 (decision_tree::gen): Make 'type' const.
19947 (write_predicate): Likewise.
19949 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19951 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19952 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19953 (rs6000_option_override_internal): Likewise.
19954 (rs6000_expand_vector_set): Likewise.
19955 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19956 (TARGET_UPPER_REGS_SF): Likewise.
19957 (TARGET_UPPER_REGS_DI): Likewise.
19958 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19959 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19960 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19961 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19962 (Splitters for DI constants in Altivec registers): Likewise.
19963 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19964 (vsx_set_v4sf_p9): Likewise.
19965 (vsx_set_v4sf_p9_zero): Likewise.
19966 (vsx_insert_extract_v4sf_p9): Likewise.
19967 (vsx_insert_extract_v4sf_p9_2): Likewise.
19969 2017-07-25 Carl Love <cel@us.ibm.com>
19971 * doc/extend.texi: Update the built-in documentation file for the
19972 existing built-in functions
19973 vector signed char vec_cnttz (vector signed char);
19974 vector unsigned char vec_cnttz (vector unsigned char);
19975 vector signed short vec_cnttz (vector signed short);
19976 vector unsigned short vec_cnttz (vector unsigned short);
19977 vector signed int vec_cnttz (vector signed int);
19978 vector unsigned int vec_cnttz (vector unsigned int);
19979 vector signed long long vec_cnttz (vector signed long long);
19980 vector unsigned long long vec_cnttz (vector unsigned long long);
19982 2017-07-25 Andrew Pinski <apinski@cavium.com>
19984 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19985 accesses where the use is for the first operand of a BIT_INSERT.
19987 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19990 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19991 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19993 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19995 * config/i386/gstabs.h: Delete.
19996 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19998 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20000 * config/i386/i386.c (ix86_decompose_address): Do not check for
20001 register RTX when looking at index_reg or base_reg.
20002 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20004 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20006 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20007 to update EH info here.
20009 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20011 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20013 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20015 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20017 2017-07-25 Torsten Duwe <duwe@suse.de>
20019 * common.opt: Introduce -fpatchable-function-entry
20020 command line option, and its variables function_entry_patch_area_size
20021 and function_entry_patch_area_start.
20022 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20023 including a two-value parser.
20024 * target.def (print_patchable_function_entry): New target hook.
20025 * targhooks.h (default_print_patchable_function_entry): New function.
20026 * targhooks.c (default_print_patchable_function_entry): Likewise.
20027 * toplev.c (process_options): Switch off IPA-RA if
20028 patchable function entries are being generated.
20029 * varasm.c (assemble_start_function): Look at the
20030 patchable-function-entry command line switch and current
20031 function attributes and maybe generate NOP instructions by
20032 calling the print_patchable_function_entry hook.
20033 * doc/extend.texi: Document patchable_function_entry attribute.
20034 * doc/invoke.texi: Document -fpatchable_function_entry
20035 command line option.
20036 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20038 * doc/tm.texi: Re-generate.
20040 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20043 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20044 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20046 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20048 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20050 2017-07-25 Richard Biener <rguenther@suse.de>
20052 PR tree-optimization/81455
20053 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20054 not walk in cycles when looking for guards.
20056 2017-07-25 Richard Biener <rguenther@suse.de>
20058 PR tree-optimization/81529
20059 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20060 when optimizing backedge uses.
20062 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20064 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20066 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20067 to dl_section_ref. On AIX, append an expression to subtract
20068 the size of the section length to dl_section_ref.
20070 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20072 * configure.ac: If any of the config.* scripts fail, exit 1.
20073 * configure: Regenerate.
20075 2017-07-25 Richard Biener <rguenther@suse.de>
20077 PR middle-end/81546
20078 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20079 of immediate uses, be more verbose on errors.
20081 2017-07-25 Richard Biener <rguenther@suse.de>
20083 PR tree-optimization/81510
20084 * tree-vect-loop.c (vect_is_simple_reduction): When the
20085 reduction stmt is not inside the loop bail out.
20087 2017-07-25 Richard Biener <rguenther@suse.de>
20089 PR tree-optimization/81303
20090 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20091 profitability check against LOOP_VINFO_NITERSM1.
20093 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20095 * domwalk.c (cmp_bb_postorder): Simplify.
20096 (sort_bbs_postorder): New function. Use it...
20097 (dom_walker::walk): ...here to optimize common cases.
20099 2017-07-25 Martin Liska <mliska@suse.cz>
20102 * ipa-visibility.c (function_and_variable_visibility): Make the
20103 redirection just on target that supports aliasing.
20104 Fix GNU coding style.
20106 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20109 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20111 * config/arm/rtems.h: Likewise.
20112 * config/bfin/rtems.h: Likewise.
20113 * config/i386/rtemself.h: Likewise.
20114 * config/lm32/rtems.h: Likewise.
20115 * config/m32c/rtems.h: Likewise.
20116 * config/m68k/rtemself.h: Likewise.
20117 * config/microblaze/rtems.h: Likewise.
20118 * config/mips/rtems.h: Likewise.
20119 * config/moxie/rtems.h: Likewise.
20120 * config/nios2/rtems.h: Likewise.
20121 * config/powerpcspe/rtems.h: Likewise.
20122 * config/rs6000/rtems.h: Likewise.
20123 * config/rtems.h: Likewise.
20124 * config/sh/rtems.h: Likewise.
20125 * config/sh/rtemself.h: Likewise.
20126 * config/sparc/rtemself.h: Likewise.
20128 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20131 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20132 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20133 * tree-ssa-structalias.c (alias_get_name): Same.
20135 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20138 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20139 instructions if no du chain is found.
20141 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20143 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20145 2017-07-25 Richard Biener <rguenther@suse.de>
20147 PR middle-end/81505
20148 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20151 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20153 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20154 upper-regs options.
20155 (ISA_2_7_MASKS_SERVER): Likewise.
20156 (ISA_3_0_MASKS_IEEE): Likewise.
20157 (OTHER_P8_VECTOR_MASKS): Likewise.
20158 (OTHER_VSX_VECTOR_MASKS): Likewise.
20159 (POWERPC_MASKS): Likewise.
20160 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20161 duplicate list of options.
20162 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20163 explicit -mupper-regs options.
20164 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20165 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20166 alias for -mupper-regs-df.
20167 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20168 (rs6000_init_hard_regno_mode_ok): Likewise.
20169 (rs6000_option_override_internal): Likewise.
20170 (rs6000_opt_masks): Likewise.
20171 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20172 options in terms of whether -mvsx or -mpower8-vector was used.
20173 (TARGET_UPPER_REGS_DI): Likewise.
20174 (TARGET_UPPER_REGS_SF): Likewise.
20175 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20176 -mupper-regs-* options.
20178 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20180 * passes.c (emergency_dump_function): Print some empty lines and a
20181 header before the RTL dump.
20183 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20185 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20187 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20190 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20191 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20193 2017-07-24 Carl Love <cel@us.ibm.com>
20195 * config/rs6000/rs6000-c.c: Add support for built-in functions
20196 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20197 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20198 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20199 vec_extract_fp_from_shortl): Add defines for the two builtins.
20200 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20201 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20203 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20204 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20205 * doc/extend.texi: Update the built-in documentation file for the
20206 new built-in function.
20208 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20211 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20213 * doc/generic.texi: Likewise.
20214 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20215 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20217 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20219 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20220 (aarch64_mls_elt_merge<mode>): Likewise.
20222 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20224 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20225 having __cxa_atexit.
20227 2017-07-23 Michael Collison <michael.collison@arm.com>
20229 * config/arm/arm.c (arm_option_override): Deprecate
20230 use of -mstructure-size-boundary.
20231 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20232 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20234 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20237 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20238 Reduce cost estimate for direct moves.
20240 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20243 * config/i386/i386.c (ix86_option_override_internal): Disable
20244 BMI, BMI2 and TBM instructions for -m16.
20246 2017-07-21 Carl Love <cel@us.ibm.com>
20248 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20249 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20250 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20251 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20252 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20253 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20254 VMULOSW): New enum "unspec" values.
20255 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20256 altivec_vmulosw): New patterns.
20257 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20258 VMULOSW): Add definitions.
20260 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
20262 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
20263 (qdf24xx): Likewise.
20264 * config/aarch64/aarch64-options-extensions.def (rdma); New.
20265 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
20266 (AARCH64_FL_V8_1): Renumber.
20267 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
20268 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
20269 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
20270 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
20271 rdma to feature modifiers list.
20273 2017-07-21 Yury Gribov <tetra2005@gmail.com>
20275 PR middle-end/56727
20276 * ipa-visibility (function_and_variable_visibility): Convert
20277 recursive PLT call to direct call if appropriate.
20279 2017-07-21 Andrew Pinski <apinski@cavium.com>
20281 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
20282 operand 1 to see if the types precision matches.
20283 * fold-const.c (operand_equal_p): Likewise.
20285 2017-07-21 Richard Biener <rguenther@suse.de>
20287 PR tree-optimization/81303
20288 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
20289 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
20290 (vect_peeling_hash_get_lowest_cost): Adjust.
20291 (vect_enhance_data_refs_alignment): Likewise. Use
20292 vect_get_peeling_costs_all_drs to compute the penalty for no
20293 peeling to match up costs.
20295 2017-07-21 Richard Biener <rguenther@suse.de>
20297 PR tree-optimization/81500
20298 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
20299 we didn't identify a reduction path.
20301 2017-07-21 Tom de Vries <tom@codesourcery.com>
20302 Cesar Philippidis <cesar@codesourcery.com>
20304 PR gcov-profile/81442
20305 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
20308 2017-07-21 Tom de Vries <tom@codesourcery.com>
20311 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
20313 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
20314 nvptx_override_options_after_change.
20316 2017-07-21 Ulrich Drepper <drepper@redhat.com>
20318 * dwarf2out.c (output_file_names): Avoid double testing for
20319 dwarf_version >= 5.
20321 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
20323 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
20325 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20327 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
20329 (try_crossjump_to_edge): Do not punt on partitioned functions.
20331 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
20333 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
20334 Put all BBs reachable only via paths crossing cold region to cold
20336 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
20338 2016-07-21 Richard Biener <rguenther@suse.de>
20340 PR tree-optimization/81303
20341 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
20342 into account prologue and epilogue iterations when raising
20343 min_profitable_iters to sth at least covering one vector iteration.
20345 2017-07-21 Tamar Christina <tamar.christina@arm.com>
20347 * config/arm/arm.c (arm_test_cpu_arch_dat):
20350 2017-07-20 Nathan Sidwell <nathan@acm.org>
20352 Remove TYPE_METHODS.
20353 * tree.h (TYPE_METHODS): Delete.
20354 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
20355 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
20356 (dbxout_type_methods): Scan TYPE_FIELDS.
20357 (dbxout_type): Don't check TYPE_METHODS here.
20358 * function.c (use_register_for_decl): Always ignore register for
20359 class types when not optimizing.
20360 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
20361 * tree.c (free_lang_data_in_type): Stitch out member functions and
20362 templates from TYPE_FIELDS.
20363 (build_distinct_type_copy, verify_type_variant,
20364 verify_type): Member fns are on TYPE_FIELDS.
20365 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
20366 * tree-pretty-print.c (dump_generic_node): Likewise.
20368 2017-07-20 Jakub Jelinek <jakub@redhat.com>
20371 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
20372 V2TImode and V4TImode.
20373 (ix86_expand_vector_extract): Likewise.
20374 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
20375 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
20376 (ssescalarmode): Handle V4TImode and V2TImode.
20377 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
20378 (*vec_extractv2ti, *vec_extractv4ti): New insns.
20379 (VEXTRACTI128_MODE): New mode iterator.
20380 (splitter for *vec_extractv?ti first element): New.
20381 (VEC_INIT_MODE): New mode iterator.
20382 (vec_init<mode>): Consolidate 3 expanders into one using
20383 VEC_INIT_MODE mode iterator.
20385 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20387 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
20388 non_spilled_static_chain_regno_p.
20390 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
20392 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
20394 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
20396 * bb-reorder.c (connect_traces): Allow copying of blocks within
20399 2017-07-20 Richard Biener <rguenther@suse.de>
20401 * gimple.h (gimple_phi_result): Add gphi * overload.
20402 (gimple_phi_result_ptr): Likewise.
20403 (gimple_phi_arg): Likewise. Adjust index assert to only
20404 allow actual argument accesses rather than all slots available
20406 (gimple_phi_arg_def): Add gphi * overload.
20407 * tree-phinodes.c (make_phi_node): Initialize only actual
20409 (resize_phi_node): Clear memory not covered by old node,
20410 do not initialize excess argument slots.
20411 (reserve_phi_args_for_new_edge): Initialize new argument slot
20414 2017-07-20 Bin Cheng <bin.cheng@arm.com>
20416 PR tree-optimization/81388
20418 2016-07-21 Bin Cheng <bin.cheng@arm.com>
20420 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
20421 by removing computation of may_be_zero.
20423 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20424 Tom de Vries <tom@codesourcery.com>
20426 PR middle-end/81030
20427 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
20428 when gimple level profile disagrees with what RTL expander did.
20430 2017-07-20 Richard Biener <rguenther@suse.de>
20432 PR tree-optimization/61171
20433 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
20434 (vect_analyze_stmt): Add slp instance parameter.
20435 (vectorizable_reduction): Likewise.
20436 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
20437 (vect_is_simple_reduction): Deal with chains not detected
20438 as SLP reduction chain, specifically not properly associated
20439 chains containing a mix of plus/minus.
20440 (get_reduction_op): Remove.
20441 (get_initial_defs_for_reduction): Simplify, pass in whether
20442 this is a reduction chain, pass in the SLP node for the PHIs.
20443 (vect_create_epilog_for_reduction): Get the SLP instance as
20445 (vectorizable_reduction): Get the SLP instance as arg.
20446 During analysis remember the SLP node with the PHIs in the
20447 instance. Simplify getting at the vectorized reduction PHIs.
20448 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
20449 through SLP instance.
20450 (vect_slp_analyze_operations): Likewise.
20451 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
20452 (vect_transform_stmt): Likewise.
20454 2017-07-20 Tom de Vries <tom@codesourcery.com>
20456 PR tree-optimization/81489
20457 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
20458 read of phi arg location to before loop that modifies phi.
20460 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20462 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
20465 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20467 PR middle-end/81331
20468 * except.c (execute): Fix ordering issue.
20470 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20472 PR rtl-optimization/81423
20473 * combine.c (make_compound_operation_int): Don't try to optimize
20474 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
20476 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
20478 PR rtl-optimization/81423
20479 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
20480 with a constant that is -1 in the truncated to mode.
20482 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20484 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
20485 (determine_unlikely_bbs): ... here.
20486 * predict.h (propagate_unlikely_bbs_forward): Declare.
20487 * cfgexpand.c (pass_expand::execute): Use it.
20488 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
20490 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
20491 propagate_unlikely_bbs_forward.
20493 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
20495 PR middle-end/81331
20496 * except.c (maybe_add_nop_after_section_switch): New function.
20499 2017-07-19 Tom de Vries <tom@codesourcery.com>
20501 * gimple.h (gimple_phi_set_arg): Make assert more strict.
20503 2017-07-19 Tom de Vries <tom@codesourcery.com>
20505 * gimple.h (gimple_phi_arg): Make assert more strict.
20507 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
20509 * config.gcc (powerpc*-*-*): Add mmintrin.h.
20510 * config/rs6000/mmintrin.h: New file.
20511 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
20513 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20515 PR tree-optimization/81346
20516 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
20518 2017-07-19 Tom de Vries <tom@codesourcery.com>
20520 * config/nvptx/nvptx.md (VECIM): Add V2DI.
20522 2017-07-19 Tom de Vries <tom@codesourcery.com>
20524 * config/nvptx/nvptx-modes.def: Add V2DImode.
20525 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
20526 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
20527 (nvptx_output_mov_insn): Handle lack of mov.b128.
20528 (nvptx_print_operand): Handle 'H' and 'L' codes.
20529 (nvptx_vector_mode_supported): Allow V2DImode.
20530 (nvptx_preferred_simd_mode): New function.
20531 (nvptx_data_alignment): New function.
20532 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
20533 nvptx_preferred_simd_mode.
20534 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
20536 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
20538 2017-07-19 Tom de Vries <tom@codesourcery.com>
20540 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
20541 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
20542 (nvptx_vector_mode_supported): New function. Allow V2SImode.
20543 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
20544 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
20545 (mov<VECIM>_insn): New define_insn.
20546 (define_expand "mov<VECIM>): New define_expand.
20548 2017-07-19 Tom de Vries <tom@codesourcery.com>
20550 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
20552 2017-07-19 Jakub Jelinek <jakub@redhat.com>
20554 PR tree-optimization/81346
20555 * fold-const.h (fold_div_compare, range_check_type): Declare.
20556 * fold-const.c (range_check_type): New function.
20557 (build_range_check): Use range_check_type.
20558 (fold_div_compare): No longer static, rewritten into
20559 a match.pd helper function.
20560 (fold_comparison): Don't call fold_div_compare here.
20561 * match.pd (X / C1 op C2): New optimization using fold_div_compare
20562 as helper function.
20564 2017-07-19 Nathan Sidwell <nathan@acm.org>
20566 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
20567 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
20568 * tree.c (find_decls_types_r, verify_type): Use
20569 TYPE_{MIN,MAX}_VALUE_RAW.
20570 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
20571 (hash_tree): Likewise.
20572 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20574 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20577 2017-07-18 Tom de Vries <tom@codesourcery.com>
20579 PR middle-end/81464
20580 * omp-expand.c (expand_omp_for_static_chunk): Handle
20581 equal-argument loop exit phi.
20583 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
20586 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
20587 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
20588 operand 2 predicate.
20589 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
20590 operand 2 predicate.
20591 (ror,rol -> rorx splitters): Use const_int_operand as
20592 operand 2 predicate.
20594 2017-06-18 Richard Biener <rguenther@suse.de>
20596 PR tree-optimization/81410
20597 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
20598 the gap in the ! slp_perm SLP case after each group.
20600 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20602 PR middle-end/81463
20603 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
20606 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
20608 PR middle-end/81462
20609 * predict.c (set_even_probabilities): Cleanup; do not affect
20610 probabilities that are already known.
20611 (combine_predictions_for_bb): Call even when count is set.
20613 2017-07-18 Nathan Sidwell <nathan@acm.org>
20615 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
20618 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20621 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
20622 optimization for loop niter analysis.
20624 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
20627 * config/avr/avr.c (avr_optimize_casesi): Don't use
20628 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
20630 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
20632 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
20633 body_cost_vec from _vect_peel_extended_info.
20634 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
20635 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
20638 2017-07-18 Bin Cheng <bin.cheng@arm.com>
20640 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
20642 2017-07-18 Richard Biener <rguenther@suse.de>
20644 PR tree-optimization/80620
20645 PR tree-optimization/81403
20646 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
20647 info when re-using a VN table entry.
20649 2017-07-18 Richard Biener <rguenther@suse.de>
20651 PR tree-optimization/81418
20652 * tree-vect-loop.c (vectorizable_reduction): Properly compute
20653 vectype_in. Verify that with lane-reducing reduction operations
20654 we have a single def-use cycle.
20656 2017-07-17 Carl Love <cel@us.ibm.com>
20658 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
20660 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20661 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20662 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20663 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20664 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20665 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20666 VMULOSW): New enum "unspec" values.
20667 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20668 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20669 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20670 altivec_vmulosw): New patterns.
20671 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20672 VMULOSW): Add definitions.
20674 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
20676 * config/alpha/alpha.c: Include predict.h.
20678 2017-07-17 Yury Gribov <tetra2005@gmail.com>
20680 * tree-vrp.c (compare_assert_loc): Fix comparison function
20681 to return predictable results.
20683 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20685 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
20687 (subdi3): Likewise.
20688 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
20689 * doc/invoke.texi (mexpand-adddi): Update text.
20691 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20693 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
20694 that also clobbers the CC register. The old expand code is moved
20696 (*arc_clzsi2): ... here.
20697 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
20698 the CC register. The old expand code is moved to ...
20699 (arc_ctzsi2): ... here.
20701 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20703 * config/arc/arc.opt (mindexed-loads): Use initial value
20704 TARGET_INDEXED_LOADS_DEFAULT.
20705 (mauto-modify-reg): Use initial value
20706 TARGET_AUTO_MODIFY_REG_DEFAULT.
20707 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20708 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20709 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
20710 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
20712 2017-07-17 Martin Liska <mliska@suse.cz>
20715 * opts.c (finish_options): Do not allow -fgnu-tm
20716 w/ -fsanitize={kernel-,}address. Say sorry.
20718 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20721 * tree-loop-distribution.c (classify_partition): Only assert on
20722 numer of iterations.
20723 (merge_dep_scc_partitions): Delete prameter. Update function call.
20724 (distribute_loop): Remove code handling loop with unknown niters.
20725 (pass_loop_distribution::execute): Skip loop with unknown niters.
20727 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20730 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
20731 function sort_partitions_by_post_order.
20733 2017-07-17 Bin Cheng <bin.cheng@arm.com>
20735 PR tree-optimization/81374
20736 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
20737 the max index of basic blocks, rather than number of basic blocks.
20739 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20741 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
20743 (arc_legitimate_pic_operand_p): Likewise.
20744 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
20746 (arc_needs_pcl_p): Likewise.
20747 (arc_legitimate_pc_offset_p): Likewise.
20748 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
20749 function is also used in constrains.md.
20750 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
20751 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
20752 PLUS. Only return true/false in known cases, otherwise assert.
20753 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
20754 is already called in arc_legitimate_constant_p.
20755 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
20757 (LEGITIMATE_PIC_OPERAND_P): Use
20758 arc_raw_symbolic_reference_mentioned_p function.
20759 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
20764 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
20765 Andrew Burgess <andrew.burgess@embecosm.com>
20767 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
20768 (arc_return_address_register): New function.
20769 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
20770 (arc_handle_fndecl_attribute): Add naked attribute.
20771 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20772 (TARGET_WARN_FUNC_RETURN): Likewise.
20773 (arc_allocate_stack_slots_for_args): New function.
20774 (arc_warn_func_return): Likewise.
20775 (machine_function): Change type fn_type.
20776 (arc_compute_function_type): Consider new naked function type,
20777 change function return type.
20778 (arc_must_save_register): Adapt to handle new
20779 arc_compute_function_type's return type.
20780 (arc_expand_prologue): Likewise.
20781 (arc_expand_epilogue): Likewise.
20782 (arc_return_address_regs): Delete.
20783 (arc_return_address_register): New function.
20784 (arc_epilogue_uses): Use above function.
20785 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
20786 (arc_function_type): Change encoding, add naked type.
20787 (ARC_INTERRUPT_P): Change to handle the new encoding.
20788 (ARC_FAST_INTERRUPT_P): Likewise.
20789 (ARC_NORMAL_P): Define.
20790 (ARC_NAKED_P): Likewise.
20791 (arc_compute_function_type): Delete prototype.
20792 * config/arc/arc.md (in_ret_delay_slot): Use
20793 arc_return_address_register function.
20794 (simple_return): Likewise.
20795 (p_return_i): Likewise.
20797 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20799 PR tree-optimization/81428
20800 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
20801 can't be built for those types.
20803 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20805 Remove stuff dead since r239246.
20807 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
20808 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
20809 (avr_inform_devices): Remove dead stuff.
20811 2017-07-17 Tamar Christina <tamar.christina@arm.com>
20813 * config/arm/arm_neon.h: Fix softp typo.
20815 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20817 PR tree-optimization/81365
20818 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
20819 aggregate moves onto bb predecessor edges, make sure there are no
20820 loads that could alias the lhs in between the start of bb and the
20823 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
20826 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
20827 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
20828 [LSHIFTRT, outer_code = TRUNCATE]: Same.
20830 2017-07-17 Jakub Jelinek <jakub@redhat.com>
20832 PR tree-optimization/81396
20833 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
20834 (init_symbolic_number): Initialize it to 1.
20835 (perform_symbolic_merge): Add n_ops from both operands into the new
20837 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
20838 without base_addr as useless if they need more than one operation.
20839 (bswap_replace): Handle !bswap case for NULL base_addr.
20841 2017-07-17 Tom de Vries <tom@codesourcery.com>
20844 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
20847 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
20849 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
20850 conditional builtin define __FIX_LEON3FT_B2BST.
20852 2017-07-17 Daniel Cederman <cederman@gaisler.com>
20854 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
20855 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
20858 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20860 PR rtl-optimization/81424
20861 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
20862 to remove potential trapping from operands if -fnon-call-exceptions.
20864 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20866 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
20867 profile_proability for scalling.
20868 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
20870 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20872 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
20874 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20876 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
20877 fixpoint arithmetics.
20879 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20881 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
20882 fixpoint arithmetics.
20884 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20886 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20887 fixpoint arithmetics.
20889 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20891 * profile-count.h (profile_probability::from_reg_br_prob_note,
20892 profile_probability::to_reg_br_prob_note): New functions.
20893 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20894 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20895 * predict.c (probability_reliable_p): Update.
20896 (edge_probability_reliable_p): Update.
20897 (br_prob_note_reliable_p): Update.
20898 (invert_br_probabilities): Update.
20899 (add_reg_br_prob_note): New function.
20900 (combine_predictions_for_insn): Update.
20901 * asan.c (asan_clear_shadow): Update.
20902 * cfgbuild.c (compute_outgoing_frequencies): Update.
20903 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20904 (update_br_prob_note): Update.
20905 (rtl_verify_edges): Update.
20906 (purge_dead_edges): Update.
20907 (fixup_reorder_chain): Update.
20908 * emit-rtl.c (try_split): Update.
20909 * ifcvt.c (cond_exec_process_insns): Update.
20910 (cond_exec_process_if_block): Update.
20911 (dead_or_predicable): Update.
20912 * internal-fn.c (expand_addsub_overflow): Update.
20913 (expand_neg_overflow): Update.
20914 (expand_mul_overflow): Update.
20915 * loop-doloop.c (doloop_modify): Update.
20916 * loop-unroll.c (compare_and_jump_seq): Update.
20917 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20918 * predict.h: Update.
20919 * reorg.c (mostly_true_jump): Update.
20921 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20922 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20923 * config/arc/arc.c: (emit_unlikely_jump): Update.
20924 * config/arm/arm.c: (emit_unlikely_jump): Update.
20925 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20926 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20927 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20928 (ix86_print_operand): Update.
20929 (ix86_split_fp_branch): Update.
20930 (predict_jump): Update.
20931 * config/ia64/ia64.c (ia64_print_operand): Update.
20932 * config/mmix/mmix.c (mmix_print_operand): Update.
20933 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20934 (rs6000_expand_split_stack_prologue): Update.
20935 * config/rs6000/rs6000.c: Update.
20936 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20937 (s390_expand_vec_movstr): Update.
20938 (s390_expand_cs_tdsi): Update.
20939 (s390_expand_split_stack_prologue): Update.
20940 * config/sh/sh.c (sh_print_operand): Update.
20941 (expand_cbranchsi4): Update.
20942 (expand_cbranchdi4): Update.
20943 * config/sparc/sparc.c (output_v9branch): Update.
20944 * config/spu/spu.c (get_branch_target): Update.
20945 (ea_load_store_inline): Update.
20946 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20947 * config/tilepro/tilepro.c: Update.
20949 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20951 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20952 (gimplify_save_expr): Likewise.
20954 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20956 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20958 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20960 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20962 * timevar.def (TV_IPA_FNSUMMARY): Define.
20964 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20966 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20967 to back store errata sensitive sequence from being generated.
20968 (sqrtdf2_fix): Likewise.
20970 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20972 * tree-ssa-threadupdate.c (compute_path_counts,
20973 update_joiner_offpath_counts): Use profile_probability.
20975 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20978 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20980 * config/arm/arm-c.c (arm_cpu_builtins): Define
20981 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20983 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20985 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20986 array entries to represent __ieee128 versions of the
20987 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20988 scalar_extract_sig, and scalar_insert_exp built-in functions.
20989 (altivec_resolve_overloaded_builtin): Add special case handling
20990 for the __builtin_scalar_insert_exp function, as represented by
20991 the P9V_BUILTIN_VEC_VSIEDP constant.
20992 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20993 exponent support for __ieee128 argument.
20994 (VSESQP): Add scalar extract signature support for __ieee128
20996 (VSTDCNQP): Add scalar test negative support for __ieee128
20998 (VSIEQP): Add scalar insert exponent support for __int128 argument
20999 with __ieee128 result.
21000 (VSIEQPF): Add scalar insert exponent support for __ieee128
21001 argument with __ieee128 result.
21002 (VSTDCQP): Add scalar test data class support for __ieee128
21004 (VSTDCNQP): Add overload support for scalar test negative with
21005 __ieee128 argument.
21006 (VSTDCQP): Add overload support for scalar test data class
21007 __ieee128 argument.
21008 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21009 UNSPEC_VSX_SXSIGDP.
21010 (UNSPEC_VSX_SIEXPQP): New constant.
21011 (xsxexpqp): New insn for VSX scalar extract exponent quad
21013 (xsxsigqp): New insn for VSX scalar extract significand quad
21015 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21016 precision with floating point argument.
21017 (xststdcqp): New expand for VSX scalar test data class quad
21019 (xststdcnegqp): New expand for VSX scalar test negative quad
21021 (xststdcqp): New insn to match expansions for VSX scalar test data
21022 class quad precision and VSX scalar test negative quad precision.
21023 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21024 special case operand checking to enforce that second operand of
21025 VSX scalar test data class with quad precision argument is a 7-bit
21027 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21028 prototypes and descriptions of __ieee128 versions of
21029 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21030 scalar_test_data_class, and scalar_test_neg built-in functions.
21032 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21034 PR tree-optimization/81162
21035 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21036 replace a negate with an add.
21038 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21040 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21042 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21044 * config/arm/arm-c.c (arm_cpu_builtins): Define
21045 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21047 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21049 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21050 (armv8-r): Set ARM Cortex-R52 as default CPU.
21051 * config/arm/arm-tables.opt: Regenerate.
21052 * config/arm/arm-tune.md: Regenerate.
21053 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21055 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21056 extension for -mcpu=cortex-r52.
21058 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21060 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21061 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21062 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21063 (fp-armv8): Define it as FP_ARMv8 only.
21064 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21065 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21067 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21068 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21069 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21070 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21071 than TARGET_FPU_ARMV8.
21072 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21073 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21074 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21076 * config/arm/neon.md (neon_vrint): Likewise.
21077 (neon_vcvt): Likewise.
21078 (neon_<fmaxmin_op><mode>): Likewise.
21079 (<fmaxmin><mode>3): Likewise.
21080 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21081 * config/arm/predicates.md (arm_cond_move_operator): Check against
21082 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21084 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21086 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21087 to top of function.
21089 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21091 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21092 loop in comment with memset.
21094 2017-07-14 Martin Liska <mliska@suse.cz>
21096 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21097 * dwarf2out.c (is_java): Remove the function.
21098 (output_pubname): Remove usage of the function.
21099 (lower_bound_default): Remove usage of DW_LANG_Java.
21100 (gen_compile_unit_die): Likewise.
21101 * gcc.c: Remove compiler defaults for .java and .zip files.
21102 * gimple-expr.c (remove_suffix): Change as there's no longer
21103 extension than 4-letter one.
21104 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21105 (gimplify_save_expr): Likewise.
21106 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21107 as it's possible even for other languages than Java.
21108 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21109 * lto-opts.c (lto_write_options): Remove reference to Java.
21110 * opts.c (strip_off_ending): Update file extension handling.
21111 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21112 * tree-eh.c (lower_resx): Likewise.
21113 * tree.c (free_lang_data_in_type): Remove dead code.
21114 (find_decls_types_r): Likewise.
21115 (build_common_builtin_nodes): Remove Java from a comment.
21116 (verify_type): Remove dead code.
21117 * varasm.c (assemble_external): Remove Java from a comment.
21119 2017-07-14 Martin Liska <mliska@suse.cz>
21121 * opts.c (finish_options): Add quotes.
21122 (common_handle_option): Likewise.
21124 2017-07-14 Martin Liska <mliska@suse.cz>
21126 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21127 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21128 Remove N_SO_PASCAL.
21129 * dwarf2out.c (lower_bound_default): Do not handle
21131 (gen_compile_unit_die): Likewise.
21132 * gcc.c: Remove default extension binding for GNU Pascal.
21133 * stmt.c: Remove Pascal language from a comment.
21134 * xcoffout.c: Likewise.
21136 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21139 * diagnostic-show-locus.c (fixit_cmp): New function.
21140 (layout::layout): Sort m_fixit_hints.
21141 (column_range::column_range): Assert that the values are valid.
21142 (struct char_span): New struct.
21143 (correction::overwrite): New method.
21144 (struct source_line): New struct.
21145 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21146 calls in terms of classes source_line and char_span, and
21147 correction::overwrite.
21148 (selftest::test_overlapped_fixit_printing_2): New function.
21149 (selftest::diagnostic_show_locus_c_tests): Call it.
21151 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21153 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21154 early if there is no lhs.
21156 2017-07-13 Martin Liska <mliska@suse.cz>
21158 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21159 (gen_reference_type_die): Likewise.
21160 * stor-layout.c: Remove Pascal-related comment.
21162 2017-07-13 Martin Liska <mliska@suse.cz>
21164 * opts.c (finish_options): Add quotes to error messages.
21165 (parse_sanitizer_options): Likewise.
21167 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21169 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21171 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21173 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21175 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21177 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21179 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21181 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21184 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21185 provides the hardware capability bits, define the macro
21186 __BUILTIN_CPU_SUPPORTS__.
21187 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21188 if GLIBC does not provide the hardware capability bits. Add a
21189 gcc_unreachable call if the built-in cpu function is neither
21190 __builtin_cpu_is nor __builtin_cpu_supports.
21191 (rs6000_get_function_versions_dispatcher): Change the warning
21192 that an old GLIBC is used which does not export the capability
21193 bits to be an error.
21194 * doc/extend.texi (target_clones attribute): Document the
21195 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21196 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21197 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21198 the macros defined by GCC if the newer GLIBC is available.
21200 2017-07-12 Jeff Law <law@redhat.com>
21202 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21203 remaining includes slightly.
21204 * config/riscv/riscv-builtins.c: Include profile-count.h.
21206 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21209 * config/avr/avr.c (avr_set_current_function): In diagnostic
21210 messages: Quote keywords and (parts of) identifiers.
21211 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21214 2017-07-12 Carl Love <cel@us.ibm.com>
21216 * config/rs6000/rs6000-c.c: Add support for built-in functions
21217 vector bool char vec_revb (vector bool char);
21218 vector bool short vec_revb (vector short char);
21219 vector bool int vec_revb (vector bool int);
21220 vector bool long long vec_revb (vector bool long long);
21221 * doc/extend.texi: Update the built-in documentation file for the
21222 new built-in functions.
21224 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21226 * config/s390/s390.md: Remove movcc splitter.
21228 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21230 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21231 load/store on condition.
21233 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21236 * config/avr/avr.c (avr_encode_section_info)
21237 [progmem && !TREE_READONLY]: Error if progmem object needs
21240 2017-07-11 Michael Collison <michael.collison@arm.com>
21242 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21245 2017-07-11 Carl Love <cel@us.ibm.com>
21247 * config/rs6000/rs6000-c.c: Add support for builtins
21248 vector unsigned int vec_parity_lsbb (vector signed int);
21249 vector unsigned int vec_parity_lsbb (vector unsigned int);
21250 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
21251 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
21252 vector unsigned long long vec_parity_lsbb (vector signed long long);
21253 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
21254 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
21255 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
21256 * doc/extend.texi: Update the built-in documentation file for the
21257 new built-in functions.
21259 2017-07-11 David Malcolm <dmalcolm@redhat.com>
21261 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
21262 (layout::m_primary_loc): New field.
21263 (layout::layout): Initialize new field. Move location filtering
21264 logic from here to...
21265 (layout::maybe_add_location_range): ...this new method. Add
21266 support for filtering to just the lines already specified by other
21268 (layout::will_show_line_p): New method.
21269 (gcc_rich_location::add_location_if_nearby): New method.
21270 (selftest::test_add_location_if_nearby): New test function.
21271 (selftest::diagnostic_show_locus_c_tests): Call it.
21272 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
21275 2017-07-11 Tom de Vries <tom@codesourcery.com>
21277 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
21278 (bb_first_real_insn): New function.
21279 (nvptx_single): Add extra initialization of broadcasted condition
21282 2017-07-11 Nathan Sidwell <nathan@acm.org>
21284 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
21286 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
21288 * doc/extend.texi (AVR Function Attributes): Remove weblink to
21289 Binutils doc as TEXI will mess them up.
21290 * doc/invoke.texi (AVR Options): Same here.
21292 2017-07-11 Daniel Cederman <cederman@gaisler.com>
21294 * config/sparc/sparc.opt (mfix-ut700): New option.
21295 (mfix-gr712rc): Likewise.
21296 (sparc_fix_b2bst): New variable.
21297 * doc/invoke.texi (SPARC options): Document them.
21298 (ARM options): Fix warnings.
21299 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
21300 instructions to prevent sequences that can trigger the store-store
21301 errata for certain LEON3FT processors.
21302 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
21303 (sparc_option_override): Set sparc_fix_b2bst appropriately.
21304 * config/sparc/sparc.md (fix_b2bst): New attribute.
21305 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
21307 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
21310 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
21312 (*rsqrtsf2_sse): Ditto.
21314 (div<mode>3): Macroize insn from divdf3 and divsf3
21315 using MODEF mode iterator.
21317 2017-07-10 Martin Sebor <msebor@redhat.com>
21319 PR tree-optimization/80397
21320 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
21321 instead of testing for equality to INTEGER_TYPE.
21323 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
21325 * config.gcc: Remove uclibc from arc target spec.
21327 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
21329 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
21331 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21334 * lto-wrapper.c (remove_option): New function.
21335 (merge_and_complain): Merge PIC/PIE options more realistically.
21337 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21339 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
21343 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
21344 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
21345 * config.in: Regenerate.
21346 * configure: Regenerate.
21347 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
21348 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
21349 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
21350 (TARGET_GASISR_PROLOGUES): ...target mask.
21351 * common/config/avr/avr-common.c
21352 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
21353 Set -mgas-isr-prologues.
21354 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
21355 INSERT_PASS_BEFORE for it.
21356 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
21357 * config/avr/avr.c (avr_option_override)
21358 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
21359 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
21360 (avr_attribute_table) <no_gccisr>: Add new function attribute.
21361 (avr_set_current_function) <is_no_gccisr>: Init machine field.
21362 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
21364 (make_avr_pass_pre_proep): New function.
21365 (emit_push_sfr) <treg>: Add argument to function and use it
21366 instead of TMP_REG.
21367 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
21368 and set machine->gasisr.yes.
21369 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
21370 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
21371 __gcc_isr.n_pushed to .L__stack_usage.
21372 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
21373 (avr_asm_final_postscan_insn): ...this new static function.
21374 * config/avr/avr.h (machine_function)
21375 <is_no_gccisr, use_L__stack_usage>: New fields.
21376 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
21377 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
21378 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
21379 (gasisr, *gasisr): New expander and insn.
21380 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21381 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
21382 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
21384 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
21386 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
21389 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
21391 Move jump-tables out of .text again.
21394 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
21395 (ASM_OUTPUT_ADDR_VEC): New function.
21396 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
21397 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
21398 INSN_ADDRESSes as asm comment.
21399 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
21400 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
21401 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
21402 * config/avr/avr.md (*tablejump): Adjust comment.
21403 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
21404 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
21406 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
21407 (avr_output_addr_vec): New proto.
21408 (avr_log_t) <insn_addresses>: New field.
21410 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
21413 * config/i386/i386.c (ix86_function_arg_advance): Set
21414 outgoing_args_on_stack to true if there are outgoing arguments
21416 (ix86_function_arg): Likewise.
21417 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
21418 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
21419 * config/i386/i386.h (machine_function): Add
21420 outgoing_args_on_stack.
21422 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
21424 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21425 supporting pthreds.
21426 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
21428 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
21430 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
21431 (REAL_H): Remove $(MACHMODE_H).
21432 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
21434 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
21435 $(MACHMODE_H) and double-int.h.
21436 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
21438 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
21441 2017-07-07 Andrew Pinski <apinski@cavium.com>
21443 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
21444 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
21446 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21448 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
21449 Add warning if GCC was not configured to link against a GLIBC that
21450 exports the hardware capability bits.
21451 (make_resolver_func): Make resolver function private and not a
21452 COMDAT function. Create the name with clone_function_name instead
21453 of make_unique_name.
21456 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
21457 correct operand in doing the split.
21459 2017-07-07 Carl Love <cel@us.ibm.com>
21461 * config/rs6000/rs6000-c: Add support for built-in function
21462 vector unsigned short vec_pack_to_short_fp32 (vector float,
21464 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
21465 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
21466 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
21467 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
21468 (convert_4f32_8i16): Add define_expand.
21469 * doc/extend.texi: Update the built-in documentation file for the
21470 new built-in function.
21472 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21474 * config/sparc/m8.md: New file.
21475 * config/sparc/sparc.md: Include m8.md.
21477 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21479 * config/sparc/sparc.opt: New option -mvis4b.
21480 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
21481 (sparc_option_override): Handle VIS4B.
21482 (enum sparc_builtins): Define
21483 SPARC_BUILTIN_DICTUNPACK{8,16,32},
21484 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
21485 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
21486 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
21487 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
21488 (check_constant_argument): New function.
21489 (sparc_vis_init_builtins): Define builtins
21490 __builtin_vis_dictunpack{8,16,32},
21491 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
21492 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
21493 __builtin_vis_fpcmpde{8,16,32}shl and
21494 __builtin_vis_fpcmpur{8,16,32}shl.
21495 (sparc_expand_builtin): Check that the constant operands to
21496 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
21497 constant and in range.
21498 * config/sparc/sparc-c.c (sparc_target_macros): Handle
21500 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
21501 (SPARC_IMM5_P): Likewise.
21502 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
21503 (enabled): Handle vis4b.
21504 (UNSPEC_DICTUNPACK): New unspec.
21505 (UNSPEC_FPCMPSHL): Likewise.
21506 (UNSPEC_FPUCMPSHL): Likewise.
21507 (UNSPEC_FPCMPDESHL): Likewise.
21508 (UNSPEC_FPCMPURSHL): Likewise.
21509 (cpu_feature): New CPU feature `vis4b'.
21510 (dictunpack{8,16,32}): New insns.
21511 (FPCSMODE): New mode iterator.
21512 (fpcscond): New code iterator.
21513 (fpcsucond): Likewise.
21514 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
21515 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
21516 (fpcmpde{8,16,32}{si,di}shl): Likewise.
21517 (fpcmpur{8,16,32}{si,di}shl): Likewise.
21518 * config/sparc/constraints.md: Define constraints `q' for unsigned
21519 2-bit integer constants and `t' for unsigned 5-bit integer
21521 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
21523 (imm5_operand_dictunpack16): Likewise.
21524 (imm5_operand_dictunpack32): Likewise.
21525 (imm2_operand): Likewise.
21526 * doc/invoke.texi (SPARC Options): Document -mvis4b.
21527 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
21528 ditunpack* and fpcmp*shl builtins.
21530 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21532 * config.gcc: Handle m8 in --with-{cpu,tune} options.
21533 * config.in: Add HAVE_AS_SPARC6 define.
21534 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
21536 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
21538 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
21539 (CPP_CPU_SPEC): Handle m8.
21540 (ASM_CPU_SPEC): Likewise.
21541 * config/sparc/sparc-opts.h (enum processor_type): Add
21543 * config/sparc/sparc.c (m8_costs): New struct.
21544 (sparc_option_override): Handle TARGET_CPU_m8.
21545 (sparc32_initialize_trampoline): Likewise.
21546 (sparc64_initialize_trampoline): Likewise.
21547 (sparc_issue_rate): Likewise.
21548 (sparc_register_move_cost): Likewise.
21549 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
21550 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
21551 (ASM_CPU64_DEFAULT_SPEC): Likewise.
21552 (CPP_CPU_SPEC): Handle M8.
21553 (ASM_CPU_SPEC): Likewise.
21554 (AS_M8_FLAG): Define.
21555 * config/sparc/sparc.md: Add m8 to the cpu attribute.
21556 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
21557 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
21559 * configure: Regenerate.
21560 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
21563 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21565 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
21567 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
21568 ("*movdi_insn_sp32"): Do not set v3pipe.
21569 ("*movsi_insn"): Likewise.
21570 ("*movdi_insn_sp64"): Likewise.
21571 ("*movsf_insn"): Likewise.
21572 ("*movdf_insn_sp32"): Likewise.
21573 ("*movdf_insn_sp64"): Likewise.
21574 ("*zero_extendsidi2_insn_sp64"): Likewise.
21575 ("*sign_extendsidi2_insn"): Likewise.
21576 ("*mov<VM32:mode>_insn"): Likewise.
21577 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21578 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21579 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21580 ("<vlop:code><VL:mode>3"): Likewise.
21581 ("*not_<vlop:code><VL:mode>3"): Likewise.
21582 ("*nand<VL:mode>_vis"): Likewise.
21583 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
21584 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
21585 ("one_cmpl<VL:mode>2"): Likewise.
21586 ("faligndata<VM64:mode>_vis"): Likewise.
21587 ("alignaddrsi_vis"): Likewise.
21588 ("alignaddrdi_vis"): Likweise.
21589 ("alignaddrlsi_vis"): Likewise.
21590 ("alignaddrldi_vis"): Likewise.
21591 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21592 ("bmaskdi_vis"): Likewise.
21593 ("bmasksi_vis"): Likewise.
21594 ("bshuffle<VM64:mode>_vis"): Likewise.
21595 ("cmask8<P:mode>_vis"): Likewise.
21596 ("cmask16<P:mode>_vis"): Likewise.
21597 ("cmask32<P:mode>_vis"): Likewise.
21598 ("pdistn<P:mode>_vis"): Likewise.
21599 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21601 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21603 * config/sparc/sparc.md ("subtype"): New insn attribute.
21604 ("*wrgsr_sp64"): Set insn subtype.
21605 ("*rdgsr_sp64"): Likewise.
21606 ("alignaddrsi_vis"): Likewise.
21607 ("alignaddrdi_vis"): Likewise.
21608 ("alignaddrlsi_vis"): Likewise.
21609 ("alignaddrldi_vis"): Likewise.
21610 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
21611 ("fexpand_vis"): Likewise.
21612 ("fpmerge_vis"): Likewise.
21613 ("faligndata<VM64:mode>_vis"): Likewise.
21614 ("bshuffle<VM64:mode>_vis"): Likewise.
21615 ("cmask8<P:mode>_vis"): Likewise.
21616 ("cmask16<P:mode>_vis"): Likewise.
21617 ("cmask32<P:mode>_vis"): Likewise.
21618 ("fchksm16_vis"): Likewise.
21619 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
21620 ("fmean16_vis"): Likewise.
21621 ("fp<plusminus_insn>64_vis"): Likewise.
21622 ("<plusminus_insn>v8qi3"): Likewise.
21623 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
21624 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
21625 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
21626 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
21627 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
21628 ("*movqi_insn"): Likewise.
21629 ("*movhi_insn"): Likewise.
21630 ("*movsi_insn"): Likewise.
21631 ("movsi_pic_gotdata_op"): Likewise.
21632 ("*movdi_insn_sp32"): Likewise.
21633 ("*movdi_insn_sp64"): Likewise.
21634 ("movdi_pic_gotdata_op"): Likewise.
21635 ("*movsf_insn"): Likewise.
21636 ("*movdf_insn_sp32"): Likewise.
21637 ("*movdf_insn_sp64"): Likewise.
21638 ("*zero_extendhisi2_insn"): Likewise.
21639 ("*zero_extendqihi2_insn"): Likewise.
21640 ("*zero_extendqisi2_insn"): Likewise.
21641 ("*zero_extendqidi2_insn"): Likewise.
21642 ("*zero_extendhidi2_insn"): Likewise.
21643 ("*zero_extendsidi2_insn_sp64"): Likewise.
21644 ("ldfsr"): Likewise.
21645 ("prefetch_64"): Likewise.
21646 ("prefetch_32"): Likewise.
21647 ("tie_ld32"): Likewise.
21648 ("tie_ld64"): Likewise.
21649 ("*tldo_ldub_sp32"): Likewise.
21650 ("*tldo_ldub1_sp32"): Likewise.
21651 ("*tldo_ldub2_sp32"): Likewise.
21652 ("*tldo_ldub_sp64"): Likewise.
21653 ("*tldo_ldub1_sp64"): Likewise.
21654 ("*tldo_ldub2_sp64"): Likewise.
21655 ("*tldo_ldub3_sp64"): Likewise.
21656 ("*tldo_lduh_sp32"): Likewise.
21657 ("*tldo_lduh1_sp32"): Likewise.
21658 ("*tldo_lduh_sp64"): Likewise.
21659 ("*tldo_lduh1_sp64"): Likewise.
21660 ("*tldo_lduh2_sp64"): Likewise.
21661 ("*tldo_lduw_sp32"): Likewise.
21662 ("*tldo_lduw_sp64"): Likewise.
21663 ("*tldo_lduw1_sp64"): Likewise.
21664 ("*tldo_ldx_sp64"): Likewise.
21665 ("*mov<VM32:mode>_insn"): Likewise.
21666 ("*mov<VM64:mode>_insn_sp64"): Likewise.
21667 ("*mov<VM64:mode>_insn_sp32"): Likewise.
21669 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21671 * config/sparc/sparc.md ("type"): New insn type viscmp.
21672 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
21674 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
21675 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
21676 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
21677 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
21679 ("n7_vis_logical_11cycle"): Likewise.
21680 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
21681 * config/sparc/niagara2.md ("niag3_vis": Likewise.
21682 * config/sparc/niagara.md ("niag_vis"): Likewise.
21683 * config/sparc/ultra3.md ("us3_fga"): Likewise.
21684 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
21686 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
21688 * config/sparc/sparc.md: New instruction type `bmask'.
21689 (bmaskdi_vis): Use the `bmask' type.
21690 (bmasksi_vis): Likewise.
21691 * config/sparc/ultra3.md (us3_array): Likewise.
21692 * config/sparc/niagara7.md (n7_array): Likewise.
21693 * config/sparc/niagara4.md (n4_array): Likewise.
21694 * config/sparc/niagara2.md (niag2_vis): Likewise.
21695 (niag3_vis): Likewise.
21696 * config/sparc/niagara.md (niag_vis): Likewise.
21698 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21700 * ipa-comdats.c: Remove optimize check from gate.
21701 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
21702 for functions not optimized.
21703 (ipa_fn_summary_read): Skip optimize check.
21704 (ipa_fn_summary_write): Likewise.
21705 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
21707 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
21709 (can_inline_edge_p): Check flag_pcc_struct_return for match.
21710 (check_callers): Give up on caller which is not optimized.
21711 (inline_small_functions): Likewise.
21712 (ipa_inline): Do not give up when not optimizing.
21713 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
21714 away unoptimizes cdtors.
21715 (whole_program_function_and_variable_visibility): Do
21716 ipa_discover_readonly_nonaddressable_vars in LTO mode.
21717 * ipa.c (process_references): Do not check optimize.
21718 (symbol_table::remove_unreachable_nodes): Update optimize check.
21719 (set_writeonly_bit): Update optimize check.
21720 (pass_ipa_cdtor_merge::gate): Do not check optimize.
21721 (pass_ipa_single_use::gate): Remove.
21723 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21725 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
21726 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
21727 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
21728 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
21729 permute_load, permute_store, adjust_extract, adjust_splat,
21730 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
21731 replace_swap_with_copy, dump_swap_insn_table,
21732 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
21733 recombine_lvx_pattern, recombine_stvx_pattern,
21734 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
21735 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
21736 to file rs6000-p8swap.c.
21737 * config/rs6000/rs6000-p8swap.c: New file.
21738 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
21739 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
21740 and rs6000*-*-* targets.
21742 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21744 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
21746 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21748 * lto-wrapper.c (merge_and_complain): Do not merge
21749 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
21750 fsigned_zeros, ftrapping_math, fwrapv.
21751 (append_compiler_options): Do not track these options.
21752 (append_linker_options): Likewie
21754 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21756 * cgraphunit.c (cgraph_node::finalize_function): When
21757 !flag_toplevel_reorde set no_reorder flag.
21758 (varpool_node::finalize_decl): Likewise.
21759 (symbol_table::compile): Drop no toplevel reorder path.
21761 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21763 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
21764 edges; zero probability is not better than uninitialized.
21766 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21768 * asan.h (asan_sanitize_allocas_p): Declare.
21769 * asan.c (asan_sanitize_allocas_p): New function.
21770 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
21771 (handle_builtin_alloca): Likewise.
21772 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
21773 if !asan_sanitize_allocas_p.
21774 * params.def (asan-instrument-allocas): Add new option.
21775 * params.h (ASAN_PROTECT_ALLOCAS): Define.
21776 * opts.c (common_handle_option): Disable allocas sanitization for
21779 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
21781 * asan.c: Include gimple-fold.h.
21782 (get_last_alloca_addr): New function.
21783 (handle_builtin_stackrestore): Likewise.
21784 (handle_builtin_alloca): Likewise.
21785 (asan_emit_allocas_unpoison): Likewise.
21786 (get_mem_refs_of_builtin_call): Add new parameter, remove const
21787 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
21788 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
21789 (instrument_builtin_call): Pass gimple iterator to
21790 get_mem_refs_of_builtin_call.
21791 (last_alloca_addr): New global.
21792 * asan.h (asan_emit_allocas_unpoison): Declare.
21793 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
21794 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
21795 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
21796 if function calls alloca.
21797 * gimple-fold.c (replace_call_with_value): Remove static keyword.
21798 * gimple-fold.h (replace_call_with_value): Declare.
21799 * internal-fn.c: Include asan.h.
21800 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
21801 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
21803 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21805 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
21806 (C_SELFTEST_FLAGS): New.
21807 (CPP_SELFTEST_FLAGS): New.
21808 (SELFTEST_DEPS): New, from deps of s-selftest.
21809 (C_SELFTEST_DEPS): New, from deps of s-selftest.
21810 (CPP_SELFTEST_DEPS): New.
21811 (selftest): Add dependency on s-selftest-c++.
21812 (s-selftest): Rename to...
21813 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
21814 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
21815 than SELFTEST_FLAGS.
21816 (selftest-gdb): Rename to...
21817 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
21819 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
21820 (selftest-valgrind): Rename to...
21821 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
21823 (selftest-valgrind): Reintroduce as an alias for
21824 selftest-c-valgrind.
21825 (s-selftest-c++): New.
21826 (selftest-c++-gdb): New.
21827 (selftest-c++-valgrind): New.
21829 2017-07-06 Olivier Hainque <hainque@adacore.com>
21831 * gcc.c (process_command): When deciding if undefined variables
21832 should be ignored when processing specs, accept "gcc -v" as well.
21834 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
21836 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
21837 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
21839 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21841 * config/arm/arm-cpus.in (armv8-r): Add new entry.
21842 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
21843 * config/arm/arm-tables.opt: Regenerate.
21844 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
21846 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
21848 2017-07-06 Carl Love <cel@us.ibm.com>
21850 * ChangeLog: Clean up from mid air collision
21852 2017-07-06 Carl Love <cel@us.ibm.com>
21854 * config/rs6000/rs6000-c.c: Add support for built-in functions
21855 vector signed int vec_subc (vector signed int, vector signed int);
21856 vector signed __int128 vec_subc (vector signed __int128,
21857 vector signed __int128);
21858 vector unsigned __int128 vec_subc (vector unsigned __int128,
21859 vector unsigned __int128);
21860 vector signed int vec_sube (vector signed int, vector signed int,
21861 vector signed int);
21862 vector unsigned int vec_sube (vector unsigned int,
21863 vector unsigned int,
21864 vector unsigned int);
21865 vector signed __int128 vec_sube (vector signed __int128,
21866 vector signed __int128,
21867 vector signed__int128);
21868 vector unsigned __int128 vec_sube (vector unsigned __int128,
21869 vector unsigned __int128,
21870 vector unsigned __int128);
21871 vector signed int vec_subec (vector signed int, vector signed int,
21872 vector signed int);
21873 vector unsigned int vec_subec (vector unsigned int,
21874 vector unsigned int,
21875 vector unsigned int);
21876 vector signed __int128 vec_subec (vector signed __int128,
21877 vector signed __int128,
21878 vector signed__int128);
21879 vector unsigned __int128 vec_subec (vector unsigned __int128,
21880 vector unsigned __int128,
21881 vector unsigned __int128);
21882 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
21883 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
21884 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
21885 BU_ALTIVEC_OVERLOAD_X definitions.
21886 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21887 * doc/extend.texi: Update the built-in documentation file for the new
21888 built-in functions.
21890 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21893 * diagnostic-show-locus.c (layout::layout): Use start and finish
21894 spelling location for the start and finish of each range.
21895 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21896 Add unused aspect param.
21897 * input.c (expand_location_1): Add "aspect" param, and use it
21898 to access the correct part of the location.
21899 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21901 (expand_location_to_spelling_point): Likewise.
21902 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21903 param, and pass it to expand_location_1.
21905 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21907 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21908 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21909 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21910 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21911 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21912 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21913 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21914 _mm_maskz_getmant_ss): New intrinsics.
21915 (__builtin_ia32_getexpss128_mask): Changed to ...
21916 __builtin_ia32_getexpss128_round ... this.
21917 (__builtin_ia32_getexpsd128_mask): Changed to ...
21918 __builtin_ia32_getexpsd128_round ... this.
21919 * config/i386/i386-builtin-types.def
21920 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21921 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21922 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21923 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21924 __builtin_ia32_getmantss_mask_round): New builtins.
21925 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21926 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21927 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21928 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21929 * config/i386/sse.md
21930 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21931 avx512f_sgetexp<mode><mask_scalar_name>
21932 <round_saeonly_scalar_name> ... this.
21933 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21934 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21935 vgetexp<ssescalarmodesuffix>
21936 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21937 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21938 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21939 avx512f_vgetmant<mode><mask_scalar_name>
21940 <round_saeonly_scalar_name> ... this.
21941 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21942 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21943 vgetmant<ssescalarmodesuffix>
21944 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21945 %0<mask_scalar_operand4>, %1, %2
21946 <round_saeonly_scalar_mask_op4>, %3} ... this.
21947 * config/i386/subst.md (mask_scalar_operand4,
21948 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21949 round_saeonly_scalar_nimm_predicate): New subst attributes.
21951 2017-07-06 Julia Koval <julia.koval@intel.com>
21953 * config/i386/i386.c (ix86_erase_embedded_rounding):
21954 Remove code for old rounding pattern.
21956 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21958 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21960 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21962 * doc/sourcebuild.texi (Test Directives, Variants of
21963 dg-require-support): Add documentation for dg-require-stack-check.
21965 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21967 * config/i386/subst.md (mask_scalar, round_scalar,
21968 round_saeonly_scalar): New meta-templates.
21969 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21970 round_scalar_mask_operand3, round_scalar_mask_op3,
21971 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21972 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21973 round_saeonly_scalar_constraint,
21974 round_saeonly_scalar_prefix): New subst attribute.
21975 * config/i386/sse.md
21976 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21977 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21978 <round_scalar_name> ... this.
21979 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21980 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21981 <round_scalar_name> ... this.
21982 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21983 <sse>_vm<code><mode>3<mask_scalar_name>
21984 <round_saeonly_scalar_name> ... this.
21985 (v<plusminus_mnemonic><ssescalarmodesuffix>
21986 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21987 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21988 v<plusminus_mnemonic><ssescalarmodesuffix>
21989 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21990 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21991 (v<multdiv_mnemonic><ssescalarmodesuffix>
21992 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21993 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21994 v<multdiv_mnemonic><ssescalarmodesuffix>
21995 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21996 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21997 (v<maxmin_float><ssescalarmodesuffix>
21998 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21999 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22000 v<maxmin_float><ssescalarmodesuffix>
22001 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22002 %0<mask_scalar_operand3>, %1, %<iptr>2
22003 <round_saeonly_scalar_mask_op3>} ... this.
22005 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22007 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22008 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22010 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22011 Alan Hayward <alan.hayward@arm.com>
22012 David Sherwood <david.sherwood@arm.com>
22014 * combine.c (simplify_if_then_else): Remove "enum" before
22016 * compare-elim.c (can_eliminate_compare): Likewise.
22017 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22019 (aarch64_lookup_simd_builtin_type): Likewise.
22020 (aarch64_simd_builtin_type): Likewise.
22021 (aarch64_init_simd_builtin_types): Likewise.
22022 (aarch64_simd_expand_args): Likewise.
22023 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22025 (aarch64_reverse_mask): Likewise.
22026 (aarch64_simd_emit_reg_reg_move): Likewise.
22027 (aarch64_gen_adjusted_ldpstp): Likewise.
22028 (aarch64_ccmp_mode_to_code): Likewise.
22029 (aarch64_operands_ok_for_ldpstp): Likewise.
22030 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22031 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22033 (aarch64_min_divisions_for_recip_mul): Likewise.
22034 (aarch64_reassociation_width): Likewise.
22035 (aarch64_get_condition_code_1): Likewise.
22036 (aarch64_simd_emit_reg_reg_move): Likewise.
22037 (aarch64_simd_attr_length_rglist): Likewise.
22038 (aarch64_reverse_mask): Likewise.
22039 (aarch64_operands_ok_for_ldpstp): Likewise.
22040 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22041 (aarch64_gen_adjusted_ldpstp): Likewise.
22042 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22044 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22045 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22046 (arm_lookup_simd_builtin_type): Likewise.
22047 (arm_simd_builtin_type): Likewise.
22048 (arm_init_simd_builtin_types): Likewise.
22049 (arm_expand_builtin_args): Likewise.
22050 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22051 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22052 (ft32_setup_incoming_varargs): Likewise.
22053 (ft32_function_arg): Likewise.
22054 (ft32_function_arg_advance): Likewise.
22055 (ft32_pass_by_reference): Likewise.
22056 (ft32_arg_partial_bytes): Likewise.
22057 (ft32_valid_pointer_mode): Likewise.
22058 (ft32_addr_space_pointer_mode): Likewise.
22059 (ft32_addr_space_legitimate_address_p): Likewise.
22060 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22062 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22063 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22064 (iamcu_alignment): Likewise.
22065 (canonicalize_vector_int_perm): Likewise.
22066 (ix86_noce_conversion_profitable_p): Likewise.
22067 (ix86_mpx_bound_mode): Likewise.
22068 (ix86_operands_ok_for_move_multiple): Likewise.
22069 * config/microblaze/microblaze-protos.h
22070 (microblaze_expand_conditional_branch_reg): Likewise.
22071 * config/microblaze/microblaze.c
22072 (microblaze_expand_conditional_branch_reg): Likewise.
22073 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22075 (rs6000_reassociation_width): Likewise.
22076 (rs6000_invalid_binary_op): Likewise.
22077 (fusion_p9_p): Likewise.
22078 (emit_fusion_p9_load): Likewise.
22079 (emit_fusion_p9_store): Likewise.
22080 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22082 (riscv_hard_regno_mode_ok_p): Likewise.
22083 (riscv_address_insns): Likewise.
22084 (riscv_split_symbol): Likewise.
22085 (riscv_legitimize_move): Likewise.
22086 (riscv_function_value): Likewise.
22087 (riscv_hard_regno_nregs): Likewise.
22088 (riscv_expand_builtin): Likewise.
22089 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22090 (riscv_build_integer): Likewise.
22091 (riscv_split_integer): Likewise.
22092 (riscv_legitimate_constant_p): Likewise.
22093 (riscv_cannot_force_const_mem): Likewise.
22094 (riscv_regno_mode_ok_for_base_p): Likewise.
22095 (riscv_valid_base_register_p): Likewise.
22096 (riscv_valid_offset_p): Likewise.
22097 (riscv_valid_lo_sum_p): Likewise.
22098 (riscv_classify_address): Likewise.
22099 (riscv_legitimate_address_p): Likewise.
22100 (riscv_address_insns): Likewise.
22101 (riscv_load_store_insns): Likewise.
22102 (riscv_force_binary): Likewise.
22103 (riscv_split_symbol): Likewise.
22104 (riscv_force_address): Likewise.
22105 (riscv_legitimize_address): Likewise.
22106 (riscv_move_integer): Likewise.
22107 (riscv_legitimize_const_move): Likewise.
22108 (riscv_legitimize_move): Likewise.
22109 (riscv_address_cost): Likewise.
22110 (riscv_subword): Likewise.
22111 (riscv_output_move): Likewise.
22112 (riscv_canonicalize_int_order_test): Likewise.
22113 (riscv_emit_int_order_test): Likewise.
22114 (riscv_function_arg_boundary): Likewise.
22115 (riscv_pass_mode_in_fpr_p): Likewise.
22116 (riscv_pass_fpr_single): Likewise.
22117 (riscv_pass_fpr_pair): Likewise.
22118 (riscv_get_arg_info): Likewise.
22119 (riscv_function_arg): Likewise.
22120 (riscv_function_arg_advance): Likewise.
22121 (riscv_arg_partial_bytes): Likewise.
22122 (riscv_function_value): Likewise.
22123 (riscv_pass_by_reference): Likewise.
22124 (riscv_setup_incoming_varargs): Likewise.
22125 (riscv_print_operand): Likewise.
22126 (riscv_elf_select_rtx_section): Likewise.
22127 (riscv_save_restore_reg): Likewise.
22128 (riscv_for_each_saved_reg): Likewise.
22129 (riscv_register_move_cost): Likewise.
22130 (riscv_hard_regno_mode_ok_p): Likewise.
22131 (riscv_hard_regno_nregs): Likewise.
22132 (riscv_class_max_nregs): Likewise.
22133 (riscv_memory_move_cost): Likewise.
22134 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22135 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22136 (rl78_addr_space_address_mode): Likewise.
22137 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22139 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22140 (rs6000_reassociation_width): Likewise.
22141 (rs6000_invalid_binary_op): Likewise.
22142 (fusion_p9_p): Likewise.
22143 (emit_fusion_p9_load): Likewise.
22144 (emit_fusion_p9_store): Likewise.
22145 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22146 (ok_for_simple_move_operands): Likewise.
22147 (ok_for_simple_move_strict_operands): Likewise.
22148 (ok_for_simple_arith_logic_operands): Likewise.
22149 (visium_legitimize_reload_address): Likewise.
22150 (visium_select_cc_mode): Likewise.
22151 (output_cbranch): Likewise.
22152 (visium_split_double_move): Likewise.
22153 (visium_expand_copysign): Likewise.
22154 (visium_expand_int_cstore): Likewise.
22155 (visium_expand_fp_cstore): Likewise.
22156 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22157 (visium_function_arg): Likewise.
22158 (visium_function_arg_advance): Likewise.
22159 (visium_libcall_value): Likewise.
22160 (visium_setup_incoming_varargs): Likewise.
22161 (visium_legitimate_constant_p): Likewise.
22162 (visium_legitimate_address_p): Likewise.
22163 (visium_legitimize_address): Likewise.
22164 (visium_secondary_reload): Likewise.
22165 (visium_register_move_cost): Likewise.
22166 (visium_memory_move_cost): Likewise.
22167 (prepare_move_operands): Likewise.
22168 (ok_for_simple_move_operands): Likewise.
22169 (ok_for_simple_move_strict_operands): Likewise.
22170 (ok_for_simple_arith_logic_operands): Likewise.
22171 (visium_function_value_1): Likewise.
22172 (rtx_ok_for_offset_p): Likewise.
22173 (visium_legitimize_reload_address): Likewise.
22174 (visium_split_double_move): Likewise.
22175 (visium_expand_copysign): Likewise.
22176 (visium_expand_int_cstore): Likewise.
22177 (visium_expand_fp_cstore): Likewise.
22178 (visium_split_cstore): Likewise.
22179 (visium_select_cc_mode): Likewise.
22180 (visium_split_cbranch): Likewise.
22181 (output_cbranch): Likewise.
22182 (visium_print_operand_address): Likewise.
22183 * expmed.c (flip_storage_order): Likewise.
22184 * expmed.h (emit_cstore): Likewise.
22185 (flip_storage_order): Likewise.
22186 * genrecog.c (validate_pattern): Likewise.
22187 * hsa-gen.c (gen_hsa_addr): Likewise.
22188 * internal-fn.c (expand_arith_overflow): Likewise.
22189 * ira-color.c (allocno_copy_cost_saving): Likewise.
22190 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22191 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22192 (process_invariant_for_inheritance): Likewise.
22193 * lra-eliminations.c (move_plus_up): Likewise.
22194 * omp-low.c (lower_oacc_reductions): Likewise.
22195 * simplify-rtx.c (simplify_subreg): Likewise.
22196 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22197 (TARGET_CHKP_BOUND_MODE): Likewise..
22198 * targhooks.c (default_chkp_bound_mode): Likewise.
22199 (default_setup_incoming_vararg_bounds): Likewise.
22200 * targhooks.h (default_chkp_bound_mode): Likewise.
22201 (default_setup_incoming_vararg_bounds): Likewise.
22202 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22203 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22204 (have_whole_vector_shift): Likewise.
22205 * tree-vect-stmts.c (vectorizable_load): Likewise.
22206 * doc/tm.texi: Regenerate.
22208 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22210 Graceful degrade if Binutils PR21472 is not available.
22213 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22214 .rodata in flash test fails.
22215 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22216 * confgure: Regenerate.
22217 * config.in: Regenerate.
22218 * config/avr/avr.c (avr_asm_named_section)
22219 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22220 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22221 (avr_asm_init_sections): Same.
22223 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22225 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22226 (fma<VH:mode>4_intrinsic): Likewise.
22227 (*fmsub<VCVTF:mode>4): Likewise.
22228 (*fmsub<VH:mode>4_intrinsic): Likewise.
22230 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22233 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22234 Don't depend on "optimize > 0".
22235 (out_movhi_r_mr, out_movqi_mr_r): Same.
22236 (out_movhi_mr_r, out_movqi_r_mr): Same.
22237 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22238 io_address_operand on "optimize > 0".
22240 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22242 * tree-loop-distribution.c: Add general explanantion on the pass.
22243 (generate_loops_for_partition): Mark distributed loop.
22244 (pg_add_dependence_edges): New parameter. Handle alias data
22245 dependence specially and record it in the parameter if asked.
22246 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22247 (init_partition_graph_vertices, add_partition_graph_edge): New.
22248 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22249 (free_partition_graph_vdata, build_partition_graph): New.
22250 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
22251 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
22252 (data_ref_segment_size, latch_dominated_by_data_ref): New.
22253 (compute_alias_check_pairs, version_loop_by_alias_check): New.
22254 (version_for_distribution_p, finalize_partitions): New.
22255 (distribute_loop): Handle alias data dependence specially. Factor
22256 out loop fusion code as functions and call these functions.
22258 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22260 * tree-loop-distribution.c (classify_partition): New parameter and
22261 better handle reduction statement.
22262 (rdg_build_partitions): Revise comment.
22263 (distribute_loop): Compute statements in all partitions and pass it
22264 to classify_partition.
22266 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22268 * tree-loop-distribution.c (enum partition_type): New.
22269 (struct partition): New field type.
22270 (partition_merge_into): Add parameter. Update partition type.
22271 (data_dep_in_cycle_p, update_type_for_merge): New functions.
22272 (build_rdg_partition_for_vertex): Compute partition type.
22273 (rdg_build_partitions): Dump partition type.
22274 (distribute_loop): Update calls to partition_merge_into.
22276 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22278 * tree-loop-distribution.c (struct ddr_hasher): New.
22279 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
22281 (classify_partition): Call get_data_dependence.
22282 (pg_add_dependence_edges): Ditto.
22283 (distribute_loop): Release data dependence hash table.
22285 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22287 * tree-loop-distribution.c (ref_base_address): Delete.
22288 (similar_memory_accesses): Rename ...
22289 (share_memory_accesses): ... to this. Check if partitions access
22290 the same memory reference.
22291 (distribute_loop): Call share_memory_accesses.
22293 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22295 * tree-loop-distribution.c (struct partition): New field recording
22296 its data reference.
22297 (partition_alloc, partition_free): Init and release data refs.
22298 (partition_merge_into): Merge data refs.
22299 (build_rdg_partition_for_vertex): Collect data refs for partition.
22300 (pg_add_dependence_edges): Change parameters from vector to bitmap.
22302 (distribute_loop): Remve data refs from vertice data of partition
22305 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22307 * tree-loop-distribution.c (params.h): Include header file.
22308 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
22309 (datarefs_vec): New global var.
22310 (create_rdg_vertices): Use datarefs_vec directly.
22311 (free_rdg): Don't free data references.
22312 (build_rdg): Update use. Don't free data references.
22313 (distribute_loop): Compute global variable for data references.
22314 Bail out if there are too many data references.
22316 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22318 * tree-loop-distribution.c (loop_nest): New global var.
22319 (build_rdg): Use loop directly, rather than loop nest.
22320 (pg_add_dependence_edges): Remove loop nest parameter. Use global
22322 (distribute_loop): Compute global variable loop nest. Update use.
22324 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22326 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
22327 (partition_merge_into): New parameter. Dump reason for fusion.
22328 (distribute_loop): Update use of partition_merge_into.
22330 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22332 * tree-loop-distribution.c (bb_top_order_index): New.
22333 (bb_top_order_index_size, bb_top_order_cmp): New.
22334 (stmts_from_loop): Use topological order.
22335 (pass_loop_distribution::execute): Compute and release topological
22336 order for basic blocks.
22338 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22340 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
22343 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22345 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
22346 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
22347 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
22348 * internal-fn.def (LOOP_DIST_ALIAS): New.
22349 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
22350 (fold_loop_internal_call): ... this.
22351 (vect_loop_dist_alias_call): New function.
22352 (set_uid_loop_bbs): Call fold_loop_internal_call.
22353 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
22356 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22359 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
22360 Require dead FLAGS_REG at the beginning of a peephole.
22362 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
22365 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
22366 arguments in the call to __builtin_ia32_sbb_u32.
22367 (_subborrow_u64): Swap _X and _Y arguments in the call to
22368 __builtin_ia32_sbb_u64.
22370 2017-07-04 Jakub Jelinek <jakub@redhat.com>
22373 * tree-vrp.c (compare_assert_loc): Turn into a function template
22374 with stable template parameter. Only test if a->e is NULL,
22375 !a->e == !b->e has been verified already. Use e == NULL or
22376 e != NULL instead of e or ! e tests. If stable is true, don't use
22377 iterative_hash_expr, on the other side allow a or b or both NULL
22378 and sort the NULLs last.
22379 (process_assert_insertions): Sort using compare_assert_loc<false>
22380 instead of compare_assert_loc, later sort using
22381 compare_assert_loc<true> before calling process_assert_insertions_for
22382 in a loop. Use break instead of continue once seen NULL pointer.
22384 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22386 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22387 Cortex-R7 and Cortex-R8 processors.
22389 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22391 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
22392 uninitialized while src is not.
22394 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
22396 * common/config/arm/arm-common.c: Adjust include path for
22398 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
22399 (arm-cpu.h): Create in build directory. Adjust dependency rules.
22400 (arm-cpu-data.h): Likewise.
22401 (arm-cpu-cdata.h): Likewise.
22402 * config/arm/arm-cpu.h: Delete.
22403 * config/arm/arm-cpu-cdata.h: Delete.
22404 * config/arm/arm-cpu-data.h: Delete.
22406 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
22408 * config/arm/arm-cpus.in (cortex-a55): New.
22409 (cortex-a75): Likewise.
22410 (cortex-a75.cortex-a55): Likewise.
22411 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
22413 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
22414 * config/arm/arm-cpu-cdata.h: Regenerate.
22415 * config/arm/arm-cpu-data.h: Regenerate.
22416 * config/arm/arm-cpu.h: Regenerate.
22417 * config/arm/arm-tables.opt: Regenerate.
22418 * config/arm/arm-tune.md: Regenerate.
22420 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22422 * haifa-sched.c (sched_create_recovery_edges): Update profile.
22424 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
22426 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
22429 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
22431 PR tree-optimization/81292
22432 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
22433 full_string_p, also call adjust_related_strinfos if the adjustment
22434 is simple, otherwise invalidate related strinfos.
22436 2017-07-04 Martin Liska <mliska@suse.cz>
22439 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
22440 newly created variable as DECL_IGNORED_P.
22442 2017-07-04 Martin Liska <mliska@suse.cz>
22445 * ipa-inline.c (inline_small_functions):
22446 Use xstrdup_for_dump.
22448 2017-07-04 Tom de Vries <tom@codesourcery.com>
22450 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
22452 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22455 * config/darwin.c (darwin_function_switched_text_sections):
22458 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
22460 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
22462 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22464 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
22466 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22468 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
22469 min_profitable_iters, and th as inclusive lower bounds.
22470 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
22471 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
22472 for min_profitable_iters and min_profitable_estimate.
22473 (vect_transform_loop): Treat th as an inclusive lower bound.
22474 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22476 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
22479 * config/darwin.c (darwin_function_switched_text_sections):
22480 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
22481 in two pieces, and suppress the use of buf.
22483 2017-07-03 Nathan Sidwell <nathan@acm.org>
22485 * hash-table.h (hash_table_mod1): Fix indentation.
22487 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22489 PR middle-end/81290
22490 * predict.c (force_edge_cold): Be more careful about propagation
22492 * profile-count.h (profile_probability::guessed,
22493 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
22495 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
22497 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
22499 * doc/invoke.texi (rcpc architecture extension): Document it.
22501 2017-07-03 Richard Biener <rguenther@suse.de>
22503 PR tree-optimization/60510
22504 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
22505 the scalar reduction PHI and use it.
22506 (vectorizable_reduction): Properly guard the single_defuse_cycle
22507 path for non-SLP reduction chains where we cannot use it.
22508 Rework reduc_def/index and vector type deduction. Rework
22509 vector operand gathering during reduction op code-gen.
22510 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
22511 chains dissolve the chain and leave it to non-SLP reduction
22514 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22516 * tree-data-ref.h (dr_alignment): Declare.
22517 * tree-data-ref.c (dr_alignment): New function.
22518 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
22519 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22521 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
22523 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22525 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
22526 and base_misalignment fields.
22527 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
22528 * tree-data-ref.c: Include builtins.h.
22529 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
22530 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
22531 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
22532 * tree-vect-data-refs.c: Include tree-cfg.h.
22533 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
22534 fields instead of calculating an alignment here.
22535 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
22536 innermost_loop_behavior fields.
22538 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22540 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
22542 (DR_STEP_ALIGNMENT): New macro.
22543 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
22544 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
22545 (create_data_ref): Print it.
22546 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
22547 to tell whether the step preserves vector (mis)alignment.
22548 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22549 Move the check for an integer step and generalise to all INTEGER_CST.
22550 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
22551 Print the outer step alignment.
22553 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22555 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
22556 with offset_alignment.
22557 (DR_ALIGNED_TO): Delete.
22558 (DR_OFFSET_ALIGNMENT): New macro.
22559 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
22560 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
22561 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
22562 (create_data_ref): Likewise.
22563 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
22564 (vect_analyze_data_refs): Likewise.
22565 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
22566 creating dummy innermost behavior.
22568 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22570 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
22571 with a "innermost_loop_behavior *" and refeence tree.
22572 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22573 (create_data_ref): Update call accordingly.
22574 * tree-predcom.c (find_looparound_phi): Likewise.
22576 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22578 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
22579 fields with dr_wrt_vec_loop.
22580 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
22581 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
22582 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
22583 (vect_dr_behavior): New function.
22584 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22585 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22586 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
22587 track whether the step preserves the misalignment.
22588 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
22589 Use vect_dr_behavior.
22590 (vect_setup_realignment): Update call accordingly.
22591 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
22592 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
22593 call to vect_create_addr_base_for_vector_ref.
22594 (vect_create_cond_for_align_checks): Likewise.
22595 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
22596 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
22597 (vect_recog_mask_conversion_pattern): Likewise.
22598 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
22599 (new_stmt_vec_info): Remove redundant zeroing.
22601 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
22603 * common/config/arm/arm-common.c (arm_be8_option): New function.
22604 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
22605 (ISA_ARMv6): Add isa_bit_be8.
22606 * config/arm/arm.h (arm_be8_option): Add prototype.
22607 (BE8_SPEC_FUNCTION): New define.
22608 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
22609 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
22610 (mlittle-endian): Similarly.
22611 (mbe8, mbe32): New options.
22612 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
22613 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
22615 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22617 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
22619 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22621 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
22622 (cleanup_tree_cfg_bb): Use it.
22623 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
22625 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
22627 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22630 * loop-doloop.c (add_test): Update profile.
22632 2017-07-03 Martin Liska <mliska@suse.cz>
22635 * sanopt.c (rewrite_usage_of_param): New function.
22636 (sanitize_rewrite_addressable_params): Likewise.
22637 (pass_sanopt::execute): Call rewrite_usage_of_param.
22639 2017-07-03 Richard Biener <rguenther@suse.de>
22641 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
22642 back to using VIEW_CONVERT_EXPR.
22644 2017-07-03 Martin Liska <mliska@suse.cz>
22647 * doc/extend.texi: Document when a resolver function is
22648 generated for target_clones.
22650 2017-07-03 Martin Liska <mliska@suse.cz>
22652 * asan.c (asan_emit_stack_protection): Unpoison just red zones
22653 and shadow memory of auto variables which are subject of
22654 use-after-scope sanitization.
22655 (asan_expand_mark_ifn): Add do set only when is_poison.
22657 2016-07-03 Richard Biener <rguenther@suse.de>
22659 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
22661 (vect_force_simple_reduction): Record reduction def -> phi mapping.
22662 (vectorizable_reduction): Perform reduction PHI creation when
22663 visiting a reduction PHI and adjust and simplify code generation
22664 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
22665 (vect_transform_loop): Visit reduction PHIs.
22666 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
22667 defs into the SLP tree.
22668 (vect_build_slp_tree): Reduction defs terminate the recursion.
22669 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
22671 (vect_get_vec_defs_for_stmt_copy): Export.
22672 (vect_get_vec_defs): Likewise.
22673 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
22675 (vect_get_vec_defs_for_stmt_copy): Declare.
22676 (vect_get_vec_defs): Likewise.
22678 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
22680 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
22681 parameter with a "loop" parameter and use it instead of the
22682 loop containing DR_STMT. Don't check simple_iv when doing
22683 BB analysis. Describe the two analysis modes in the comment.
22685 2017-07-03 Tom de Vries <tom@codesourcery.com>
22687 PR tree-optimization/69468
22688 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
22689 (find_same_succ_bb): Handle ignore_edge_flags.
22691 2017-07-03 Tom de Vries <tom@codesourcery.com>
22693 PR tree-optimization/81192
22694 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
22696 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
22698 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
22700 2017-07-03 Tom de Vries <tom@codesourcery.com>
22702 PR tree-optimization/81192
22703 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
22704 BB_SAME_SUCC (bb) == NULL.
22706 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22708 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
22711 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22713 * dumpfile.c: Include profile-count.h
22714 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
22716 (insert_cond_bb): Update profile.
22717 * tree-cfg.h (insert_cond_bb): Update prototype.
22718 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
22719 * tree-dump.c: Do not include tree-cfg.
22721 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22723 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
22725 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22727 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
22730 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
22732 * tree-complex.c (expand_complex_div_wide): update profile.
22734 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22735 Alan Hayward <alan.hayward@arm.com>
22736 David Sherwood <david.sherwood@arm.com>
22738 * Makefile.in (MACHMODE_H): Remove insn-modes.h
22739 (CORETYPES_H): New define.
22740 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
22741 (insn-modes-inline.h, s-modes-inline-h): New rules.
22742 (generated_files): Add insn-modes-inline.h.
22743 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
22744 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
22745 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
22746 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
22747 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
22748 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
22749 (build/gencodes.o, build/genconditions.o): Likewise.
22750 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
22751 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
22752 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
22753 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
22754 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
22755 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
22756 * coretypes.h: Include everything up to real.h for generators.
22757 Include insn-modes.h first. Include wide-int-print.h after
22758 wide-int.h. Include insn-modes-inline.h and then machmode.h.
22759 * machmode.h: Don't include insn-modes.h here.
22760 * function-tests.c: Remove includes of signop.h, machmode.h,
22761 double-int.h and wide-int.h.
22763 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
22765 * optc-save-gen.awk: Likewise.
22766 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
22767 * godump.c: Remove include of wide-int-print.h.
22768 * pretty-print.h: Likewise.
22769 * wide-int-print.cc: Likewise.
22770 * wide-int.cc: Likewise.
22771 * hash-map-tests.c: Remove include of signop.h.
22772 * hash-set-tests.c: Likewise.
22773 * rtl-tests.c: Likewise.
22774 * mkconfig.sh: Remove include of machmode.h.
22775 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
22777 (emit_insn_modes_inline_h): ...this new function. Emit the code
22778 into an insn-modes-inline.h header file, adding appropriate
22779 include guards and end comments.
22780 (emit_insn_modes_c_header): Remove include of machmode.h.
22781 (emit_min_insn_modes_c_header): Include coretypes.h rather than
22783 (main): Handle -i flag and call emit_insn_modes_inline_h when
22786 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22788 * tree-ssa-strlen.c (strinfo): Rename the length field to
22789 nonzero_chars. Add a full_string_p field.
22790 (compare_nonzero_chars, zero_length_string_p): New functions.
22791 (get_addr_stridx): Add an offset_out parameter.
22792 Use compare_nonzero_chars.
22793 (get_stridx): Update accordingly. Use compare_nonzero_chars.
22794 (new_strinfo): Update after above changes to strinfo.
22795 (set_endptr_and_length): Set full_string_p.
22796 (get_string_length): Update after above changes to strinfo.
22797 (unshare_strinfo): Update call to new_strinfo.
22798 (maybe_invalidate): Likewise.
22799 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
22800 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
22801 as a uhwi instead of an shwi. Update after above changes to
22802 strinfo and new_strinfo.
22803 (zero_length_string): Assert that chainsi contains full strings.
22804 Use zero_length_string_p. Update call to new_strinfo.
22805 (adjust_related_strinfos): Update after above changes to strinfo.
22806 Copy full_string_p from origsi.
22807 (adjust_last_stmt): Use zero_length_string_p.
22808 (handle_builtin_strlen): Update after above changes to strinfo and
22809 new_strinfo. Install the lhs as the string length if the previous
22810 entry didn't describe a full string.
22811 (handle_builtin_strchr): Update after above changes to strinfo
22813 (handle_builtin_strcpy): Likewise.
22814 (handle_builtin_strcat): Likewise.
22815 (handle_builtin_malloc): Likewise.
22816 (handle_pointer_plus): Likewise.
22817 (handle_builtin_memcpy): Likewise. Track nonzero characters
22818 that aren't necessarily followed by a nul terminator.
22819 (handle_char_store): Likewise.
22821 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22823 PR tree-optimization/80769
22824 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
22825 for malloc and calloc. Document the new invariant that all related
22826 strinfos have delayed lengths or none do.
22827 (verify_related_strinfos): Move earlier in file.
22828 (set_endptr_and_length): New function, split out from...
22829 (get_string_length): ...here. Also set the lengths of related
22831 (zero_length_string): Assert that chainsi has known (rather than
22833 (adjust_related_strinfos): Likewise.
22835 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
22837 PR tree-optimization/81136
22838 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
22839 assert that two references with the same misalignment have the same
22840 compile-time misalignment if those compile-time misalignments
22843 2017-07-01 Andi Kleen <ak@linux.intel.com>
22845 * print-tree.c (print_node): Print all attributes.
22847 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22849 * cfg.c (scale_bbs_frequencies): New function.
22850 * cfg.h (scale_bbs_frequencies): Declare it.
22851 * cfgloopanal.c (single_likely_exit): Cleanup.
22852 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
22854 (scale_loop_profile): Likewise.
22855 (loop_version): Likewise.
22856 (create_empty_loop_on_edge): Update.
22857 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
22858 scale_loop_frequencies, scale_loop_profile, loopify,
22859 loop_version): Update prototypes.
22860 * modulo-sched.c (sms_schedule): Update.
22861 * predict.c (unlikely_executed_edge_p): Also check probability.
22862 (probably_never_executed_edge_p): Fix typo.
22863 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22864 * tree-parloops.c (gen_parallel_loop): Update.
22865 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
22866 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22867 * tree-ssa-loop-split.c (split_loop): Update.
22868 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22869 * tree-vect-loop-manip.c (vect_do_peeling): Update.
22870 (vect_loop_versioning): Update.
22871 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22873 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22875 * trans-mem.c (split_bb_make_tm_edge): Update profile.
22877 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22879 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
22880 to keep profile consistent.
22882 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22884 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
22885 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
22886 * profile-count.h (max_safe_multiplier): Make unsigned.
22887 (profile_count::guessed_zero): New.
22889 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22891 * bb-reorder.c (fix_up_crossing_landing_pad,
22892 fix_crossing_conditional_branches): Use make_single_succ_edge
22893 to keep profile consistent.
22895 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22897 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22900 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22903 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22904 the right scopes, make sure cond_jump isn't preserved between multiple
22905 iterations. Search for fallthru edge whenever there are 3+ edges and
22906 use find_fallthru_edge for it.
22908 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22910 Patch by Alexander Monakov <amonakov@ispras.ru>
22911 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22912 probabilities consistently.
22914 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22916 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22917 emit_cmp_and_jump_insns.
22919 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22922 * tree-inline.c (expand_call_inline): Combine profile statuses.
22924 2017-06-30 Andrew Pinski <apinski@cavium.com>
22926 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22927 fold_stmt returned true.
22929 2017-06-30 Nathan Sidwell <nathan@acm.org>
22931 * ggc.h (empty_string): Delete.
22932 * cfgexpand.c (expand_asm_stmt): Use plain "".
22933 * optabs.c (expand_asm_memory_barrier): Likewise.
22934 * stringpool.c (empty_string): Delete.
22935 (digit_vector, digit_string): Delete.
22936 (ggc_alloc_string): Use plain "", don't optimize single digit
22937 strings. Use ggc_alloc_atomic.
22939 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22941 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22942 comparison set and one other set, use the cost of the non-comparison
22945 2017-06-30 Nathan Sidwell <nathan@acm.org>
22947 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22950 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22952 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22953 loops. Remove now unneeded calls to gimple_switch_set_label() that
22954 just set removed labels to NULL_TREE.
22956 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22958 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22959 (set_range_info): ...here. Only call set_range_info_raw if domain
22961 (set_nonzero_bits): Call set_range_info_raw.
22962 * tree-ssanames.h (set_range_info_raw): New.
22964 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22967 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22968 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22969 of nonimmediate_operand and <store_mask_constraint> instead of m
22970 for the input operand. For V8FI iterator, always split if input
22971 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22972 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22973 <store_mask_predicate> instead of register_operand and
22974 <store_mask_constraint> instead of v for the input operand. Make
22975 sure both operands aren't MEMs for if not <mask_applied>.
22977 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22979 * lto-wrapper.c (copy_file) Close both file descriptors before
22982 2017-06-30 Martin Liska <mliska@suse.cz>
22985 * multiple_target.c (create_dispatcher_calls): Make ifunc
22986 also for function that don't have calls or are not referenced.
22988 2017-06-30 Richard Biener <rguenther@suse.de>
22990 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22991 analyze the first scalar stmt. Move vector type computation
22992 for the BB case here from ...
22993 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22994 live operation processing in the SLP case properly.
22996 2017-06-30 Richard Biener <rguenther@suse.de>
22998 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23000 2017-06-30 Martin Liska <mliska@suse.cz>
23003 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23004 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23006 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23008 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23011 2017-06-30 Richard Biener <rguenther@suse.de>
23013 PR tree-optimization/81249
23014 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23015 condition reduction result to original scalar type.
23017 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23019 * profile-count.h (enum profile_quality): Fix typos and whitespace
23022 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23024 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23025 type for branch probabilities.
23027 2017-06-29 Julian Brown <julian@codesourcery.com>
23028 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23030 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23031 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23032 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23033 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23035 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23037 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23038 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23039 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23040 CC usage from generic code to here.
23041 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23042 CC usage into the target macros.
23044 2017-06-29 Maya Rashish <coypu@sdf.org>
23046 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23049 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23051 * arm/arm-builtins.c: Include profile-count.h
23052 * except.c (sjlj_emit_function_enter): Use
23053 profile_probability::unlikely.
23055 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23057 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23058 and tocrel_offset be pointer args rather than implicitly using
23060 (legitimate_constant_pool_address_p, rs6000_emit_move,
23061 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23062 tocrel_offset and use in toc_relative_expr_p call.
23063 (print_operand, print_operand_address): Use static tocrel_base_oac
23064 and tocrel_offset_oac.
23065 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23068 2017-06-29 Maya Rashish <coypu@sdf.org>
23070 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23072 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23074 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23075 objects, take into account only the alignment of 'op0' and 'mode1' if
23078 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23080 * ccmp.c (ccmp_tree_comparison_p): New function.
23081 (ccmp_candidate_p): Update to use above function.
23082 (get_compare_parts): New function.
23083 (expand_ccmp_next): Update to use new functions.
23084 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23086 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23087 take mode as argument.
23088 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23089 * expr.c (expand_expr_real_1): Pass mode as argument.
23091 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23093 * combine.c (combine_instructions): Print insns to dump_file, together
23096 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23098 * asan.c (asan_emit_stack_protection): Update.
23099 (create_cond_insert_point): Update.
23100 * auto-profile.c (afdo_propagate_circuit): Update.
23101 * basic-block.h (struct edge_def): Turn probability to
23102 profile_probability.
23103 (EDGE_FREQUENCY): Update.
23104 * bb-reorder.c (find_traces_1_round): Update.
23105 (better_edge_p): Update.
23106 (sanitize_hot_paths): Update.
23107 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23108 (make_single_succ_edge): Update.
23109 (check_bb_profile): Update.
23110 (dump_edge_info): Update.
23111 (update_bb_profile_for_threading): Update.
23112 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23113 probabilitycount to 0.
23114 * cfgbuild.c (compute_outgoing_frequencies): Update.
23115 * cfgcleanup.c (try_forward_edges): Update.
23116 (outgoing_edges_match): Update.
23117 (try_crossjump_to_edge): Update.
23118 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23119 (expand_gimple_tailcall): Update.
23120 (construct_init_block): Use make_single_succ_edge.
23121 (construct_exit_block): Use make_single_succ_edge.
23122 * cfghooks.c (verify_flow_info): Update.
23123 (redirect_edge_succ_nodup): Update.
23124 (split_edge): Update.
23125 (account_profile_record): Update.
23126 * cfgloopanal.c (single_likely_exit): Update.
23127 * cfgloopmanip.c (scale_loop_profile): Update.
23128 (set_zero_probability): Remove.
23129 (duplicate_loop_to_header_edge): Update.
23130 * cfgloopmanip.h (loop_version): Update prototype.
23131 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23132 (force_nonfallthru_and_redirect): Update.
23133 (update_br_prob_note): Update.
23134 (rtl_verify_edges): Update.
23135 (purge_dead_edges): Update.
23136 (rtl_lv_add_condition_to_bb): Update.
23137 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23138 * cgraphunit.c (init_lowered_empty_function): Update.
23139 (cgraph_node::expand_thunk): Update.
23140 * cilk-common.c: Include profile-count.h
23141 * dojump.c (inv): Remove.
23142 (jumpifnot): Update.
23143 (jumpifnot_1): Update.
23144 (do_jump_1): Update.
23146 (do_jump_by_parts_greater_rtx): Update.
23147 (do_compare_rtx_and_jump): Update.
23148 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23149 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23150 * dwarf2cfi.c: Include profile-count.h
23151 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23152 (sjlj_emit_dispatch_table): Likewise.
23153 * explow.c: Include profile-count.h
23154 * expmed.c (emit_store_flag_force): Update.
23155 (do_cmp_and_jump): Update.
23156 * expr.c (compare_by_pieces_d::generate): Update.
23157 (compare_by_pieces_d::finish_mode): Update.
23158 (emit_block_move_via_loop): Update.
23159 (store_expr_with_bounds): Update.
23160 (store_constructor): Update.
23161 (expand_expr_real_2): Update.
23162 (expand_expr_real_1): Update.
23163 * expr.h (try_casesi, try_tablejump): Update prototypes.
23164 * gimple-pretty-print.c (dump_probability): Update.
23165 (dump_profile): New.
23166 (dump_gimple_label): Update.
23167 (dump_gimple_bb_header): Update.
23168 * graph.c (draw_cfg_node_succ_edges): Update.
23169 * hsa-gen.c (convert_switch_statements): Update.
23170 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23171 (find_if_case_1): Update.
23172 (find_if_case_2): Update.
23173 * internal-fn.c (expand_arith_overflow_result_store): Update.
23174 (expand_addsub_overflow): Update.
23175 (expand_neg_overflow): Update.
23176 (expand_mul_overflow): Update.
23177 (expand_vector_ubsan_overflow): Update.
23178 * ipa-cp.c (good_cloning_opportunity_p): Update.
23179 * ipa-split.c (split_function): Use make_single_succ_edge.
23180 * ipa-utils.c (ipa_merge_profiles): Update.
23181 * loop-doloop.c (add_test): Update.
23182 (doloop_modify): Update.
23183 * loop-unroll.c (compare_and_jump_seq): Update.
23184 (unroll_loop_runtime_iterations): Update.
23185 * lra-constraints.c (lra_inheritance): Update.
23186 * lto-streamer-in.c (input_cfg): Update.
23187 * lto-streamer-out.c (output_cfg): Update.
23188 * mcf.c (adjust_cfg_counts): Update.
23189 * modulo-sched.c (sms_schedule): Update.
23190 * omp-expand.c (expand_omp_for_init_counts): Update.
23191 (extract_omp_for_update_vars): Update.
23192 (expand_omp_ordered_sink): Update.
23193 (expand_omp_for_ordered_loops): Update.
23194 (expand_omp_for_generic): Update.
23195 (expand_omp_for_static_nochunk): Update.
23196 (expand_omp_for_static_chunk): Update.
23197 (expand_cilk_for): Update.
23198 (expand_omp_simd): Update.
23199 (expand_omp_taskloop_for_outer): Update.
23200 (expand_omp_taskloop_for_inner): Update.
23201 * omp-simd-clone.c (simd_clone_adjust): Update.
23202 * optabs.c (expand_doubleword_shift): Update.
23203 (expand_abs): Update.
23204 (emit_cmp_and_jump_insn_1): Update.
23205 (expand_compare_and_swap_loop): Update.
23206 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23207 * predict.c (predictable_edge_p): Update.
23208 (edge_probability_reliable_p): Update.
23209 (set_even_probabilities): Update.
23210 (combine_predictions_for_insn): Update.
23211 (combine_predictions_for_bb): Update.
23212 (propagate_freq): Update.
23213 (estimate_bb_frequencies): Update.
23214 (force_edge_cold): Update.
23215 * profile-count.c (profile_count::dump): Add missing space into dump.
23216 (profile_count::debug): Add newline.
23217 (profile_count::differs_from_p): Explicitly convert to unsigned.
23218 (profile_count::stream_in): Update.
23219 (profile_probability::dump): New member function.
23220 (profile_probability::debug): New member function.
23221 (profile_probability::differs_from_p): New member function.
23222 (profile_probability::differs_lot_from_p): New member function.
23223 (profile_probability::stream_in): New member function.
23224 (profile_probability::stream_out): New member function.
23225 * profile-count.h (profile_count_quality): Rename to ...
23226 (profile_quality): ... this one.
23227 (profile_probability): New.
23228 (profile_count): Update.
23229 * profile.c (compute_branch_probabilities): Update.
23230 * recog.c (peep2_attempt): Update.
23231 * sched-ebb.c (schedule_ebbs): Update.
23232 * sched-rgn.c (find_single_block_region): Update.
23233 (compute_dom_prob_ps): Update.
23234 (schedule_region): Update.
23235 * sel-sched-ir.c (compute_succs_info): Update.
23236 * stmt.c (struct case_node): Update.
23237 (do_jump_if_equal): Update.
23238 (get_outgoing_edge_probs): Update.
23239 (conditional_probability): Update.
23240 (emit_case_dispatch_table): Update.
23241 (expand_case): Update.
23242 (expand_sjlj_dispatch_table): Update.
23243 (emit_case_nodes): Update.
23244 * targhooks.c: Update.
23245 * tracer.c (better_p): Update.
23246 (find_best_successor): Update.
23247 * trans-mem.c (expand_transaction): Update.
23248 * tree-call-cdce.c: Update.
23249 * tree-cfg.c (gimple_split_edge): Upate.
23250 (move_sese_region_to_fn): Upate.
23251 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
23252 * tree-eh.c (lower_resx): Upate.
23253 (cleanup_empty_eh_move_lp): Upate.
23254 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23255 * tree-inline.c (copy_edges_for_bb): Update.
23256 (copy_cfg_body): Update.
23257 * tree-parloops.c (gen_parallel_loop): Update.
23258 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
23259 (gimple_gen_time_profiler): Update.
23260 * tree-ssa-dce.c (remove_dead_stmt): Update.
23261 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23262 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
23263 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
23264 (unloop_loops): Update.
23265 (try_peel_loop): Update.
23266 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23267 * tree-ssa-loop-split.c (connect_loops): Update.
23268 (split_loop): Update.
23269 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23270 (hoist_guard): Update.
23271 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
23272 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
23273 (value_replacement): Update.
23274 * tree-ssa-reassoc.c (branch_fixup): Update.
23275 * tree-ssa-tail-merge.c (replace_block_by): Update.
23276 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
23277 (create_edge_and_update_destination_phis): Update.
23278 (compute_path_counts): Update.
23279 (recompute_probabilities): Update.
23280 (update_joiner_offpath_counts): Update.
23281 (freqs_to_counts_path): Update.
23282 (duplicate_thread_path): Update.
23283 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
23284 (struct switch_conv_info): Update.
23285 (gen_inbound_check): Update.
23286 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23287 (vect_do_peeling): Update.
23288 (vect_loop_versioning): Update.
23289 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23290 (optimize_mask_stores): Update.
23291 * ubsan.c (ubsan_expand_null_ifn): Update.
23292 * value-prof.c (gimple_divmod_fixed_value): Update.
23293 (gimple_divmod_fixed_value_transform): Update.
23294 (gimple_mod_pow2): Update.
23295 (gimple_mod_pow2_value_transform): Update.
23296 (gimple_mod_subtract): Update.
23297 (gimple_mod_subtract_transform): Update.
23298 (gimple_ic): Update.
23299 (gimple_stringop_fixed_value): Update.
23300 (gimple_stringops_transform): Update.
23301 * value-prof.h: Update.
23303 2017-06-29 Carl Love <cel@us.ibm.com>
23305 * config/rs6000/rs6000-c.c: Add support for built-in functions
23306 vector signed int vec_signed (vector float);
23307 vector signed long long vec_signed (vector double);
23308 vector signed int vec_signed2 (vector double, vector double);
23309 vector signed int vec_signede (vector double);
23310 vector signed int vec_signedo (vector double);
23311 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
23312 instruction generator.
23313 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23314 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
23315 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
23317 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
23318 vunsignede_v2df): Add define_expands.
23319 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
23320 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
23321 VEC_UNSIGNEDO): Add definitions.
23322 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
23323 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
23324 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
23325 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
23326 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
23327 * config/rs6000/altivec.h (vec_signed, vec_signed2,
23328 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
23329 vec_unsignede, vec_unsignedo): Add builtin defines.
23330 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
23332 * doc/extend.texi: Update the built-in documentation file for the
23333 new built-in functions.
23335 2017-06-29 Richard Biener <rguenther@suse.de>
23337 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
23338 reduction chains to LOOP_VINFO_REDUCTIONS.
23339 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
23340 SLP reductions after processing reduction chains.
23342 2017-06-29 Nathan Sidwell <nathan@acm.org>
23344 * builtins.c (fold_builtin_FUNCTION): Use
23345 lang_hooks.decl_printable_name.
23347 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
23349 PR middle-end/81194
23350 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
23351 with only one label.
23352 * stmt.c (expand_case): Assert NCASES is greater than one.
23354 2017-06-29 Richard Biener <rguenther@suse.de>
23356 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
23358 (group_case_labels): Likewise.
23359 (find_taken_edge): Push sanity checking on val to workers...
23360 (find_taken_edge_cond_expr): ... here
23361 (find_taken_edge_switch_expr): ... and here, handle cases
23362 with just a default label.
23363 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
23364 (group_case_labels): Likewise.
23365 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
23366 group_case_labels does anything cleanup the CFG again.
23368 2017-06-29 Bin Cheng <bin.cheng@arm.com>
23370 PR tree-optimization/81196
23371 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
23372 exit condition comparing two IVs.
23374 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
23376 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
23377 profile to the dummy entry at the end of the list of architectures.
23378 * config/arm/arm-cpu-cdata.h: Regenerated.
23380 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23381 Michael Collison <michael.collison@arm.com>
23384 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
23386 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
23387 (*aarch64_reg_<mode>3_minus_mask): New pattern.
23388 (*aarch64_<optab>_reg_di3_mask2): New pattern.
23389 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
23390 of shift when the shift amount is masked with constant equal to
23391 the size of the mode.
23392 * config/aarch64/predicates.md (subreg_lowpart_operator): New
23395 2017-06-29 Martin Liska <mliska@suse.cz>
23397 * config/i386/i386.opt: Change range from [1,5] to [0,5].
23399 2017-06-29 Yury Gribov <tetra2005@gmail.com>
23402 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
23405 (edge_growth_cache_entry::edge_growth_cache_entry): New
23407 (reset_edge_growth_cache): Update to use constructor.
23409 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23411 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
23412 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23413 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
23415 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
23417 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
23418 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
23420 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
23422 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
23423 (*-linux-uclibc*): Add t-uclibc tmake_file.
23424 * config/t-musl: New.
23425 * config/t-uclibc: New.
23427 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
23429 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
23431 (gen_comm_data): Emit architectural setting of arch_prof.
23432 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
23434 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
23435 (armv8-m.base, armv8-m.main): Likewise.
23436 * arm-protos.h (arm_build_target): Add profile field.
23437 (arch_option): Likewise.
23438 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
23440 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
23441 arm_active_target.profile.
23443 2017-06-28 Richard Biener <rguenther@suse.de>
23445 PR middle-end/81227
23446 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
23447 TYPE_OVERFLOW_WRAPS.
23448 * match.pd (negate_expr_p): Likewise.
23449 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
23450 fold_build2, not fold_binary.
23452 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23454 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
23455 Convert memory address to Pmode.
23456 (aarch64_print_operand): Assert MEM operands are always Pmode.
23458 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
23461 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
23462 Remove redundant if.
23463 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
23464 * config/arm/aarch-common-protos.h
23465 (aarch_forward_to_shift_is_not_shifted_re): Remove.
23466 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
23468 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
23471 * multiple_target.c (create_dispatcher_calls): Set the default
23472 clone to be static, not public.
23474 2017-06-28 Richard Biener <rguenther@suse.de>
23476 * tree-vect-loop.c (vectorizable_reduction): Move special
23477 cond reduction IV var creation ...
23478 (vect_create_epilog_for_reduction): ... here. Remove induction_index
23479 parameter. Use STMT_VINFO_VECTYPE.
23480 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
23483 2017-06-28 Martin Liska <mliska@suse.cz>
23486 * ipa-visibility.c (non_local_p): Handle visibility.
23488 2017-06-28 Martin Liska <mliska@suse.cz>
23491 * common.opt: Add IntegerRange to various options.
23492 * opt-functions.awk (integer_range_info): New function.
23493 * optc-gen.awk: Add integer_range_info to cl_options struct.
23494 * opts-common.c (decode_cmdline_option): Handle
23495 CL_ERR_INT_RANGE_ARG.
23496 (cmdline_handle_error): Likewise.
23497 * opts.c (print_filtered_help): Show valid interval in
23498 when --help is provided.
23499 * opts.h (struct cl_option): Add range_min and range_max fields.
23500 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
23502 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
23504 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
23505 (x * C EQ/NE y * C): New transformation.
23507 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
23509 * genmultilib (combination_space): Accept '+' in option names.
23511 2017-06-28 Martin Liska <mliska@suse.cz>
23514 * asan.c (instrument_derefs): Bail out inner references
23515 that are hard register variables.
23517 2017-06-28 Jakub Jelinek <jakub@redhat.com>
23520 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
23521 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
23523 2017-06-28 Richard Biener <rguenther@suse.de>
23525 * tree-vectorizer.h (vect_get_vec_defs): Remove.
23526 (vect_get_slp_defs): Adjust.
23527 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
23529 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
23531 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
23532 get_initial_defs_for_reduction instead of vect_get_vec_defs.
23533 (vectorizable_reduction): Adjust.
23534 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
23536 (vect_get_slp_defs): Likewise.
23537 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
23538 (vectorizable_bswap): Adjust.
23539 (vectorizable_call): Likewise.
23540 (vectorizable_conversion): Likewise.
23541 (vectorizable_assignment): Likewise.
23542 (vectorizable_shift): Likewise.
23543 (vectorizable_operation): Likewise.
23544 (vectorizable_store): Likewise.
23545 (vectorizable_condition): Likewise.
23546 (vectorizable_comparison): Likewise.
23548 2017-06-28 Michael Collison <michael.collison@arm.com>
23551 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
23553 (arm_gen_movmemqi): Removed unused variable 'i'.
23554 Convert 'for' loop into 'while' loop.
23555 (arm_expand_prologue): Remove last unnecessary set of insn.
23556 (thumb_pop): Remove unused variable 'pushed_words'.
23557 (thumb_exit): Remove last unnecessary set of regs_to_pop.
23559 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23561 * config/s390/predicates.md: Use s390_rel_address_ok_p.
23562 * config/s390/s390-protos.h: Add prototype of
23563 s390_rel_address_ok_p.
23564 * config/s390/s390.c (s390_got_symbol): New function.
23565 (s390_rel_address_ok_p): New function.
23566 (legitimize_pic_address): Use s390_rel_address_ok_p.
23567 (s390_load_got): Use s390_got_symbol.
23568 (s390_option_override): Issue error if
23569 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
23570 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
23572 * config/s390/s390.opt: New option mpic-data-is-text-relative.
23574 2017-06-27 Andrew Pinski <apinski@cavium.com>
23576 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
23577 (X * copysign (1.0, X)): New pattern.
23578 (X * copysign (1.0, -X)): New pattern.
23579 (copysign (-1.0, CST)): New pattern.
23581 2017-06-27 Joseph Myers <joseph@codesourcery.com>
23583 * genmultilib (combination_space): Remove variable.
23584 Validate reuse rules against regular expression for any sequence
23585 of multilib options in any order.
23587 2017-06-27 Michael Collison <michael.collison@arm.com>
23589 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
23590 call aarch64_split_simd_combine.
23591 * (aarch64_combine_internal<mode>): Delete pattern.
23592 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
23593 Allow register and subreg operands.
23595 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23597 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
23598 specific need, just fallback on defaults.
23599 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
23601 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23602 Olivier Hainque <hainque@adacore.com>
23604 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
23606 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
23607 targets. Pick a default if no particular attempt applied.
23608 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
23611 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23613 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
23614 (x86_64-wrs-vxworks7): Likewise.
23616 2017-06-27 Marek Polacek <polacek@redhat.com>
23619 * ubsan.c (instrument_null): Check get_base_address's result for null.
23621 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23623 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
23625 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
23627 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
23628 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
23629 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
23630 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
23631 New function types.
23632 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
23633 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
23634 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
23635 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
23636 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
23637 BUILT_IN_FEUPDATEENV): New builtins.
23638 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
23639 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
23640 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
23641 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
23643 (builtin_structptr_types): Adjust size.
23644 * tree.c (builtin_structptr_types): Add four entries.
23646 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23647 Olivier Hainque <hainque@adacore.com>
23649 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
23650 (TLS_SYM): New local macro, forcing reference to __tls__ on
23651 link command lines for VxWorks 7 RTPs, triggering initialization
23653 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
23654 OS features TLS support, true for RTPs on VxWorks 7.
23655 * config/vxworks.c (vxworks_override_options): Setup emutls
23658 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23660 * predict.c (test_prediction_value_range): Use -1U instead of -1
23661 to avoid narrowing conversion warning.
23662 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
23663 to avoid narrowing conversion warning.
23664 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
23666 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
23668 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
23670 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
23671 64bit configurations.
23672 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
23673 (SIZE_TYPE): Likewise.
23674 * config/vxworks.c (vxworks_emutls_var_fields): Use
23675 long_unsigned_type_node instead of unsigned_type_node as the offset
23676 field type, which is "pointer" mode in emutls.c.
23678 2017-06-27 Jakub Jelinek <jakub@redhat.com>
23681 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
23683 PR middle-end/81207
23684 * gimple-fold.c (replace_call_with_call_and_fold): Handle
23685 gimple_vuse copying separately from gimple_vdef copying.
23687 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23689 * value-prof.c (free_hist): Remove call to memset and the enclosing if
23692 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23693 Olivier Hainque <hainque@adacore.com>
23695 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
23696 for all vxworks7 targets.
23697 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
23698 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
23699 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
23700 variations for VX6/VX7 and 32/64bits later on in ...
23701 (VXWORKS_LIB_SPEC): Leverage new macros.
23702 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
23703 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
23705 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
23707 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
23708 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
23710 2017-06-26 Carl Love <cel@us.ibm.com>
23712 * config/rs6000/rs6000-c.c: Add support for built-in functions
23713 vector bool char vec_reve (vector bool char);
23714 vector signed char vec_reve (vector signed char);
23715 vector unsigned char vec_reve (vector unsigned char);
23716 vector bool int vec_reve (vector bool int);
23717 vector signed int vec_reve (vector signed int);
23718 vector unsigned int vec_reve (vector unsigned int);
23719 vector bool long long vec_reve (vector bool long long);
23720 vector signed long long vec_reve (vector signed long long);
23721 vector unsigned long long vec_reve (vector unsigned long long);
23722 vector bool short vec_reve (vector bool short);
23723 vector signed short vec_reve (vector signed short);
23724 vector double vec_reve (vector double);
23725 vector float vec_reve (vector float);
23726 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
23727 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
23728 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
23729 (altivec_vreve): New pattern.
23730 * config/rs6000/altivec.h (vec_reve): New define.
23731 * doc/extend.texi (vec_rev): Update the built-in documentation file
23732 for the new built-in functions.
23734 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23736 PR tree-optimization/71815
23737 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
23739 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
23741 (slsr_process_phi): Likewise.
23742 (replace_uncond_cands_and_profitable_phis): Don't replace a
23743 multiply candidate with a stride of 1 (copy or cast).
23744 (phi_incr_cost): Call uses_consumed_by_stmt rather than
23746 (lowest_cost_path): Likewise.
23747 (total_savings): Likewise.
23749 2017-06-26 Richard Biener <rguenther@suse.de>
23752 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
23753 Use def_builtin_pure for all gather builtins.
23755 2017-06-26 Richard Biener <rguenther@suse.de>
23757 PR tree-optimization/81203
23758 * tree-tailcall.c (find_tail_calls): Do not move stmts into
23759 non-dominating BBs.
23761 2017-06-26 Marek Polacek <polacek@redhat.com>
23764 * doc/invoke.texi: Document -Wmultistatement-macros.
23766 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
23768 * doc/sourcebuild.texi (ARM-specific attributes): Document new
23769 arm_neon_ok_no_float_abi effective target.
23771 2017-06-26 Richard Biener <rguenther@suse.de>
23773 PR tree-optimization/80928
23774 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
23775 (copy_bbs): Set BB_DUPLICATED flag early.
23776 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
23778 (execute_on_shrinking_pred): Likewise.
23779 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
23780 BB_DUPLICATED blocks.
23781 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
23782 iterate over all PHIs considering removal of *gsi.
23784 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
23786 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
23789 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23791 * config/rs6000/rs6000-string.c: (expand_block_clear,
23792 do_load_for_compare, select_block_compare_mode,
23793 compute_current_alignment, expand_block_compare,
23794 expand_strncmp_align_check, expand_strn_compare,
23795 expand_block_move, rs6000_output_load_multiple)
23796 Move functions related to string/block move/compare
23797 to a separate file.
23798 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
23799 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
23800 for this function which is now used in two files.
23801 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
23802 * config.gcc: Add rs6000-string.o to extra_objs for
23803 targets powerpc*-*-* and rs6000*-*-*.
23805 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
23808 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
23809 32-bit, since indexed is not valid for DImode.
23810 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
23811 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
23812 (define_peephole2 for Altivec d-form load): Add 32-bit support.
23813 (define_peephole2 for Altivec d-form store): Likewise.
23816 * multiple_target.c (create_dispatcher_calls): Only create the
23817 dispatcher call if the function is the default clone of a
23818 versioned function.
23820 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
23822 PR middle-end/80902
23823 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
23824 a call, force the call to not be a tail call.
23826 2017-06-23 Jeff Law <law@redhat.com>
23828 * doc/contrib.texi: Add entry for Steven Pemberton's work on
23831 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
23833 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
23834 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
23835 handling for early expansion of vector shifts (sl,sr,sra,rl).
23836 (builtin_function_type): Add vector shift right instructions
23837 to the unsigned argument list.
23839 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
23841 rtl-optimizatoin/79286
23842 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
23843 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
23844 trap. PIC register plus a const unspec without offset can never trap.
23846 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
23848 * tree.h (builtin_structptr_type): New type.
23849 (builtin_structptr_types): Declare new array.
23850 * tree.c (builtin_structptr_types): New array.
23851 (free_lang_data, build_common_tree_nodes): Use it.
23853 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
23856 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
23859 2017-06-22 Matt Turner <mattst88@gmail.com>
23861 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
23862 Lake models to skylake case. Assume skylake for unknown
23863 models with clflushopt.
23865 2017-06-22 Jeff Law <law@redhat.com>
23867 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
23868 frame sizes that do not satisfy aarch64_uimm12_shift.
23870 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
23872 * profile-count.h (apply_probability,
23873 apply_scale, probability_in): Fix checks for zero.
23875 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23877 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
23878 * doc/cppdiropts.texi (-I @var{dir}): Document it.
23880 2016-06-22 Richard Biener <rguenther@suse.de>
23882 * tree-vect-loop.c (vect_model_reduction_cost): Handle
23883 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
23884 REDUC_MAX_EXPR support.
23885 (vectorizable_reduction): Likewise.
23886 (vect_create_epilog_for_reduction): Likewise.
23888 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23890 * match.pd (A / (1 << B) -> A >> B): New.
23891 * generic-match-head.c: Include optabs-tree.h.
23892 * gimple-match-head.c: Likewise.
23893 * optabs-tree.h (target_supports_op_p): New.
23894 * optabs-tree.c (target_supports_op_p): New.
23896 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23898 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23899 $gcc_cv_ld --help output.
23900 (gcc_cv_ld_demangle): Likewise.
23901 (gcc_cv_ld_eh_frame_hdr): Likewise.
23902 (gcc_cv_ld_pie): Likewise.
23903 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23904 (gcc_cv_ld_buildid): Likewise.
23905 (gcc_cv_ld_sysroot): Likewise.
23906 (ld_bndplt_support): Likewise.
23907 (ld_pushpopstate_support): Likewise.
23908 * configure: Regenerate.
23909 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23911 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23914 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23915 operands indexes to avoid gap between operands and match_dups.
23917 2017-06-21 Andrew Pinski <apinski@cavium.com>
23919 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23920 Increment Arith_shift and Arith_shift_reg by 1.
23921 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23923 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23924 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23925 (aarch64_strip_extend): Add new argument and test for it.
23926 (aarch64_cheap_mult_shift_p): New function.
23927 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23928 add a cost if it is true.
23929 Update calls to aarch64_strip_extend.
23930 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23932 2017-06-21 Andrew Pinski <apinski@cavium.com>
23934 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23936 (thunderxt88): Likewise.
23937 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23938 (thunderx_prefetch_tune): New variable.
23939 (thunderx2t99_prefetch_tune): Update for the correct values.
23940 (thunderxt88_tunings): New variable.
23941 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23942 generic_prefetch_tune.
23943 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23945 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23947 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23948 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23949 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23950 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23951 (aarch64_atomic_cas<mode>, GPI): Likewise.
23953 2017-06-21 Martin Liska <mliska@suse.cz>
23955 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23956 statements on cold and hot labels.
23957 * predict.c (tree_estimate_probability_bb): Remove the
23958 prediction from this place.
23960 2017-06-21 Martin Liska <mliska@suse.cz>
23962 PR tree-optimization/79489
23963 * gimplify.c (maybe_add_early_return_predict_stmt): New
23965 (gimplify_return_expr): Call the function.
23966 * predict.c (tree_estimate_probability_bb): Remove handling
23968 * predict.def: Update comment about early return predictor.
23969 * gimple-predict.h (is_gimple_predict): New function.
23970 * predict.def: Change default value of early return to 66.
23971 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23973 * passes.def: Put pass_strip_predict_hints to the beginning of
23976 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23978 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23979 FUNCTION_DECL declarations.
23980 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23982 (dwaf2out_decl): Likewise.
23983 * godump.c (go_early_global_decl): Skip call to the real debug hook
23984 for FUNCTION_DECL declarations.
23985 * passes.c (rest_of_decl_compilation): Skip call to the
23986 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23987 -fdump-go-spec is passed.
23989 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23991 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23992 Reorder for compactness.
23993 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23994 (def_builtin_pure, def_builtin_pure2): New functions.
23995 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23997 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23999 * match.pd (nop_convert): New predicate.
24000 ((A +- CST1) +- CST2): Allow some NOP conversions.
24002 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24005 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24006 with ctors/dtors if GOVD_SHARED is set.
24008 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24010 * config/aarch64/aarch64.md (movti_aarch64):
24011 Emit mov rather than orr.
24012 (movtf_aarch64): Likewise.
24013 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24014 Emit mov rather than orr.
24016 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24018 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24019 Swap alternatives, make integer dup more expensive.
24021 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24023 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24024 Return true for non-tls symbols.
24026 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24028 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24029 (cortex-a75): Likewise.
24030 (cortex-a75.cortex-a55): Likewise.
24031 * config/aarch64/aarch64-tune.md: Regenerate.
24032 * doc/invoke.texi (-mtune): Document new values for -mtune.
24034 2017-06-21 Tom de Vries <tom@codesourcery.com>
24036 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24037 stack_size feature.
24038 (Effective-Target Keywords, Other attributes): Suggest using
24039 dg-add-options stack_size feature to get stack limit in stack_size
24040 effective target documentation.
24042 2017-06-21 Julian Brown <julian@codesourcery.com>
24043 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24045 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24046 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24047 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24049 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24050 attribute type list for neon_multiply.
24051 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24052 attribute type list for neon_multiply.
24053 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24054 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24055 attribute type list for neon_multiply.
24056 * config/arm/types.md (crypto_pmull): Add.
24057 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24058 attribute type list.
24060 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24062 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24065 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24067 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24068 to make sure not to dereference a NULL cost_classes_ptr pointer.
24070 2017-06-20 Carl Love <cel@us.ibm.com>
24072 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24073 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24074 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24075 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24076 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24077 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24078 VMULOSW): New enum "unspec" values.
24079 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24080 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24081 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24082 altivec_vmulosw): New patterns.
24083 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24084 VMULOSW): Add definitions.
24086 2017-06-20 Julia Koval <julia.koval@intel.com>
24088 * config/i386/i386.c: Fix rounding expand for new pattern.
24089 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24091 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24093 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24094 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24096 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24098 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24101 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24103 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24104 architecture, then by implementer ID.
24105 * config/aarch64/aarch64-tune.md: Regenerate.
24107 2017-06-20 Richard Biener <rguenther@suse.de>
24109 PR middle-end/81097
24110 * fold-const.c (split_tree): Fold to type before negating.
24112 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24114 * diagnostic-show-locus.c
24115 (selftest::test_fixit_deletion_affecting_newline): New function.
24116 (selftest::diagnostic_show_locus_c_tests): Call it.
24118 2017-06-20 Andreas Schwab <schwab@suse.de>
24121 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24124 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24126 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24127 __ARM_FEATURE_COPROC according to support.
24129 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24131 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24132 Rewritten to avoid overflow for > 32-bit pointers.
24135 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24136 by removing enum keyword.
24137 (ubsan_type_descriptor): Likewise. Formatting fix.
24140 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24141 splitter): Require TARGET_SSE2 in the condition.
24143 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24146 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24147 for doing vector set of SFmode on ISA 3.0.
24148 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24149 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24151 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24152 SFmode value into a V4SF variable that was extracted from another
24153 V4SF variable without converting the element to double precision
24154 and back to single precision vector format.
24155 (vsx_insert_extract_v4sf_p9_2): Likewise.
24157 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24159 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24160 in UWHI to avoid undefined overflow.
24163 * ubsan.h (enum ubsan_encode_value_phase): New.
24164 (ubsan_encode_value): Change second argument to
24165 enum ubsan_encode_value_phase with default value of
24166 UBSAN_ENCODE_VALUE_GENERIC.
24167 * ubsan.c (ubsan_encode_value): Change second argument to
24168 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24169 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24170 create_tmp_var_raw instead of create_tmp_var and use a
24172 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24173 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24174 ubsan_encode_value callers.
24177 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24178 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24179 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24181 2017-06-19 Richard Biener <rguenther@suse.de>
24183 PR middle-end/81118
24184 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24185 estimates if we changed anything.
24187 2017-06-19 Richard Biener <rguenther@suse.de>
24189 PR tree-optimization/80887
24190 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24191 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24192 simplified lookups, then reset mprts_hook.
24193 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24195 (try_to_simplify): Likewise.
24197 2017-06-19 Martin Liska <mliska@suse.cz>
24200 * gimplify.c (gimplify_switch_expr):
24201 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24203 2017-06-19 Martin Liska <mliska@suse.cz>
24205 * doc/install.texi: Document that PGO runs in 4 stages.
24207 2017-06-19 Martin Liska <mliska@suse.cz>
24210 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24211 to dispatcher function name.
24212 * multiple_target.c (replace_function_decl): New function.
24213 (create_dispatcher_calls): Redirect both edges and references.
24215 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24217 * profile-count.c (profile_count::dump): Dump quality.
24218 (profile_count::differs_from_p): Update for unsigned val.
24219 * profile-count.h (profile_count_quality): New enum.
24220 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24222 2017-06-19 Richard Biener <rguenther@suse.de>
24224 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24225 struct function as arg.
24226 (estimate_numbers_of_iterations): Export overload with loop arg.
24227 (free_numbers_of_iterations_estimates_loop): Use an overload of
24228 free_numbers_of_iterations_estimates instead.
24229 * tree-cfg.c (remove_bb): Adjust.
24230 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24231 * tree-parloops.c (gen_parallel_loop): Likewise.
24232 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24234 (tree_unroll_loops_completely): Likewise.
24235 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24236 Use an overload instead and export.
24237 (estimated_loop_iterations): Adjust.
24238 (max_loop_iterations): Likewise.
24239 (likely_max_loop_iterations): Likewise.
24240 (estimate_numbers_of_iterations): Take struct function as arg
24242 (loop_exits_before_overflow): Adjust.
24243 (free_numbers_of_iterations_estimates_loop): Use an overload.
24244 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24245 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24247 2017-06-19 Richard Biener <rguenther@suse.de>
24250 * ipa-prop.c (find_constructor_constant_at_offset): Handle
24251 RANGE_EXPR conservatively.
24253 2017-06-16 Carl Love <cel@us.ibm.com>
24255 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24256 definitions for vec_float, vec_float2, vec_floato,
24257 vec_floate built-ins.
24258 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
24259 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
24261 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
24262 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
24263 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
24264 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
24265 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
24266 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
24267 vec_floato): Add builtin defines.
24268 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
24269 Update the built-in documentation file for the new built-in
24272 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24274 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
24275 (mthumb): Mark as the negative of -marm.
24277 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24279 * doc/invoke.texi (ARM Options, -mcpu): Document supported
24281 (ARM Options, -mtune): Document that this accepts the same
24282 extension options as -mcpu.
24283 (ARM Options, -mfpu): Document addition of -mfpu=auto.
24285 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24287 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
24288 permitted extensions.
24290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24292 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
24293 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
24294 (armv8-m.main): Add option +nodsp.
24295 * config/arm/arm-cpu-cdata.h: Regenerated.
24297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24299 * config/arm/t-fuchsia: New file.
24300 * config.gcc (arm*-*-fuchsia*): Use it.
24302 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24304 * config/arm/t-symbian: Rewrite for new option infrastructure.
24306 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24308 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
24309 (MULTILIB_REQUIRED): Likewise.
24311 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24313 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
24314 (MULTILIB_RESUE): Likewise.
24315 (MULTILIB_MATCHES): Likewise.
24316 (MULTLIB_REQUIRED): Likewise.
24318 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24320 * config/arm/t-rtems: Rewrite for new option framework.
24322 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24324 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
24325 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
24326 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
24327 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
24328 * config/arm/t-multilib: ... here.
24329 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
24330 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
24331 armv7-a and armv8*-a when A-profile libraries have not been built.
24332 * config/arm/t-rmprofile: Rewrite.
24334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24336 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
24337 with a backslash. Remove the backslash after substituting unescaped
24339 * doc/fragments.texi (MULTILIB_REUSE): Document it.
24341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24343 * config.gcc: (arm*-*-*): When building a-profile libraries, force
24344 the driver to pass through the default setting of -mfloat-abi.
24345 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
24347 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
24348 (all_feat_combs): New rule.
24349 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
24351 * config/arm/t-aprofile: Rewrite.
24353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24355 * config/arm/arm.h (FPUTYPE_AUTO): Define.
24356 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
24357 fpu is not specified by the user/command-line.
24358 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
24359 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
24360 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
24361 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
24362 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
24363 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
24365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24367 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
24368 * config/arm/t-arm-elf: Rewritten.
24370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24372 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
24373 have some floating-point instructions.
24374 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
24375 (TARGET_MAYBE_HARD_FLOAT): New macro.
24376 * config/arm/arm-builtins.c (arm_init_builtins): Use
24377 TARGET_MAYBE_HARD_FLOAT.
24378 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
24380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24382 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
24383 (configargs.h): Include it.
24384 (arm_print_hint_for_fpu_option): New function.
24385 (arm_parse_fpu_option): New function.
24386 (candidate_extension): New class.
24387 (arm_canon_for_multilib): New function.
24388 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
24389 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
24390 (ARCH_CANONICAL_SPECS): New macro.
24391 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
24393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24395 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
24396 are set after handling multilib fragments. Set target_cpu_default2
24399 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24401 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
24403 (arm*-*-*): Set target_cpu_default2 to a quoted string.
24404 * config/arm/parsecpu.awk (check_cpu): Validate any extension
24406 (check_arch): Likewise.
24407 * config/arm/arm.c (arm_configure_build_target): Handle
24408 TARGET_CPU_DEFAULT being a string constant. Scan any feature
24409 options in the default.
24411 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24413 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
24414 when an option is an alias of another.
24415 * config/arm/parsecpu.awk (optalias): New parser token.
24416 (gen_comm_data): Mark non-alias options as such. Emit entries
24417 for extension aliases.
24418 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
24419 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
24420 (armv6kz, armv6zk, armv6t2): Likewise.
24421 (armv7): Make vfpv3-d16 an alias.
24422 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
24424 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
24425 Sort in canonical order.
24426 (armv8-a): Sort in canonical order.
24427 (armv8.1-a, armv8.2-a): Likewise.
24428 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
24430 (cortex-a9): Sort in canonical order.
24431 * config/arm/arm.c (selftests.h): Include it.
24432 (arm_test_cpu_arch_data): New function.
24433 (arm_run_self_tests): New function.
24434 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
24435 (targetm): Move declaration to the end of the file.
24436 * arm-cpu-cdata.h: Regenerated.
24438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24440 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
24441 call to target_mode_check describing the type of option passed.
24442 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
24443 (arm_target_thumb_only): Use arm_parse_arch_option_name or
24444 arm_parse_cpu_option_name to match parameters against list of
24446 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
24447 arm_arch_core_flags data structure.
24448 * config/arm/arm-cpu_cdata.h: Regenerated.
24450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24452 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
24454 (arm_print_hint_for_cpu_option): Likewise.
24455 (arm_print_hint_for_arch_option): Likewise.
24456 (arm_parse_cpu_option_name): Likewise.
24457 (arm_parse_arch_option_name): Likewise.
24458 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
24459 of entries in the all_fpus list.
24460 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
24461 (arm_parse_cpu_option_name): Declare.
24462 (arm_parse_arch_option_name): Declare.
24463 (arm_parse_option_features): Declare.
24464 (arm_intialize_isa): Declare.
24465 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
24467 (gen_comm_data): ... here. Make definitions non-static.
24468 * config/arm/arm-cpu-data.h: Regenerated.
24469 * config/arm/arm-cpu-cdata.h: Regenerated.
24471 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24473 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
24474 (cpu_arch_extension): New structure.
24475 (cpu_arch_option, arch_option, cpu_option): New structures.
24476 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
24477 architecture types.
24478 (gen_data): Generate new format data tables.
24479 * config/arm/arm.c (cpu_tune): New structure.
24480 (cpu_option, processors): Delete.
24481 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
24482 (arm_print_hint_for_cpu_option): ... this and ...
24483 (arm_print_hint_for_arch_option): ... this.
24484 (arm_parse_arch_cpu_name): Delete. Replace with ...
24485 (arm_parse_cpu_option_name): ... this and ...
24486 (arm_parse_arch_option_name): ... this.
24487 (arm_unrecognized_feature): Change type of target parameter to
24489 (arm_parse_arch_cpu_features): Delete. Replace with ...
24490 (arm_parse_option_features): ... this.
24491 (arm_configure_build_target): Rework to use new configuration data
24493 (arm_print_tune_info): Rework for new configuration data tables.
24494 * config/arm/arm-cpu-data.h: Regenerated.
24495 * config/arm/arm-cpu.h: Regenerated.
24497 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24499 * Makefile.in (OBJS): Move sbitmap.o from here ...
24500 (OBJS-libcommon): ... to here.
24502 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24504 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
24505 (ISA_ALL_CRYPTO): New macro.
24506 (ISA_ALL_SIMD): New macro
24507 (ISA_ALL_FP): New macro.
24508 * config/arm/arm.c (fpu_bitlist): Update initializer.
24509 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
24511 (arm9e): Add fpu. Add option for nofp
24512 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
24513 (arm926ej-s, arm1026ej-s): Likewise.
24514 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
24515 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
24516 neon-fp16, neon-vfpv4, nofp and nosimd.
24517 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
24518 (cortex-a8): Add fpu. Add option for nofp.
24519 (cortex-a9): Add fpu. Add options for nosimd and nofp.
24520 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
24521 (cortex-r4f): Add fpu.
24522 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
24523 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
24525 (cortex-r8): Likewise.
24526 (cortex-m4): Add fpu. Add option for nofp.
24527 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
24528 (cortex-a17.cortex-a7): Likewise.
24529 (cortex-a32): Add fpu. Add options for crypto and nofp.
24530 (cortex-a35, cortex-a53): Likewise.
24531 (cortex-a57): Add fpu. Add option for crypto.
24532 (cortex-a72, cortex-a73): Likewise.
24533 (exynos-m1): Likewise.
24534 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
24535 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
24536 (cortex-m33): Add fpu. Add option for nofp.
24537 * config/arm/arm-cpu-cdata.h: Regenerated
24538 * config/arm/arm-cpu-data.h: Regenerated.
24540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24542 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
24543 (armv5te, armv5tej): Likewise.
24544 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
24545 (armv7): Add options fp and vfpv3-d16.
24546 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
24547 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
24549 (armv7ve): Likewise.
24550 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
24551 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
24552 (armv8-a): Add nocrypto option.
24553 (armv8.1-a, armv8.2-a): Likewise.
24554 (armv8-m.main): add options fp, fp.dp and nofp.
24556 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24558 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
24560 (armv8-a+crc): Delete.
24561 (armv8.1-a): Add options simd, crypto and nofp.
24562 (armv8.2-a): Add options fp16, simd, crypto and nofp.
24563 (armv8.2-a+fp16): Delete.
24564 (armv8-m.main): Add option dsp.
24565 (armv8-m.main+dsp): Delete.
24566 (cortex-a8): Add fpu. Add nofp option.
24567 (cortex-a9): Add fpu. Add nofp and nosimd options.
24568 * config/arm/parsecpu.awk (gen_data): Generate option tables and
24569 link to main cpu and architecture data structures.
24570 (gen_comm_data): Only put isa attributes from the main architecture
24572 (option): New statement for architecture and CPU entries.
24573 * arm.c (struct cpu_option): New structure.
24574 (struct processors): Add entry for options.
24575 (arm_unrecognized_feature): New function.
24576 (arm_parse_arch_cpu_name): Ignore any characters after the first
24578 (arm_parse_arch_cpu_feature): New function.
24579 (arm_configure_build_target): Separate out any CPU and architecture
24580 features and parse separately. Don't error out if -mfpu=auto is
24581 used with only an architecture string.
24582 (arm_print_asm_arch_directives): New function.
24583 (arm_file_start): Call it.
24584 * config/arm/arm-cpu-cdata.h: Regenerated.
24585 * config/arm/arm-cpu-data.h: Likewise.
24586 * config/arm/arm-tables.opt: Likewise.
24588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24590 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
24591 assembler when it is not -mfpu=auto.
24593 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24595 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
24596 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
24597 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
24598 (ASM_CPU_SPEC): Rewrite.
24599 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
24600 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
24601 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
24602 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
24603 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
24604 copied string is NUL-terminated. Also strip any characters prefixed
24606 (arm_rewrite_selected_arch): New function.
24607 (arm_rewrite_march): New function.
24609 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
24611 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
24612 (x_arm_cpu_string, x_arm_tune_string): Likewise.
24613 (march, mcpu, mtune): Convert to string-based options.
24614 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
24615 (arm_parse_arch_cpu_name): New function.
24616 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
24617 identify selected architecture or CPU.
24618 (arm_option_save): New function.
24619 (TARGET_OPTION_SAVE): Redefine.
24620 (arm_option_restore): Restore string options.
24621 (arm_option_print): Print string options.
24623 2017-06-16 Martin Sebor <msebor@redhat.com>
24625 PR tree-optimization/80933
24626 PR tree-optimization/80934
24627 * builtins.c (fold_builtin_3): Do not handle bcmp here.
24628 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
24629 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
24630 (gimple_fold_builtin): Call them.
24632 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24634 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
24635 as unlikely; update profile.
24637 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24639 * predict.c (force_edge_cold): Handle declaring edges impossible
24642 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24644 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
24646 (try_unroll_loop_completely): Fix reporting.
24648 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24650 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
24652 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
24655 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
24656 if given a non-constant argument for an intrinsic which requires a
24659 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24661 * profile.c (compare_freqs): New function.
24662 (branch_prob): Sort edge list.
24663 (find_spanning_tree): Assume that the list is priority sorted.
24665 2017-06-16 Richard Biener <rguenther@suse.de>
24667 PR tree-optimization/81090
24668 * passes.def (pass_record_bounds): Remove.
24669 * tree-pass.h (make_pass_record_bounds): Likewise.
24670 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
24671 make_pass_record_bounds): Likewise.
24672 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
24673 not free niter estimates at the beginning but at the end.
24674 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
24676 2017-06-16 Richard Biener <rguenther@suse.de>
24678 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
24679 initializer to workaround ICE in host GCC 4.8.
24681 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
24683 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
24685 (clone_inlined_nodes): Update.
24687 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24689 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
24690 prefetch settings, and enable prefetching by default at -O3.
24692 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24694 * config/aarch64/aarch64.c (aarch64_override_options_internal):
24695 Set flag_prefetch_loop_arrays according to tuning data.
24697 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24699 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
24700 New tune structure.
24701 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
24702 [Unrelated to main purpose of the patch] Place the pointer field last
24703 to enable type checking errors when tune structure are wrongly merged.
24704 * config/aarch64/aarch64.c (generic_prefetch_tune,)
24705 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
24706 (thunderx2t99_prefetch_tune): New tune constants.
24707 (tune_params *_tunings): Update all tunings (no functional change).
24708 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
24709 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
24710 from tunings structures.
24712 2017-06-16 Jakub Jelinek <jakub@redhat.com>
24715 * ubsan.c (instrument_null): Add T argument, use it instead
24716 of computing it based on IS_LHS.
24717 (instrument_object_size): Likewise.
24718 (pass_ubsan::execute): Adjust instrument_null and
24719 instrument_object_size callers to pass gimple_get_lhs or
24720 gimple_assign_rhs1 result to it. Use instrument_null instead of
24721 calling get_base_address and instrument_mem_ref. Handle
24722 aggregate call arguments for object-size sanitization.
24724 2017-06-16 Yury Gribov <tetra2005@gmail.com>
24726 PR tree-optimization/81089
24727 * tree-vrp.c (is_masked_range_test): Validate operands of
24730 2017-06-15 Martin Sebor <msebor@redhat.com>
24733 * dumpfile.c (dump_register): Avoid calling memset to initialize
24734 a class with a default ctor.
24735 * gcc.c (struct compiler): Remove const qualification.
24736 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
24737 * hash-table.h: Ditto.
24738 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
24740 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
24741 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
24743 * params.h (struct param_info): Make struct members non-const.
24744 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
24745 with default initialization.
24746 * vec.h (vec_copy_construct, vec_default_construct): New helper
24748 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
24749 with vec_copy_construct.
24750 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
24751 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
24752 * doc/invoke.texi (-Wclass-memaccess): Document.
24754 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24756 * emit-rtl.h (is_leaf): Update comment about local
24757 register allocator.
24759 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
24762 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
24763 for a variable to have a section before checking if the section has a
24765 Set section to.persistent if persistent attribute is set.
24766 Warn if .persistent attribute is used on an automatic variable.
24768 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
24770 PR rtl-optimization/80474
24771 * reorg.c (update_block): Do not ignore instructions in a delay slot.
24773 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
24775 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
24778 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
24780 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
24781 (casesi): Emit bounds checking as RTL.
24782 (casesi_internal_mips16_<mode>): Remove bounds checking.
24784 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
24786 * config/xtensa/xtensa.c (xtensa_option_override): Append
24787 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
24788 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
24789 xtensa_doloop_hooks): Define unconditionally.
24790 (xtensa_reorg_loops): Only call reorg_loops in the presence of
24792 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
24793 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
24794 for it in xtensa_option_override.
24795 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
24796 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
24798 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
24800 * doc/cppopts.texi: Document '-' special value to -MF.
24802 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
24804 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
24805 (cortex_a53_fconst): Likewise.
24806 (cortex_a53_fpmul): Likewise.
24807 (cortex_a53_f_load_64): Likewise.
24808 (cortex_a53_f_load_many): Likewise.
24809 (cortex_a53_advsimd_alu): Likewise.
24810 (cortex_a53_advsimd_alu_q): Likewise.
24811 (cortex_a53_advsimd_mul): Likewise.
24812 (cortex_a53_advsimd_mul_q): Likewise.
24813 (fpmac bypass): Add new bypass for fpmac-fpmac case.
24814 Add missing fmul, r2f_cvt and fconst cases.
24816 2017-06-14 Richard Biener <rguenther@suse.de>
24818 PR middle-end/81088
24819 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
24821 (fold_binary_loc): When associating do not treat pre-existing
24822 TREE_OVERFLOW on literal constants as a reason to allow
24823 TREE_OVERFLOW on associated literal constants.
24825 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24827 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
24828 (MASK_FEATURES): New macro.
24829 * config/sparc/sparc.c (sparc_option_override): Remove the special
24830 handling of -mfpu and generalize it to all MASK_FEATURES switches.
24832 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24834 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
24835 a division of 0 if non-call exceptions are enabled.
24837 2017-06-14 Andrew Pinski <apinski@cavium.com>
24838 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24841 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
24842 Improve vector initialization code gen for only variable case.
24844 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
24846 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
24848 2017-06-14 Richard Biener <rguenther@suse.de>
24850 PR tree-optimization/81083
24851 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
24854 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24856 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
24857 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
24858 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
24859 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
24860 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
24861 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
24863 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24865 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
24866 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
24868 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24870 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
24872 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24874 * config/rs6000/t-rtems: Don't handle SPE.
24876 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24878 * config/rs6000/t-linux: Don't handle SPE.
24880 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24882 * config/rs6000/eabispe.h: Delete file.
24884 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24886 * config/rs6000/t-spe: Delete file.
24888 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24890 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24891 (rs6000_legitimate_offset_address_p): Return false for anything in
24892 V2SImode or V2SFmode.
24894 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24896 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24897 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24898 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24900 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24901 (rs6000_legitimate_offset_address_p): Ditto.
24902 (rs6000_emit_move): Ditto.
24903 (rs6000_init_builtins): Remove V4HI_type_node.
24905 2017-06-13 Martin Liska <mliska@suse.cz>
24908 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24909 (gate_asan): Likewise.
24910 * asan.h (asan_no_sanitize_address_p): Remove the function.
24911 (sanitize_flags_p): New function.
24912 * builtins.def: Fix coding style.
24913 * common.opt: Use renamed enum value.
24914 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24915 * doc/extend.texi: Document no_sanitize attribute.
24916 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24917 to SANITIZE_UNDEFINED_NONDEFAULT.
24918 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24919 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24920 Use sanitize_flags_p.
24921 * gimplify.c (gimplify_function_tree): Likewise.
24922 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24923 * opts.c (parse_no_sanitize_attribute): New function.
24924 (common_handle_option): Use renamed enum value.
24925 * opts.h (parse_no_sanitize_attribute): Declare.
24926 * tree.c (sanitize_flags_p): New function.
24927 * tree.h: Declared here.
24928 * tsan.c: Use sanitize_flags_p.
24929 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24930 (instrument_mem_ref): Likewise.
24931 (instrument_bool_enum_load): Likewise.
24932 (do_ubsan_in_current_function): Remove the function.
24933 (pass_ubsan::execute): Use sanitize_flags_p.
24934 * ubsan.h: Remove do_ubsan_in_current_function
24935 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24936 (dump_function_to_file): Use it here.
24938 2017-06-13 Martin Jambor <mjambor@suse.cz>
24940 PR tree-optimization/80803
24941 PR tree-optimization/81063
24942 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24943 (propagate_subaccesses_across_link): Enqueue subtree whenever
24944 necessary instead of relying on the caller.
24946 2017-06-13 Martin Jambor <mjambor@suse.cz>
24948 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24949 that have a first_link.
24950 (sort_and_splice_var_accesses): Do not check first_link before
24952 (subtree_mark_written_and_enqueue): Likewise.
24953 (propagate_all_subaccesses): Likewise and do not stop at first
24954 parent with a first_link.
24956 2017-06-13 Martin Jambor <mjambor@suse.cz>
24958 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24961 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24963 * match.pd: New pattern.
24965 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24967 * tree-vrp.c (is_masked_range_test): New function.
24968 (register_edge_assert_for): Determine ranges for
24971 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24973 PR tree-optimization/67328
24974 * fold-const.c (maskable_range_p): New function.
24975 (build_range_check): Generate bittests if possible.
24977 2017-06-13 Martin Liska <mliska@suse.cz>
24979 * gimple-pretty-print.c (dump_probability): Add new argument.
24980 (dump_edge_probability): Dump both probability and count.
24981 (dump_gimple_label): Likewise.
24982 (dump_gimple_bb_header): Likewise.
24984 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24987 * config/avr/avr-devices.c: Fix indentation.
24988 * config/avr/gen-avr-mmcu-specs.c: Dito.
24990 2017-06-13 Richard Biener <rguenther@suse.de>
24992 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24993 instead get vector type from stmt_info.
24994 (vectorizable_reduction): Adjust. Remove dead code.
24996 2017-06-13 Richard Biener <rguenther@suse.de>
24998 PR middle-end/81065
24999 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25000 case of C * (x * C2 + C3).
25001 (fold_addr_of_array_ref_difference): Properly fold index difference.
25003 2017-06-12 David S. Miller <davem@davemloft.net>
25006 * config/sparc/sparc.md (return expander): Emit frame blockage if
25007 function uses alloca.
25009 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25011 * combine.c (make_field_assignment): Check len rather than the mode
25012 precision when calling force_to_mode.
25014 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25016 Support multilibs and devices that see flash in RAM address range.
25019 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25020 (avr_mcu_t) <flash_pm_offset>: New field.
25021 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25022 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25023 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25024 (AVR_TINY_PM_OFFSET): Remove macro.
25025 * config/avr/avr.opt (-mshort-calls): New option.
25026 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25027 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25028 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25029 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25030 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25031 instead of avr_arch->have_jmp_call.
25032 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25033 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25034 avr_arch->flash_pm_offset to define.
25035 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25036 new field flash_pm_offset. Add entry for avrxmega3.
25037 (avr_texinfo): Add entry for avrxmega3.
25038 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25039 attiny212, attiny214,
25040 attiny412, attiny414, attiny416, attiny417,
25041 attiny814, attiny816, attiny817,
25042 attiny1614, attiny1616, attiny1617,
25043 attiny3214, attiny3216, attiny3217.
25044 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25045 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25046 (avr_print_operand_address) [AVR_TINY]: Same.
25047 (avr_asm_init_sections) <readonly_data_section>: Only patch
25048 callback if avr_arch->flash_pm_offset = 0.
25049 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25050 for rodata if avr_arch->flash_pm_offset != 0.
25051 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25052 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25053 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25054 (m_options): Append opt_rcall.
25055 (m_dirnames): Append dir_rcall.
25056 * config/avr/t-multilib: Regenerate.
25058 * configure.ac [target=avr]: Check whether avrxmega3 default
25059 linker description file works as needed.
25060 * configure: Regenerate.
25061 * doc/avr-mmcu.texi: Regenerate.
25062 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25063 <__AVR_ARCH__>: Document avrxmega3 and 103.
25064 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25065 <__AVR_SHORT_CALLS__>: Document it.
25066 <__AVR_PM_BASE_ADDRESS__>: Document it.
25067 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25068 (AVR Variable Attributes) <progmem>: Document this is
25069 not needed for avrxmega3.
25070 (AVR Named Address Spaces) <__flash>: Dito.
25072 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25074 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25076 2017-06-12 Doug Rupp <rupp@adacore.com>
25078 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25079 Append vxworks-stdint.h to the tm_file list.
25080 * config/vxworks-stdint.h: New file.
25082 2017-06-12 Martin Liska <mliska@suse.cz>
25084 PR tree-optimization/81041
25085 * tree-profile.c (gimple_gen_ic_func_profiler):
25086 Create an extra BB in profile-generate
25087 (gimple_gen_time_profiler): Likewise.
25089 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25091 PR tree-optimization/81003
25092 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25093 (update_range_test): Use it instead of force_gimple_operand_gsi.
25095 2017-06-12 Richard Biener <rguenther@suse.de>
25097 PR tree-optimization/81053
25098 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25099 with backedge value not defined in loop. Simplify def stmt
25102 2017-06-11 Tom de Vries <tom@codesourcery.com>
25105 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25107 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25108 nvptx_cannot_force_const_mem.
25110 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25112 * opts.c (finish_options): Move test for flag_split_stack after
25113 it has been initialized.
25115 2017-06-11 Jason Merrill <jason@redhat.com>
25117 * tree.h (id_equal): New.
25118 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25119 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25120 instead of strcmp of IDENTIFIER_POINTER.
25122 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25124 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25125 (mark_all_inlined_calls_cdtor): Fix formating.
25126 (inline_transform): Rescale profile before inlining.
25128 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25130 * cgraph.h (cgraph_edge::clone): Update prototype.
25131 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25132 (cgraph_node::create_clone): Update.
25133 (cgraph_node::create_version_clone): Update.
25134 * tree-inline.c (copy_bb): Update.
25135 (expand_call_inline): Update.
25137 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25139 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25140 factored out from ...
25141 (rs6000_emit_prologue): ... here.
25143 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25145 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25146 factored out from ...
25147 (rs6000_emit_prologue): ... here.
25149 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25151 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25153 (handle_missing_profiles): Fix computation of tp_first_run.
25154 (counts_to_freqs): Do not touch freqs when count is 0.
25156 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25158 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25161 2017-06-10 Tom de Vries <tom@codesourcery.com>
25163 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25164 attributes): Document signal effective target.
25166 2017-06-10 Tom de Vries <tom@codesourcery.com>
25168 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25169 Document effective target stack_size.
25171 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25173 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25174 to the edit_context if they can be auto-applied.
25176 2017-06-9 Ian Lance Taylor <iant@golang.org>
25178 * opts.c (finish_options): If -fsplit-stack, disable implicit
25179 -forder-blocks-and-partition.
25180 * doc/invoke.texi (Optimize Options): Document that when using
25181 -fsplit-stack -forder-blocks-and-partition is not implicitly
25184 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25186 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25187 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25188 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25189 * builtins.def (abort, trap, unreachable): Declare cold.
25190 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25191 * tree-core.h (ECF_COLD): New.
25192 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25193 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25195 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25197 * predict.c (unlikely_executed_stmt_p): Cleanup.
25199 2017-06-09 Richard Biener <rguenther@suse.de>
25201 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25202 model if the ref is always written to.
25204 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25206 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25208 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25210 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25213 2017-06-09 Tom de Vries <tom@codesourcery.com>
25216 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25217 "target cannot support label values" when encountering LABEL_REF.
25219 2017-06-09 Martin Liska <mliska@suse.cz>
25221 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25222 (gimple_gen_ic_func_profiler): Emit direct comparison
25223 of __gcov_indirect_call_callee with NULL.
25224 (gimple_gen_time_profiler): Change probability from
25225 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25227 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25229 * profile.c (edge_gcov_counts): Turn to pointer.
25230 (compute_branch_probabilities, compute_branch_probabilities): Update.
25231 (branch_prob): Do not clear edge_gcov_count.
25232 * profile.h (edge_gcov_counts): Turn to pointer.
25233 (edge_gcov_count): Update.
25235 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25237 * gimple.h (gimple_check_failed): Mark cold.
25239 2017-06-09 Richard Biener <rguenther@suse.de>
25241 PR tree-optimization/66623
25242 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25243 refactor check_reduction into two parts, properly computing
25244 whether we have to check reduction validity for outer loop
25247 2017-06-09 Richard Biener <rguenther@suse.de>
25249 PR tree-optimization/79483
25250 * graphite-scop-detection.c (order): New global.
25251 (get_order): Compute bb to order mapping that satisfies code
25252 generation constraints.
25253 (cmp_pbbs): New helper.
25254 (build_scops): Start domwalk at entry block, sort generated
25257 2017-06-09 Richard Biener <rguenther@suse.de>
25259 PR middle-end/81007
25260 * ipa-polymorphic-call.c
25261 (ipa_polymorphic_call_context::restrict_to_inner_class):
25262 Skip FIELD_DECLs with error_mark_node type.
25263 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
25266 2017-06-09 Martin Liska <mliska@suse.cz>
25268 * predict.c (struct branch_predictor): New struct.
25269 (test_prediction_value_range): New test.
25270 (predict_c_tests): New function.
25271 * selftest-run-tests.c (selftest::run_tests): Run the function.
25272 * selftest.h: Declare new tests.
25274 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
25277 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
25278 gen_add3_insn did not fail.
25279 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
25280 r0, construct that number in a temporary reg and add that reg to r0.
25281 If asked to put the result in r0 as well, fail.
25283 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
25285 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25286 for early expansion of vec_eqv.
25288 2017-06-08 Jakub Jelinek <jakub@redhat.com>
25290 PR middle-end/81005
25291 * ubsan.c (instrument_null): Avoid pointless code temporary.
25292 (pass_ubsan::execute): Instrument aggregate arguments of calls.
25294 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
25298 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
25301 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
25304 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25306 * predict.c (unlikely_executed_edge_p): Move ahead.
25307 (probably_never_executed_edge_p): Use it.
25309 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25311 PR middle-end/79988
25312 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
25313 gimple_call_builtin_p call.
25315 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25317 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
25318 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
25319 rtl_check_failed_type2, rtl_check_failed_code1,
25320 rtl_check_failed_code2, rtl_check_failed_code_mode,
25321 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
25322 rtvec_check_failed_bounds, rtl_check_failed_flag,
25323 _fatal_insn_not_found, _fatal_insn): Likewise.
25324 * tree.h (tree_contains_struct_check_failed,
25325 tree_check_failed, tree_not_check_failed,
25326 tree_class_check_failed, tree_range_check_failed,
25327 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
25328 tree_vec_elt_check_failed, phi_node_elt_check_failed,
25329 tree_operand_check_failed, omp_clause_check_failed,
25330 omp_clause_operand_check_failed, omp_clause_range_check_failed):
25333 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25335 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
25336 flag_branch_probabilities.
25337 * ipa-inline.c (edge_badness): Likewise.
25338 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
25339 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
25340 * predict.c (maybe_hot_frequency_p): Likewise.
25341 (probably_never_executed): Likewise.
25342 * sched-ebb.c (schedule_ebbs): Likewise.
25343 * sched-rgn.c (find_single_block_region): Likewise.
25344 * tracer.c (tail_duplicate): Likewise.
25346 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25348 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
25349 longer requires x_flag_profile_use.
25351 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25353 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
25354 instead of flag_reorder_blocks_and_partition.
25355 * dbxout.c (dbxout_function_end): Likewise.
25356 * dwarf2out.c (gen_subprogram_die): Likewise.
25357 * haifa-sched.c (sched_create_recovery_edges): Likewise.
25358 * hw-doloop.c (reorg_loops): Likewise.
25359 * varasm.c (assemble_start_function,
25360 assemble_end_function): Likewise.
25361 (decide_function_section): Do not check for
25362 flag_reorder_blocks_and_partition.
25364 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
25366 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25368 (chkp_get_hard_register_fake_addr_expr): Ditto.
25369 (chkp_build_addr_expr): Add check for hard reg case.
25370 (chkp_parse_array_and_component_ref): Ditto.
25371 (chkp_find_bounds_1): Ditto.
25372 (chkp_process_stmt): Don't generate bounds store for
25375 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25377 * predict.c (maybe_hot_bb_p): Do not check profile status.
25378 (maybe_hot_edge_p): Likewise.
25379 (probably_never_executed): Check for zero counts even if profile
25381 (unlikely_executed_edge_p): New function.
25382 (unlikely_executed_stmt_p): New function.
25383 (unlikely_executed_bb_p): New function.
25384 (set_even_probabilities): Use unlikely predicates.
25385 (combine_predictions_for_bb): Likewise.
25386 (predict_paths_for_bb): Likewise.
25387 (predict_paths_leading_to_edge): Likewise.
25388 (determine_unlikely_bbs): New function.
25389 (estimate_bb_frequencies): Use it.
25390 (compute_function_frequency): Use zero counts even if profile is
25392 * profile-count.h: Fix typo.
25394 2017-08-08 Julia Koval <julia.koval@intel.com>
25396 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
25397 _mm512_mask_cvtsepi16_storeu_epi8,
25398 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
25399 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
25400 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
25401 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
25402 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
25403 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
25404 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
25405 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
25406 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
25407 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
25408 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
25409 __builtin_ia32_pmovuswb256mem_mask,
25410 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
25411 __builtin_ia32_pmovwb512mem_mask): New builtins.
25413 2017-08-08 Julia Koval <julia.koval@intel.com>
25415 PR target/73350,80862
25416 * config/i386/subst.md (round): Fix round pattern.
25417 * config/i386/i386.c (ix86_erase_embedded_rounding):
25418 Fix erasing rounding for the fixed pattern.
25420 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
25422 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
25424 2017-06-08 Martin Liska <mliska@suse.cz>
25426 PR gcov-profile/80911
25427 * gcov.c (block_info::block_info): New constructor.
25429 2017-06-07 Carl Love <cel@us.ibm.com>
25431 * config/rs6000/rs6000-c: The return type of the following
25432 built-in functions was implemented as int not long long. Fix sign
25433 of return value for the unsigned version of vec_mulo and vec_mule.
25434 vector unsigned long long vec_bperm (vector unsigned long long,
25435 vector unsigned char)
25436 vector signed long long vec_mule (vector signed int,
25438 vector unsigned long long vec_mule (vector unsigned int,
25439 vector unsigned int)
25440 vector signed long long vec_mulo (vector signed int,
25442 vector unsigned long long vec_mulo (vector unsigned int,
25443 vector unsigned int)
25444 * doc/extend.texi: Fix the documentation for the built-in
25447 2017-06-07 Carl Love <cel@us.ibm.com>
25450 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
25453 2017-06-07 Carl Love <cel@us.ibm.com>
25455 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
25456 support, Generate doublehv for signed int/float for BE case only.
25458 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
25460 * doc/invoke.texi (mcx16): Rewrite.
25462 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25464 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
25465 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
25466 *mov<mode>_softfloat, and an anonymous splitter): Use
25467 nonimmediate_operand instead of rs6000_nonimmediate_operand.
25469 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25471 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
25473 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25474 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
25475 (rs6000_debug_reg_global): Adjust.
25476 (rs6000_init_hard_regno_mode_ok): Adjust.
25477 (rs6000_dbx_register_number): Adjust.
25478 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
25479 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25480 Remove SPE_ACC and SPEFSCR.
25481 (REG_ALLOC_ORDER): Ditto.
25482 (FRAME_POINTER_REGNUM): Change to 111.
25483 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
25484 (REG_CLASS_NAMES): Ditto.
25485 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
25486 (REGISTER_NAMES): Ditto.
25487 (ADDITIONAL_REG_NAMES): Ditto.
25488 (rs6000_reg_names): Ditto.
25489 * config/rs6000/rs6000.md: Renumber some register number
25492 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25494 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
25496 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
25497 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
25499 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
25500 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
25501 Delete the SPE high registers.
25502 (REG_ALLOC_ORDER): Ditto.
25503 (enum reg_class): Remove SPE_HIGH_REGS.
25504 (REG_CLASS_NAMES): Ditto.
25505 (REG_CLASS_CONTENTS): Delete the SPE high registers.
25506 (REGISTER_NAMES): Ditto.
25507 (rs6000_reg_names): Ditto.
25508 * doc/tm.texi.in: Remove SPE as example.
25509 * doc/tm.texi: Regenerate.
25511 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25513 * config/rs6000/8540.md (ppc8540_brinc): Delete.
25514 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
25515 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
25516 * config/rs6000/rs6000.md (type): Remove "brinc".
25518 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25520 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
25521 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
25522 * config/rs6000/linuxspe.h: Delete file.
25523 * config/rs6000/rs6000.md: Don't include spe.md.
25524 * config/rs6000/spe.h: Delete file.
25525 * config/rs6000/spe.md: Delete file.
25526 * config/rs6000/t-rs6000: Remove spe.md.
25528 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25530 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
25531 (reg_or_none500mem_operand): Delete.
25532 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
25533 instead of reg_or_none500mem_operand.
25535 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25537 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25538 handling of SPE flags.
25539 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
25541 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25543 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
25545 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
25546 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
25547 paired_divv2sf3): Similar.
25548 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
25549 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
25550 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
25552 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
25553 Rename the paired_* instruction patterns.
25554 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
25556 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
25557 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
25558 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
25559 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
25560 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
25561 PAIRED_VECTOR_MODE.
25562 (struct machine_function): Delete field spe_insn_chain_scanned_p.
25563 (spe_func_has_64bit_regs_p): Delete.
25564 (spe_expand_predicate_builtin): Delete.
25565 (spe_expand_evsel_builtin): Delete.
25566 (TARGET_DWARF_REGISTER_SPAN): Do not define.
25567 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
25568 (invalid_e500_subreg): Delete.
25569 (rs6000_legitimize_address): Always force_reg op2 as well, for
25570 paired single memory accesses.
25571 (rs6000_member_type_forces_blk): Delete.
25572 (rs6000_spe_function_arg): Delete.
25573 (rs6000_expand_unop_builtin): Delete SPE handling.
25574 (rs6000_expand_binop_builtin): Ditto.
25575 (spe_expand_stv_builtin): Delete.
25576 (bdesc_2arg_spe): Delete.
25577 (spe_expand_builtin): Delete.
25578 (spe_expand_predicate_builtin): Delete.
25579 (spe_expand_evsel_builtin): Delete.
25580 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
25581 (spe_init_builtins): Delete.
25582 (spe_func_has_64bit_regs_p): Delete.
25583 (savres_routine_name): Delete "info" parameter. Adjust callers.
25584 (rs6000_emit_stack_reset): Ditto.
25585 (rs6000_dwarf_register_span): Delete.
25586 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
25587 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
25588 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
25590 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
25592 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
25593 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
25594 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
25595 mulv2sf3, divv2sf3): Delete expanders.
25597 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25599 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
25601 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25603 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
25604 * config/rs6000/rs6000.c: Ditto.
25606 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25608 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
25609 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
25610 comparison_operator.
25612 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25614 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
25615 * config/rs6000/rs6000.opt: Ditto.
25616 * config/rs6000/t-rtems: Ditto.
25618 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25620 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
25621 TARGET_E500_SINGLE by 0, simplify.
25622 * config/rs6000/rs6000.c: Ditto.
25623 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
25624 (spe_build_register_parallel): Delete.
25625 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
25626 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
25627 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
25628 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
25629 (E500_CONVERT): Delete.
25630 * config/rs6000/spe.md: Remove many patterns and all define_constants.
25632 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
25634 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
25635 * config/rs6000/dfp.md: Ditto.
25636 (negdd2, *negdd2_fpr): Merge.
25637 (absdd2, *absdd2_fpr): Merge.
25638 (negtd2, *negtd2_fpr): Merge.
25639 (abstd2, *abstd2_fpr): Merge.
25640 * config/rs6000/e500.h: Delete file.
25641 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
25642 TARGET_FPRS by 1 and simplify.
25643 * config/rs6000/rs6000-c.c: Ditto.
25644 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
25645 TARGET_DF_SPE by 0.
25646 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
25648 * config/rs6000/rs6000.md: Ditto.
25649 (floatdidf2, *floatdidf2_fpr): Merge.
25650 (move_from_CR_gt_bit): Delete.
25651 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
25652 (E500_CR_IOR_COMPARE): Delete.
25653 (All patterns that require !TARGET_FPRS): Delete.
25654 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
25656 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25658 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
25660 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25662 * graphds.c (add_edge): Intitialize edge's attached data.
25663 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
25664 pointer parameter. Call pointed function on each edge during
25665 graph traversing. Skip traversing the edge when the function
25667 (graphds_dfs, graphds_scc): Ditto.
25668 (for_each_edge): New parameter. Pass the new parameter to callback
25670 * graphds.h (skip_edge_callback): New function pointer type.
25671 (graphds_dfs, graphds_scc): New function pointer parameter.
25672 (graphds_edge_callback, for_each_edge): New parameter.
25674 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25676 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
25677 out code checking if runtime alias check is possible to below ...
25678 Call the new function.
25679 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
25680 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
25682 2017-06-07 Marek Polacek <polacek@redhat.com>
25685 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
25686 TYPE_OVERFLOW_WRAPS check.
25688 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25690 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
25691 if versioning is required.
25692 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
25693 peeling with the check for versioning.
25695 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25697 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
25698 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
25699 Set true to new parameter if new ssa variable is defined.
25700 (vect_gen_vector_loop_niters): Refactor. Set range information
25701 for the new vector loop bound variable.
25702 (vect_do_peeling): Ditto.
25704 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25706 * tree-affine.c (ssa.h): Include header file.
25707 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
25708 has wrapping overflow behavior.
25710 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25712 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
25714 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25716 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
25717 (tree_to_aff_combination): ... here.
25719 2017-06-07 Bin Cheng <bin.cheng@arm.com>
25721 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
25722 reg_pressure model function.
25723 (ivopts_global_cost_for_size): Delete.
25724 (determine_set_costs, iv_ca_recount_cost): Call new model function
25725 ivopts_estimate_reg_pressure.
25727 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25729 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
25730 expensive than udiv. Remove floating point cases from mod.
25732 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25734 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
25735 Increase idiv cost.
25737 2017-06-07 Tamar Christina <tamar.christina@arm.com>
25739 * config/aarch64/aarch64.md
25740 (copysignsf3): Fix mask generation.
25742 2017-06-07 Jakub Jelinek <jakub@redhat.com>
25744 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
25746 (class dump_manager): Add register_dumps method.
25747 * dumpfile.c: Include langhooks.h.
25748 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
25749 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
25750 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
25751 (dump_manager::dump_register): Start with 512 entries instead of 32.
25752 (dump_manager::register_dumps): New method.
25753 * toplev.c (general_init): Instead of invoking register_dumps
25754 langhook, invoke register_dumps method on the dump manager.
25755 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
25758 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
25760 * doc/md.texi: Clarify the restrictions on a define_insn condition.
25761 Say that # requires an associated define_split to exist, and that
25762 the define_split must be suitable for use after register allocation.
25764 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25766 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
25767 (compute_outgoing_frequencies): Also initialize zero counts.
25768 (find_many_sub_basic_blocks): Do not produce uninitialized profile
25769 around loops; preserve more of profile when nothing changes.
25771 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
25773 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
25775 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
25776 * config/arm/arm-cpu-cdata.h: Regenerate.
25777 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
25778 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
25779 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
25780 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
25781 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
25783 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
25785 * config/arm/t-rmprofile: Likewise.
25786 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
25788 2017-06-06 David S. Miller <davem@davemloft.net>
25791 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
25792 blockage if function uses alloca.
25794 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25796 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
25797 New "uid" fields to hold pretty-print IDs of group and ref.
25798 Memory references are now identified as <group_id>:<ref_id>
25799 instead of using [random] addresses.
25800 (dump_mem_details): Simplify, no functional change.
25801 (dump_mem_ref): Simplify and make output more concise.
25802 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
25803 (find_or_create_group): Initialize group uid.
25804 (record_ref): Initialize ref uid. Improve debug output.
25805 (prune_group_by_reuse, should_issue_prefetch_p,)
25806 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
25807 (mark_nontemporal_store, determine_loop_nest_reuse):
25808 Improve debug output.
25810 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25812 * dbgcnt.def (prefetch): New debug counter.
25813 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
25814 (schedule_prefetches): Stop issueing prefetches if debug counter
25817 2017-06-06 Tom de Vries <tom@codesourcery.com>
25819 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
25820 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
25822 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25824 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
25825 Use aarch64_reg_or_zero predicate for operand 4.
25826 (aarch64_compare_and_swap<mode> define_insn_and_split):
25827 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
25828 (aarch64_store_exclusive<mode>): Likewise for operand 2.
25830 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25832 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
25833 (arm_compute_save_core_reg_mask): This.
25834 (thumb1_compute_save_reg_mask): Rename into ...
25835 (thumb1_compute_save_core_reg_mask): This.
25836 (arm_compute_save_reg0_reg12_mask): Adapt comment.
25837 (arm_compute_frame_layout): Likewise.
25839 2017-06-06 Richard Biener <rguenther@suse.de>
25841 PR tree-optimization/80974
25842 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
25843 keep or clear leaders SSA info.
25845 2017-06-06 Tom de Vries <tom@codesourcery.com>
25847 * config/nvptx/nvptx.c (split_mode_p): New function.
25848 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
25850 2017-06-06 Tom de Vries <tom@codesourcery.com>
25852 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
25854 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25857 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
25860 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25862 * shrink-wrap.c (handle_simple_exit): Update profile.
25863 (try_shrink_wrapping): Upate profile.
25865 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25867 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
25868 (tree_guess_outgoing_edge_probabilities): New.
25869 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
25870 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
25872 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25874 * ipa-split.c (split_function): Initialize return bb profile.
25876 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
25878 * profile.c (compute_branch_probabilities): Also initialize
25879 EXIT_BLOCK profile.
25881 2017-06-06 Richard Biener <rguenther@suse.de>
25883 PR tree-optimization/80928
25884 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
25885 (vect_analyze_loop_operations): Properly guard analysis for
25887 (vect_transform_loop): Likewise.
25888 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25889 (vect_model_induction_cost): Do not cost for pure SLP.
25890 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25891 of induction in inner loop vectorization.
25892 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25893 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25894 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25896 (vect_analyze_slp_cost_1): Cost induction.
25897 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25898 (vect_get_slp_vect_defs): Likewise.
25899 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25900 (vect_transform_stmt): Handle SLP reductions.
25901 * tree-vectorizer.h (vectorizable_induction): Adjust.
25903 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25905 * config/rs6000/rs6000.c (make_resolver_func): Update
25906 init_lowered_empty_function call.
25908 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25910 * doc/invoke.texi: Document the -fprofile-abs-path option.
25911 * common.opt (fprofile-abs-path): New option.
25912 * gcov-io.h (gcov_write_filename): Declare.
25913 * gcov-io.c (gcov_write_filename): New function.
25914 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25915 * profile.c (output_location): Likewise.
25917 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25919 * shring-wrap.c: Revert accidental commit.
25921 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25923 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25925 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25927 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25929 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25930 profile in callgraph edge.
25931 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25932 (apply_scale): Likewise.
25933 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25934 Also scale profile when inlining function with zero profile.
25935 (initialize_cfun): Update exit block profile even when it is zero.
25936 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25937 when profile is read.
25939 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25941 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25942 (CLONE_*): New constants to define the processors we can generate
25943 code for with the target_clone attribute.
25944 (rs6000_clone_map): New array to identify which clone processors
25945 the current program is running on.
25946 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25947 target_clone attribute.
25948 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25949 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25950 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25951 (cpu_expand_builtin): Add support for target_clone attribute.
25952 (rs6000_valid_attribute_p): Allow "default" attribute.
25953 (get_decl_name): New debug function to simplify printing the
25954 current function name in debugging statements.
25955 (rs6000_clone_priority): New functions to support the target_clone
25956 attribute, and be able to generate code to switch between ISA 2.05
25957 through ISA 3.0 (power6 through power9).
25958 (rs6000_compare_version_priority): Likewise.
25959 (rs6000_get_function_versions_dispatcher): Likewise.
25960 (make_resolver_func): Likewise.
25961 (add_condition_to_bb): Likewise.
25962 (dispatch_function_versions): Likewise.
25963 (rs6000_generate_version_dispatcher_body): Likewise.
25964 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25965 (fusion_gpr_load_p): Fix a spacing issue.
25966 * doc/extend.texi (Common Function Attributes): Document that the
25967 PowerPC supports the target_clone attribute.
25969 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25971 * config/arm/arm.h: explain F symbol found in description of ARM
25972 register allocation in its legend.
25974 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25976 * config/mips/frame-header-opt.c: Include profile-count.h.
25977 * config/riscv/riscv.c: Include profile-count.h
25979 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25981 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25983 (sm_set_flag_if_changed): Add bbs field.
25984 (execute_sm_if_changed_flag_set): Pass BBS.
25985 (execute_sm): Update.
25987 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25989 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25992 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25994 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25995 (peephole2): New peephole2 to emit the above.
25996 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25998 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26000 * config/aarch64/aarch64.c (define_peephole2 above
26001 *sub_<shift>_<mode>): New peephole.
26003 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26005 * config/i386/i386.c (make_resolver_func): Update.
26006 * Makefile.in: Add profile-count.h and profile-count.o
26007 * auto-profile.c (afdo_indirect_call): Update to new API.
26008 (afdo_set_bb_count): Update.
26009 (afdo_propagate_edge): Update.
26010 (afdo_propagate_circuit): Update.
26011 (afdo_calculate_branch_prob): Update.
26012 (afdo_annotate_cfg): Update.
26013 * basic-block.h: Include profile-count.h
26014 (struct edge_def): Turn count to profile_count.
26015 (struct basic_block_def): Likewie.
26016 (REG_BR_PROB_BASE): Move to profile-count.h
26017 (RDIV): Move to profile-count.h
26018 * bb-reorder.c (max_entry_count): Turn to profile_count.
26019 (find_traces): Update.
26020 (rotate_loop):Update.
26021 (connect_traces):Update.
26022 (sanitize_hot_paths):Update.
26023 * bt-load.c (migrate_btr_defs): Update.
26024 * cfg.c (RDIV): Remove.
26025 (init_flow): Use alloc_block.
26026 (alloc_block): Uninitialize count.
26027 (unchecked_make_edge): Uninitialize count.
26028 (check_bb_profile): Update.
26029 (dump_edge_info): Update.
26030 (dump_bb_info): Update.
26031 (update_bb_profile_for_threading): Update.
26032 (scale_bbs_frequencies_int): Update.
26033 (scale_bbs_frequencies_gcov_type): Update.
26034 (scale_bbs_frequencies_profile_count): New.
26035 * cfg.h (update_bb_profile_for_threading): Update.
26036 (scale_bbs_frequencies_profile_count): Declare.
26037 * cfgbuild.c (compute_outgoing_frequencies): Update.
26038 (find_many_sub_basic_blocks): Update.
26039 * cfgcleanup.c (try_forward_edges): Update.
26040 (try_crossjump_to_edge): Update.
26041 * cfgexpand.c (expand_gimple_tailcall): Update.
26042 (construct_exit_block): Update.
26043 * cfghooks.c (verify_flow_info): Update.
26044 (dump_bb_for_graph): Update.
26045 (split_edge): Update.
26046 (make_forwarder_block): Update.
26047 (duplicate_block): Update.
26048 (account_profile_record): Update.
26049 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26050 (get_estimated_loop_iterations): Update.
26051 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26052 (single_likely_exit): Update.
26053 * cfgloopmanip.c (scale_loop_profile): Update.
26055 (set_zero_probability): Update.
26056 (lv_adjust_loop_entry_edge): Update.
26057 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26058 (purge_dead_edges): Update.
26059 (rtl_account_profile_record): Update.
26060 * cgraph.c (cgraph_node::create): Uninitialize count.
26061 (symbol_table::create_edge): Uninitialize count.
26062 (cgraph_update_edges_for_call_stmt_node): Update.
26063 (cgraph_edge::dump_edge_flags): Update.
26064 (cgraph_node::dump): Update.
26065 (cgraph_edge::maybe_hot_p): Update.
26066 * cgraph.h: Include profile-count.h
26067 (create_clone), create_edge, create_indirect_edge): Update.
26068 (cgraph_node): Turn count to profile_count.
26069 (cgraph_edge0: Likewise.
26070 (make_speculative, clone): Update.
26071 (create_edge): Update.
26072 (init_lowered_empty_function): Update.
26073 * cgraphclones.c (cgraph_edge::clone): Update.
26074 (duplicate_thunk_for_node): Update.
26075 (cgraph_node::create_clone): Update.
26076 * cgraphunit.c (cgraph_node::analyze): Update.
26077 (cgraph_node::expand_thunk): Update.
26078 * final.c (dump_basic_block_info): Update.
26079 * gimple-streamer-in.c (input_bb): Update.
26080 * gimple-streamer-out.c (output_bb): Update.
26081 * graphite.c (print_global_statistics): Update.
26082 (print_graphite_scop_statistics): Update.
26083 * hsa-brig.c: Include basic-block.h.
26084 * hsa-dump.c: Include basic-block.h.
26085 * hsa-gen.c (T sum_slice): Update.
26086 (convert_switch_statements):Update.
26087 * hsa-regalloc.c: Include basic-block.h.
26088 * ipa-chkp.c (chkp_produce_thunks): Update.
26089 * ipa-cp.c (struct caller_statistics): Update.
26090 (init_caller_stats): Update.
26091 (gather_caller_stats): Update.
26092 (ipcp_cloning_candidate_p): Update.
26093 (good_cloning_opportunity_p): Update.
26094 (get_info_about_necessary_edges): Update.
26095 (dump_profile_updates): Update.
26096 (update_profiling_info): Update.
26097 (update_specialized_profile): Update.
26098 (perhaps_add_new_callers): Update.
26099 (decide_about_value): Update.
26100 (ipa_cp_c_finalize): Update.
26101 * ipa-devirt.c (struct odr_type_warn_count): Update.
26102 (struct decl_warn_count): Update.
26103 (struct final_warning_record): Update.
26104 (possible_polymorphic_call_targets): Update.
26105 (ipa_devirt): Update.
26106 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26107 * ipa-icf.c (sem_function::merge): Update.
26108 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26109 * ipa-inline.c (compute_uninlined_call_time): Update.
26110 (compute_inlined_call_time): Update.
26111 (want_inline_small_function_p): Update.
26112 (want_inline_self_recursive_call_p): Update.
26113 (edge_badness): Update.
26114 (lookup_recursive_calls): Update.
26115 (recursive_inlining): Update.
26116 (inline_small_functions): Update.
26117 (dump_overall_stats): Update.
26118 (dump_inline_stats): Update.
26119 * ipa-profile.c (ipa_profile_generate_summary): Update.
26120 (ipa_propagate_frequency): Update.
26121 (ipa_profile): Update.
26122 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26123 * ipa-utils.c (ipa_merge_profiles): Update.
26124 * loop-doloop.c (doloop_modify): Update.
26125 * loop-unroll.c (report_unroll): Update.
26126 (unroll_loop_runtime_iterations): Update.
26127 * lto-cgraph.c (lto_output_edge): Update.
26128 (lto_output_node): Update.
26129 (input_node): Update.
26130 (input_edge): Update.
26131 (merge_profile_summaries): Update.
26132 * lto-streamer-in.c (input_cfg): Update.
26133 * lto-streamer-out.c (output_cfg): Update.
26134 * mcf.c (create_fixup_graph): Update.
26135 (adjust_cfg_counts): Update.
26136 (sum_edge_counts): Update.
26137 * modulo-sched.c (sms_schedule): Update.
26138 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26139 * predict.c (maybe_hot_count_p): Update.
26140 (probably_never_executed): Update.
26141 (dump_prediction): Update.
26142 (combine_predictions_for_bb): Update.
26143 (propagate_freq): Update.
26144 (handle_missing_profiles): Update.
26145 (counts_to_freqs): Update.
26146 (rebuild_frequencies): Update.
26147 (force_edge_cold): Update.
26148 * predict.h: Include profile-count.h
26149 (maybe_hot_count_p, counts_to_freqs): UPdate.
26150 * print-rtl-function.c: Do not include cfg.h
26151 * print-rtl.c: Include basic-block.h
26152 * profile-count.c: New file.
26153 * profile-count.h: New file.
26154 * profile.c (is_edge_inconsistent): Update.
26155 (correct_negative_edge_counts): Update.
26156 (is_inconsistent): Update.
26157 (set_bb_counts): Update.
26158 (read_profile_edge_counts): Update.
26159 (compute_frequency_overlap): Update.
26160 (compute_branch_probabilities): Update; Initialize and deinitialize
26162 (branch_prob): Update.
26163 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26164 (edge_gcov_count): New.
26165 (bb_gcov_count): New.
26166 * shrink-wrap.c (try_shrink_wrapping): Update.
26167 * tracer.c (better_p): Update.
26168 * trans-mem.c (expand_transaction): Update.
26169 (ipa_tm_insert_irr_call): Update.
26170 (ipa_tm_insert_gettmclone_call): Update.
26171 * tree-call-cdce.c: Update.
26172 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26173 (gimple_duplicate_sese_tail): Update.
26174 (gimple_account_profile_record): Update.
26175 (execute_fixup_cfg): Update.
26176 * tree-inline.c (copy_bb): Update.
26177 (copy_edges_for_bb): Update.
26178 (initialize_cfun): Update.
26179 (freqs_to_counts): Update.
26180 (copy_cfg_body): Update.
26181 (expand_call_inline): Update.
26182 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26183 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26184 (try_unroll_loop_completely): Update.
26185 (try_peel_loop): Update.
26186 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26187 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26188 * tree-ssa-loop-split.c (connect_loops): Update.
26189 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26190 * tree-ssa-reassoc.c (branch_fixup): Update.
26191 * tree-ssa-tail-merge.c (replace_block_by): Update.
26192 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26193 (compute_path_counts): Update.
26194 (update_profile): Update.
26195 (recompute_probabilities): Update.
26196 (update_joiner_offpath_counts): Update.
26197 (estimated_freqs_path): Update.
26198 (freqs_to_counts_path): Update.
26199 (clear_counts_path): Update.
26200 (ssa_fix_duplicate_block_edges): Update.
26201 (duplicate_thread_path): Update.
26202 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26203 (struct switch_conv_info): Update.
26204 * tree-tailcall.c (decrease_profile): Update.
26205 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26206 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26207 * value-prof.c (check_counter): Update.
26208 (gimple_divmod_fixed_value): Update.
26209 (gimple_mod_pow2): Update.
26210 (gimple_mod_subtract): Update.
26211 (gimple_ic_transform): Update.
26212 (gimple_stringop_fixed_value): Update.
26213 * value-prof.h (gimple_ic): Update.
26215 2017-06-02 Carl Love <cel@us.ibm.com>
26217 * config/rs6000/rs6000-c: Add support for built-in functions
26218 vector double vec_doublee (vector signed int);
26219 vector double vec_doublee (vector unsigned int);
26220 vector double vec_doublee (vector float);
26221 vector double vec_doubleh (vector signed int);
26222 vector double vec_doubleh (vector unsigned int);
26223 vector double vec_doubleh (vector float);
26224 vector double vec_doublel (vector signed int);
26225 vector double vec_doublel (vector unsigned int);
26226 vector double vec_doublel (vector float);
26227 vector double vec_doubleo (vector signed int);
26228 vector double vec_doubleo (vector unsigned int);
26229 vector double vec_doubleo (vector float);.
26230 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26231 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26233 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26234 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26235 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26237 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26238 vec_doublel, vec_doubleh.
26239 * doc/extend.texi: Update the built-in documentation file for the
26240 new built-in functions.
26242 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26245 * ipa-inline-analysis.c (free_growth_caches): Set
26246 edge_removal_hook_holder to NULL after removing it.
26248 2017-06-02 Sudakshina Das <sudi.das@arm.com>
26250 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
26251 comparision with zero.
26253 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26254 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26255 for early expansion of vec_min and vec_max builtins.
26256 (builtin_function_type): Add min/max unsigned variants to those
26257 identified as having unsigned arguments.
26259 2017-06-02 Olivier Hainque <hainque@adacore.com>
26261 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
26263 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26265 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
26266 Use VALL_F16 iterator rather than VALL.
26268 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26270 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
26271 Emit CBNZ inside loop when doing a strong exchange and comparing
26272 against zero. Generate the CC flags after the loop.
26274 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
26276 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
26277 (dl_section_ref): New.
26278 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
26279 On AIX, append an expression to subtract the size of the
26280 section length to dl_section_ref.
26282 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
26284 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26285 for early expansion of vector absolute builtins.
26287 2017-06-02 Richard Biener <rguenther@suse.de>
26289 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
26290 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
26292 2017-06-02 Richard Biener <rguenther@suse.de>
26294 PR tree-optimization/80948
26295 * tree-tailcall.c (find_tail_calls): Track stmts to move in
26296 stmt order as well.
26298 2017-06-02 Richard Biener <rguenther@suse.de>
26300 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
26302 * tree-vect-stmts.c (process_use): Do not mark backedge defs
26303 for inductions as relevant.
26305 2017-06-02 Richard Biener <rguenther@suse.de>
26307 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
26308 (vectorizable_induction): ... this. Remove dead code.
26310 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
26312 * builtins. (expand_builtin_alloca): Remove second parameter and
26313 infer its value from the first parameter instead.
26314 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
26316 2017-06-02 Jakub Jelinek <jakub@redhat.com>
26318 PR rtl-optimization/80903
26319 * loop-doloop.c (add_test): Unshare sequence.
26321 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
26323 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
26325 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
26327 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
26329 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
26330 xlogue_layout::get_instance, logue_layout::xlogue_layout,
26331 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
26332 (xlogue_layout::get_stub_rtx): Make static.
26333 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
26334 (xlogue_layout::compute_stub_managed_regs): Rename to...
26335 (xlogue_layout::count_stub_managed_regs): ...this.
26336 (xlogue_layout::is_stub_managed_reg): New function.
26337 (xlogue_layout::m_stub_names): Rename to...
26338 (xlogue_layout::s_stub_names): ...this, make static.
26339 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
26340 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
26341 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
26342 xlogue_layout::s_stub_names): Instantiate statics.
26343 (stub_managed_regs): Remove.
26344 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
26345 (disable_call_ms2sysv_xlogues): Rename to...
26346 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
26347 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
26349 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
26350 change after reload_completed.
26351 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
26353 (ix86_expand_prologue): Likewise.
26354 (ix86_expand_epilogue): Likewise.
26355 (ix86_expand_split_stack_prologue): Likewise.
26356 (ix86_compute_frame_layout): Remove frame parameter ...
26357 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
26358 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
26360 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
26361 (ix86_frame): Move from here ...
26362 * config/i386/i386.h (ix86_frame): ... to here.
26363 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
26364 complete ix86_frame data structure instead. Remove some_ld_name.
26366 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26368 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
26369 symbols that hold a DECL_VALUE_EXPR.
26371 2017-06-01 Martin Jambor <mjambor@suse.cz>
26373 PR tree-optimization/80898
26374 * tree-sra.c (process_subtree_disqualification): Removed.
26375 (disqualify_candidate): Do not acll
26376 process_subtree_disqualification.
26377 (subtree_mark_written_and_enqueue): New function.
26378 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
26379 RHS has been disqualified and re-queue LHS if necessary. Apart
26380 from that, ignore disqualified RHS.
26382 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26384 * config/s390/s390.c (s390_emit_epilogue): Disable early return
26385 address fetch for z10 or later.
26387 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26389 * config/arc/arc.md (tst_movb): Add guard when splitting.
26391 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26393 * config/arc/arc.c (arc_can_eliminate): Test against
26394 arc_frame_pointer_needed.
26396 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26398 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
26399 to prevent store reordering.
26400 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
26401 (type): Add block type.
26402 (stack_tie): Define special instruction to be used in
26405 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26407 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
26408 constraint. It is not valid for the pattern.
26409 (noncommutative_binary_comparison): Likewise.
26411 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26413 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
26416 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26418 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
26419 be used by the reg-alloc.
26421 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26423 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
26424 reg-alloc when having mul64 or mul32x16 instructions.
26425 (mulsidi3): Likewise.
26426 (umulsidi3): Likewise.
26427 (mulsi32x16): New pattern.
26428 (mulsi64): Likewise.
26429 (mulsidi64): Likewise.
26430 (umulsidi64): Likewise.
26431 (MUL32x16_REG): Define.
26432 (mul64_600): Use MUL32x16_REG.
26433 (mac64_600): Likewise.
26434 (umul64_600): Likewise.
26435 (umac64_600): Likewise.
26437 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
26439 * config/arc/arc.md (mulsi3_700): Make it commutative.
26441 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
26443 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
26445 (*sign_extendsidi2_insn): Likewise for movstosw.
26447 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
26449 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
26450 the type of the input discriminant value. Convert the
26451 discriminant value of signedness vary.
26453 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
26455 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
26456 Add to -Wall section.
26458 2017-06-01 Richard Biener <rguenther@suse.de>
26460 PR middle-end/66313
26461 * fold-const.c (fold_plusminus_mult_expr): If the factored
26462 factor may be zero use a wrapping type for the inner operation.
26463 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
26464 and handle moved defs.
26465 (process_assignment): Properly guard the unary op case. Return a
26466 tri-state indicating that moving the stmt before the call may allow
26467 to continue. Pass through to_move.
26468 (find_tail_calls): Handle moving unrelated defs before
26471 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
26474 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
26475 splitter result in the canonical way.
26477 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
26479 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
26480 also for 32bit target. Update insn attributes.
26481 (zero-extendsidi2 splitter): Allow all registers for operand 1.
26483 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
26485 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
26486 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
26487 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
26488 (_mm_maskz_min_ss): New intrinsics.
26490 2017-05-31 Martin Liska <mliska@suse.cz>
26492 * tree-vect-loop.c (vect_create_epilog_for_reduction):
26493 Change comment style to one we normally use.
26494 (vectorizable_reduction): Likewise.
26495 (vectorizable_induction): Likewise.
26496 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
26497 (vectorizable_call): Likewise.
26498 (vectorizable_simd_clone_call): Likewise.
26499 (vectorizable_conversion): Likewise.
26500 (vectorizable_assignment): Likewise.
26501 (vectorizable_shift): Likewise.
26502 (vectorizable_operation): Likewise.
26503 (vectorizable_store): Likewise.
26504 (vectorizable_load): Likewise.
26505 * tree-vectorizer.h: Likewise.
26507 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
26509 * passes.c (emergency_dump_function): New.
26510 * tree-pass.h (emergency_dump_function): Declare.
26511 * plugin.c (plugins_internal_error_function): Remove.
26512 * plugin.h (plugins_internal_error_function): Remove declaration.
26513 * toplev.c (internal_error_function): New static function. Use it...
26514 (general_init): ...here.
26516 2017-05-31 Graham Markall <graham.markall@embecosm.com>
26518 * config/arc/arc.c (arc_print_operand): Handle constant operands.
26519 (arc_rtx_costs): Add costs for new patterns.
26520 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
26521 * config/arc/predicates.md: Add _1_2_3_operand predicate.
26523 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26525 * tree-ssa-strlen.c (get_next_strinfo): New function.
26526 (get_stridx_plus_constant): Use it.
26527 (zero_length_string): Likewise.
26528 (adjust_related_strinfos): Likewise.
26529 (adjust_last_stmt): Likewise.
26531 2017-05-31 Richard Biener <rguenther@suse.de>
26534 * config/i386/i386.c (ix86_expand_builtin): Remove assert
26535 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
26537 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
26539 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
26540 loop_vinfo argument and use of dependence distance vectors.
26541 Check instead whether the two references differ only in their
26542 initial value and assume that they have the same alignment if the
26543 difference is a multiple of the vector alignment.
26544 (vect_analyze_data_refs_alignment): Update call accordingly.
26546 2017-05-31 Martin Liska <mliska@suse.cz>
26549 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
26551 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26553 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
26554 (create_intersect_range_checks): Move from ...
26555 * tree-data-ref.c (create_intersect_range_checks_index)
26556 (create_intersect_range_checks): ... to here.
26557 (create_runtime_alias_checks): New function factored from ...
26558 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
26559 here. Call above function.
26560 * tree-data-ref.h (create_runtime_alias_checks): New function.
26562 2017-05-31 Bin Cheng <bin.cheng@arm.com>
26564 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
26565 segment length for dr_b and compute it in wide_int.
26567 2017-05-31 Richard Biener <rguenther@suse.de>
26569 PR tree-optimization/80906
26570 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
26571 and pass through iv_map.
26572 (copy_bb_and_scalar_dependences): Adjust.
26573 (translate_pending_phi_nodes): Likewise.
26574 (copy_loop_close_phi_args): Handle code-generating IVs instead
26577 2017-05-30 David Malcolm <dmalcolm@redhat.com>
26579 * diagnostic-color.c (color_dict): Add "type-diff".
26580 (parse_gcc_colors): Update comment.
26581 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
26582 -fdiagnostics-show-template-tree and -fno-elide-type.
26583 (GCC_COLORS): Add type-diff to example.
26585 (-fdiagnostics-show-template-tree): New.
26586 (-fno-elide-type): New.
26587 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
26588 the pp_format_decoder callback. Call any m_format_postprocessor's
26590 (pretty_printer::pretty_printer): Initialize
26591 m_format_postprocessor.
26592 (pretty_printer::~pretty_printer): Delete any
26593 m_format_postprocessor.
26594 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
26595 (class format_postprocessor): New class.
26596 (struct pretty_printer::format_decoder): Document the new parameters.
26597 (struct pretty_printer::m_format_postprocessor): New field.
26598 * tree-diagnostic.c (default_tree_printer): Update for new
26599 bool and const char ** params.
26600 * tree-diagnostic.h (default_tree_printer): Likewise.
26602 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
26604 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
26605 (lwa_operand): Delete rs6000_gen_cell_microcode test.
26606 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26607 rs6000_gen_cell_microcode code.
26608 (rs6000_final_prescan_insn): Delete.
26609 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
26610 "warn-cell-microcode" entries.
26611 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
26612 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
26613 throughout. Change cc_reg_not_micro_cr0_operand to
26614 cc_reg_not_cr0_operand throughout.
26615 (*extendhi<mode>2_noload): Delete.
26616 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
26617 (mwarn-cell-microcode): Delete.
26618 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
26619 -mgen-cell-microcode and -mwarn-cell-microcode.
26621 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
26624 * config/i386/constraints.md (Yd): New constraint.
26626 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
26627 and (?Yd, r) alternatives. Update insn attributes.
26628 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
26629 and (?*Yd, r) alternatives. Update insn attributes.
26630 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
26632 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
26634 * gimplify.c (gimplify_modify_expr): Don't create a
26635 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
26638 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
26640 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
26642 2017-05-30 Richard Biener <rguenther@suse.de>
26644 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
26645 and reduc_def fields.
26646 (STMT_VINFO_REDUC_TYPE): New define.
26647 (STMT_VINFO_REDUC_DEF): Likewise.
26648 (vect_force_simple_reduction): Adjust prototype.
26649 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
26650 (vect_is_simple_reduction): Remove check_reduction argument.
26651 (vect_force_simple_reduction): Adjust and set
26652 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26653 (vectorizable_reduction): Do not re-do reduction analysis
26654 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
26655 * tree-parloops.c (gather_scalar_reductions): Adjust.
26657 2017-05-30 Richard Biener <rguenther@suse.de>
26659 PR middle-end/80901
26660 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
26663 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
26665 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
26666 Introduce unknown_misalignment parameter and remove vf.
26667 (vect_peeling_hash_get_lowest_cost):
26668 Pass unknown_misalignment parameter.
26669 (vect_enhance_data_refs_alignment):
26670 Fix unsupportable data ref treatment.
26672 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26674 * tree-vect-data-refs.c (vect_get_data_access_cost):
26675 Workaround for SLP handling.
26676 (vect_enhance_data_refs_alignment):
26677 Compute costs for doing no peeling at all, compare to the best
26678 peeling costs so far and avoid peeling if cheaper.
26680 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26682 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
26683 Return peeling info and set costs to zero for unlimited cost
26685 (vect_enhance_data_refs_alignment): Also inspect all datarefs
26686 with unknown misalignment. Compute and costs for unknown
26687 misalignment, compare them to the costs for known misalignment
26688 and choose the cheapest for peeling.
26690 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26692 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
26693 (vect_get_peeling_costs_all_drs): Create function.
26694 (vect_peeling_hash_get_lowest_cost):
26695 Use vect_get_peeling_costs_all_drs.
26696 (vect_peeling_supportable): Create function.
26697 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
26699 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
26701 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
26702 DR_HAS_NEGATIVE_STEP.
26703 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
26704 (vect_enhance_data_refs_alignment): Use.
26705 (vect_duplicate_ssa_name_ptr_info): Use.
26706 * tree-vectorizer.h (dr_misalignment): Use.
26707 (known_alignment_for_access_p): Use.
26709 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26712 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
26714 (has_section_name): New function.
26716 2017-05-30 Martin Liska <mliska@suse.cz>
26719 * auto-profile.c (get_function_decl_from_block): Fix
26722 2017-05-30 Richard Biener <rguenther@suse.de>
26724 PR middle-end/80876
26725 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
26727 2017-05-30 Martin Liska <mliska@suse.cz>
26729 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
26730 * dumpfile.h (struct dump_file_info): Remove ctors.
26732 2017-05-30 Martin Liska <mliska@suse.cz>
26734 * predict.def: Fix GNU coding style.
26736 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26738 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
26739 Mark 'to' argument with ATTRIBUTE_UNUSED.
26741 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
26743 * config/xtensa/xtensa.c (xtensa_emit_call): Use
26744 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
26745 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
26748 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
26750 * doc/install.texi (Options specification): Restore entry of
26751 --enable-sjlj-exceptions.
26753 2017-05-27 Michael Eager <eager@eagercon.com>
26756 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
26758 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
26760 * config/microblaze/microblaze.h
26761 (FIXED_REGISTERS): Update in macro.
26762 (CALL_USED_REGISTERS): Update in macro.
26764 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
26766 * doc/install.texi: Add links to macOS binary distributions.
26768 2017-05-27 Jakub Jelinek <jakub@redhat.com>
26772 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26774 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26776 2017-05-26 Martin Liska <mliska@suse.cz>
26778 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
26780 2017-05-26 Martin Liska <mliska@suse.cz>
26782 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
26783 always leading ';; '.
26784 (dump_bb_info): Likewise.
26785 (brief_dump_cfg): Likewise.
26786 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
26787 * dumpfile.c: Remove usage of TDF_VERBOSE.
26788 * dumpfile.h (enum dump_kind): Likewise.
26789 (dump_gimple_bb_header): Do not use TDF_COMMENT.
26790 * print-tree.c (debug_verbose): Remove.
26791 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
26792 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
26793 * tree-diagnostic.c (default_tree_printer): Replace
26794 TDF_DIAGNOSTIC with TDF_SLIM.
26796 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26798 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
26799 in parameter loop, rather than loop_vinfo.
26800 (create_intersect_range_checks): Ditto.
26801 (vect_create_cond_for_alias_checks): Update call to above functions.
26803 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26805 PR tree-optimization/80815
26806 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
26807 for merging runtime alias checks. Handle negative DR_STEPs.
26809 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26811 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
26813 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
26814 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
26815 out code pruning runtime alias checks.
26816 * tree-data-ref.c (prune_runtime_alias_test_list): New function
26817 factored out from above.
26818 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26820 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
26822 (prune_runtime_alias_test_list): New decalaration.
26824 2017-05-26 Bin Cheng <bin.cheng@arm.com>
26826 * tree-vect-data-refs.c (compare_tree): Rename and move ...
26827 * tree-data-ref.c (data_ref_compare_tree): ... to here.
26828 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
26829 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
26830 (operator==, comp_dr_with_seg_len_pair): Ditto.
26831 (vect_prune_runtime_alias_test_list): Ditto.
26833 2017-05-26 Martin Liska <mliska@suse.cz>
26836 * params.def: Bound partial-inlining-entry-probability param.
26838 2017-05-26 Marek Polacek <polacek@redhat.com>
26841 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
26844 2017-05-26 Richard Biener <rguenther@suse.de>
26846 PR tree-optimization/80842
26847 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
26850 2017-05-26 Richard Biener <rguenther@suse.de>
26852 PR tree-optimization/80844
26853 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
26855 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
26857 * doc/md.texi (Machine Constraints): Update x86 family
26858 machine constraints section to match 'config/i386/constraints.md'.
26860 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
26862 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
26864 2017-05-25 Nathan Sidwell <nathan@acm.org>
26866 * doc/invoke.texi (--enable-languages): Update documentation.
26868 2017-05-25 Martin Liska <mliska@suse.cz>
26870 * dumpfile.c: Add TDF_FOLDING.
26871 * dumpfile.h (enum dump_kind): Likewise.
26872 * genmatch.c (dt_simplify::gen_1): Use it.
26874 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26876 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
26878 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26880 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
26881 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
26883 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
26885 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
26886 * match.pd (X == C): Rewrite it here.
26887 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26888 with_certain_nonzero_bits2): New predicates.
26889 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26891 2017-05-24 Nathan Sidwell <nathan@acm.org>
26893 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26896 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26899 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26901 * config/powerpcspe: New port. Files are copied from the rs6000
26902 port, with "rs6000" in filenames replaced by "powerpcspe".
26904 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26906 PR rtl-optimization/80754
26907 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26909 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26911 * config/sparc/sparc.md (length): Return the correct value for -mflat
26912 sibcalls to match output_sibcall.
26914 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26918 * config/rs6000/rs6000.c (struct machine_function): Add new field
26920 (rs6000_get_separate_components): Init that field, use it.
26921 (rs6000_components_for_bb): Use the field.
26923 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26925 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26927 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26929 PR middle-end/80823
26930 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26932 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26935 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26936 address against address_operand predicate.
26937 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26939 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26941 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26942 parameters passed indirectly.
26944 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26946 * config/i386/i386.md (*movdi_internal): Remove SSE4
26947 alternative 18 (?r, *v). Update insn attributes.
26948 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26949 Update insn attributes.
26950 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26951 Update insn attributes.
26952 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26953 alternative 1 (r, v). Remove isa attribute.
26954 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26955 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26956 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26958 2017-05-23 Tom de Vries <tom@codesourcery.com>
26960 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26963 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26965 * cgraphunit.c (symbol_table::process_new_functions): Update.
26966 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26967 (inline_generate_summary): Rename to ...
26968 (ipa_fn_summary_generate): ... this one.
26969 (inline_read_summary): Rename to ...
26970 (ipa_fn_summary_read): ... this one.
26971 (inline_write_summary): Rename to ...
26972 (ipa_fn_summary_write): ... this one.
26973 (inline_free_summary): Rename to ...
26974 (ipa_free_fn_summary): ... this one.
26975 (pass_data_local_fn_summary, pass_local_fn_summary,
26976 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26977 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26978 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26979 make_pass_ipa_fn_summary): New.
26980 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26981 inline_write_summary, inline_free_summary): Remove.
26982 (ipa_free_fn_summary) : New.
26983 * ipa-inline.c (ipa_inline): Update.
26984 (pass_ipa_inline): Do not generate summaries.
26985 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26987 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26988 and add pass_ipa_fn_summary.
26989 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26991 (make_pass_inline_parameters): Remove.
26993 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26995 * omp-low.c (struct omp_context): Remove "default_kind" member.
26998 * omp-offload.c (execute_oacc_device_lower): Remove the
26999 parallelism dimensions function attributes for unparallelized
27000 OpenACC kernels constructs.
27002 2017-05-23 Martin Liska <mliska@suse.cz>
27004 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27006 (cgraph_edge::make_speculative): Likewise.
27007 (cgraph_edge::resolve_speculation): Likewise.
27008 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27009 (cgraph_node::dump): Likewise.
27010 * cgraph.h: Likewise.
27011 * cgraphunit.c (analyze_functions): Likewise.
27012 (symbol_table::compile): Likewise.
27013 * ipa-cp.c (print_all_lattices): Likewise.
27014 (determine_versionability): Likewise.
27015 (initialize_node_lattices): Likewise.
27016 (ipcp_verify_propagated_values): Likewise.
27017 (estimate_local_effects): Likewise.
27018 (update_profiling_info): Likewise.
27019 (create_specialized_node): Likewise.
27020 (perhaps_add_new_callers): Likewise.
27021 (decide_about_value): Likewise.
27022 (decide_whether_version_node): Likewise.
27023 (identify_dead_nodes): Likewise.
27024 (ipcp_store_bits_results): Likewise.
27025 * ipa-devirt.c (dump_targets): Likewise.
27026 (ipa_devirt): Likewise.
27027 * ipa-icf.c (sem_item::dump): Likewise.
27028 (sem_function::equals): Likewise.
27029 (sem_variable::equals): Likewise.
27030 (sem_item_optimizer::read_section): Likewise.
27031 (sem_item_optimizer::execute): Likewise.
27032 (congruence_class::dump): Likewise.
27033 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27034 (dump_inline_summary): Likewise.
27035 (estimate_node_size_and_time): Likewise.
27036 (inline_analyze_function): Likewise.
27037 * ipa-inline-transform.c (inline_call): Likewise.
27038 * ipa-inline.c (report_inline_failed_reason): Likewise.
27039 (want_early_inline_function_p): Likewise.
27040 (edge_badness): Likewise.
27041 (update_edge_key): Likewise.
27042 (inline_small_functions): Likewise.
27043 * ipa-profile.c (ipa_profile): Likewise.
27044 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27045 (ipa_make_edge_direct_to_target): Likewise.
27046 (remove_described_reference): Likewise.
27047 (ipa_impossible_devirt_target): Likewise.
27048 (propagate_controlled_uses): Likewise.
27049 (ipa_print_node_params): Likewise.
27050 (ipcp_transform_function): Likewise.
27051 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27052 (propagate_pure_const): Likewise.
27053 * ipa-reference.c (generate_summary): Likewise.
27054 (read_write_all_from_decl): Likewise.
27055 (propagate): Likewise.
27056 (ipa_reference_read_optimization_summary): Likewise.
27057 * ipa-utils.c (ipa_merge_profiles): Likewise.
27058 * ipa.c (walk_polymorphic_call_targets): Likewise.
27059 (symbol_table::remove_unreachable_nodes): Likewise.
27060 (ipa_single_use): Likewise.
27061 * passes.c (execute_todo): Likewise.
27062 * predict.c (drop_profile): Likewise.
27063 * symtab.c (symtab_node::get_dump_name): New function.
27064 (symtab_node::dump_name): Likewise.
27065 (symtab_node::dump_asm_name): Likewise.
27066 (symtab_node::dump_references): Likewise.
27067 (symtab_node::dump_referring): Likewise.
27068 (symtab_node::dump_base): Likewise.
27069 (symtab_node::debug_symtab): Likewise.
27070 * tree-sra.c (convert_callers_for_node): Likewise.
27071 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27072 * value-prof.c (init_node_map): Likewise.
27074 2017-05-23 Martin Liska <mliska@suse.cz>
27076 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27077 and symtab_node::debug_symtab to symbol_table::debug.
27078 * cgraphunit.c (analyze_functions): Use the renamed function.
27079 (symbol_table::compile): Likewise.
27080 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27081 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27082 * passes.c (execute_todo): Likewise.
27083 * symtab.c (symbol_table::dump): New function.
27084 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27086 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27088 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27089 that nonconst implies exec.
27091 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27093 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27094 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27095 (inline_edge_summary_vec): Turn into ...
27096 (ipa_call_summaries): ... this one.
27097 (redirect_to_unreachable, edge_set_predicate,
27098 evaluate_properties_for_edge, inline_summary_alloc,
27099 reset_ipa_call_summary, reset_inline_summary,
27100 inline_summary_t::duplicate): Update.
27101 (inline_edge_duplication_hook): Turn to ...
27102 (ipa_call_summary_t::duplicate): ... this one.
27103 (inline_edge_removal_hook): Turn to ...
27104 (ipa_call_summary_t::remove): ... this one.
27105 (dump_inline_edge_summary): Turn to ...
27106 (dump_ipa_call_summary): ... this one.
27107 (estimate_function_body_sizes): Update.
27108 (inline_update_callee_summaries): Update.
27109 (remap_edge_change_prob): Update.
27110 (remap_edge_summaries): Update.
27111 (inline_merge_summary): Update.
27112 (do_estimate_edge_time): Update.
27113 (inline_generate_summary): Update.
27114 (inline_read_section): Update.
27115 (inline_read_summary): Update.
27116 (inline_free_summary): Update.
27117 * ipa-inline.c (can_inline_edge_p): Update.
27118 (compute_inlined_call_time): Update.
27119 (want_inline_small_function_p): Update.
27120 (edge_badness): Update.
27121 (early_inliner): Update.
27122 * ipa-inline.h (inline_edge_summary): Turn to ...
27123 (ipa_call_summary): ... this one.
27124 (ipa_call_summary_t): New class.
27125 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27126 (ipa_call_summaries): New.
27127 (inline_edge_summary): Remove.
27128 (estimate_edge_growth): Update.
27129 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27130 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27131 * ipa-split.c (execute_split_functions): Update.
27132 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27134 2017-05-23 Tom de Vries <tom@codesourcery.com>
27136 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27137 attributes): Document rdrand effective target.
27139 2017-05-23 Tom de Vries <tom@codesourcery.com>
27141 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27142 attributes): Sort alphabetically.
27144 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27146 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27148 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27151 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27152 V2DF/V2DI splat into two separate patterns, one that handles
27153 registers, and the other that only handles memory. Drop support
27154 for splatting from a GPR on ISA 2.07 and then splitting the
27155 splat into direct move and splat.
27156 (vsx_splat_<mode>_reg): Likewise.
27157 (vsx_splat_<mode>_mem): Likewise.
27159 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27161 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27163 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27165 PR middle-end/80809
27166 * omp-low.c (finish_taskreg_remap): New function.
27167 (finish_taskreg_scan): If unit size of ctx->record_type
27168 is non-constant, unshare the size expression and replace
27169 decls in it with possible outer var refs.
27171 PR middle-end/80809
27172 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27173 GOVD_SHARED rather than GOVD_PRIVATE with it.
27174 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27175 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27177 PR middle-end/80853
27178 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27179 as last argument to build_outer_var_ref for pointer bases of array
27180 section reductions.
27182 2017-05-19 Martin Sebor <msebor@redhat.com>
27184 * print-tree.c (print_node): Print DECL_READ_P flag.
27186 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27188 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27189 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27190 * cgraph.c: Likewise.
27191 * cgraphunit.c: Likewise.
27192 * gengtype.c: Likewise.
27193 * ipa-cp.c: Likewise.
27194 * ipa-devirt.c: Likewise.
27195 * ipa-icf.c: Likewise.
27196 * ipa-predicate.c: Likewise.
27197 * ipa-profile.c: Likewise.
27198 * ipa-prop.c: Likewise.
27199 * ipa-split.c: Likewise.
27201 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27202 edge_predicate_pool, dump_inline_hints,
27203 inline_summary::account_size_time, redirect_to_unreachable,
27204 edge_set_predicate, set_hint_predicate,
27205 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27206 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27207 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27208 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27209 ipa_call_summary_t::remove, initialize_growth_caches,
27210 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27211 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27212 mark_modified, unmodified_parm_1, unmodified_parm,
27213 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27214 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27215 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27216 will_be_nonconstant_predicate, record_modified_bb_info,
27217 get_minimal_bb, record_modified, param_change_prob,
27218 phi_result_unknown_predicate, predicate_for_phi_result,
27219 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27220 estimate_function_body_sizes, compute_inline_parameters,
27221 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27222 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27223 inline_update_callee_summaries, remap_edge_change_prob,
27224 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27225 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27226 inline_analyze_function, inline_summary_t::insert,
27227 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27228 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27229 inline_free_summary): Move to ipa-fnsummary.h
27230 (predicate_t): Remove.
27231 * ipa-fnsummary.c: New file.
27232 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27233 (enum inline_hints_vals, inline_hints, agg_position_info,
27234 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27235 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27236 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27237 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27238 inline_read_summary, inline_write_summary, inline_free_summary,
27239 inline_analyze_function, initialize_inline_failed,
27240 inline_merge_summary, inline_update_overall_summary,
27241 compute_inline_parameters): Move to ipa-fnsummary.h
27242 * ipa-fnsummary.h: New file.
27243 * ipa-inline-transform.h: Include ipa-inline.h.
27244 * ipa-inline.c: LIkewise.
27246 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27248 * ipa-inline.c (edge_badness): Use inlined_time instead of
27249 inline_summaries->get.
27251 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27253 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
27255 2017-05-22 Nathan Sidwell <nathan@acm.org>
27257 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
27258 (fdump-lang): Document 'raw' option.
27259 * dumpfile.h (TDI_tu): Delete.
27260 * dumpfile.c (dump_files): Remove translation-unit.
27261 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
27263 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
27265 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
27266 command option from $(AWK) call.
27267 * config/avr/genmultilib.awk: Simplify and rewrite so that it
27268 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
27269 [FORMAT]: Remove handling of variable.
27270 * config/avr/t-multilib: Regenerate.
27272 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27274 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
27276 (dump_inline_summary): Do not print self_time.
27277 (estimate_function_body_sizes): Do not set self_time.
27278 (compute_inline_parameters): Likewise.
27279 (inline_read_section, inline_write_summary): Do not stream self_time.
27280 * ipa-inline.h (inline_summary): Drop self_time.
27282 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27284 * ipa-inline-analysis.c (account_size_time): Rename to ...
27285 (inline_summary::account_size_time): ... this one.
27286 (reset_ipa_call_summary): Turn to ...
27287 (ipa_call_summary::reset): ... this one.
27288 (reset_inline_summary): Turn to ...
27289 (inline_summary::reset): ... this one.
27290 (inline_summary_t::remove): Update.
27291 (inline_summary_t::duplicate): Update.
27292 (ipa_call_summary_t::remove): Update.
27293 (dump_inline_summary): Update.
27294 (estimate_function_body_sizes): Update.
27295 (compute_inline_parameters): Update.
27296 (estimate_node_size_and_time): Update.
27297 (inline_merge_summary): Update.
27298 (inline_update_overall_summary): Update.
27299 (inline_read_section): Update.
27300 (inline_write_summary): Update.
27301 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
27302 add account_size_time and reset member functions.
27303 (ipa_call_summary): Add reset function.
27304 * ipa-predicate.h (predicate::operator &): Constify.
27306 2017-05-22 Richard Biener <rguenther@suse.de>
27308 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
27310 2017-05-19 Jason Merrill <jason@redhat.com>
27312 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
27314 2017-05-19 Marek Polacek <polacek@redhat.com>
27317 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
27318 TYPE_OVERFLOW_WRAPS checks.
27320 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
27322 * tree-core.h (enum omp_clause_default_kind): Add
27323 "OMP_CLAUSE_DEFAULT_PRESENT".
27324 * tree-pretty-print.c (dump_omp_clause): Handle it.
27325 * gimplify.c (enum gimplify_omp_var_data): Add
27326 "GOVD_MAP_FORCE_PRESENT".
27327 (gimplify_adjust_omp_clauses_1): Map it to
27328 "GOMP_MAP_FORCE_PRESENT".
27329 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
27331 * gimplify.c (oacc_default_clause): Clarify.
27333 2017-05-19 Nathan Sidwell <nathan@acm.org>
27335 LANG_HOOK_REGISTER_DUMPS
27336 * toplev.c (general_init): Call register dump lang hook.
27337 * doc/invoke.texi: Document -fdump-lang option family.
27338 * dumpfile.c (dump_files): Remove class dump here.
27339 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
27340 * dumpfile.h (tree_dump_index): Remove TDI_class.
27341 * langhooks-def.h (lhd_register_dumps): Declare.
27342 (LANG_HOOKS_REGISTER_DUMPS): Define.
27343 (LANG_HOOKS_INITIALIZER): Add it.
27344 * langhooks.c (lhd_register_dumps): Define.
27345 * langhooks.h (struct lang_hooks): Add register_dumps.
27347 2017-05-19 Nathan Sidwell <nathan@acm.org>
27349 * context.h (context::set_passes): New.
27350 * context.c (context::context): Do not create pass manager.
27351 * toplev.c (general_init): Create pass manager here.
27353 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
27355 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
27356 use this splitter if two add or or instructions would also work for
27357 the constant we want to generate.
27359 2017-05-19 Richard Biener <rguenther@suse.de>
27362 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
27363 predicate evaluation.
27365 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27367 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
27369 * ipa-inline.c (want_inline_small_function_p): Do not cast to
27372 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
27374 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27375 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27376 (inline_edge_summary_vec): Turn into ...
27377 (ipa_call_summaries): ... this one.
27378 (redirect_to_unreachable, edge_set_predicate,
27379 evaluate_properties_for_edge, inline_summary_alloc,
27380 reset_ipa_call_summary, reset_inline_summary,
27381 inline_summary_t::duplicate): Update.
27382 (inline_edge_duplication_hook): Turn to ...
27383 (ipa_call_summary_t::duplicate): ... this one.
27384 (inline_edge_removal_hook): Turn to ...
27385 (ipa_call_summary_t::remove): ... this one.
27386 (dump_inline_edge_summary): Turn to ...
27387 (dump_ipa_call_summary): ... this one.
27388 (estimate_function_body_sizes): Update.
27389 (inline_update_callee_summaries): Update.
27390 (remap_edge_change_prob): Update.
27391 (remap_edge_summaries): Update.
27392 (inline_merge_summary): Update.
27393 (do_estimate_edge_time): Update.
27394 (inline_generate_summary): Update.
27395 (inline_read_section): Update.
27396 (inline_read_summary): Update.
27397 (inline_free_summary): Update.
27398 * ipa-inline.c (can_inline_edge_p): Update.
27399 (compute_inlined_call_time): Update.
27400 (want_inline_small_function_p): Update.
27401 (edge_badness): Update.
27402 (early_inliner): Update.
27403 * ipa-inline.h (inline_edge_summary): Turn to ...
27404 (ipa_call_summary): ... this one.
27405 (ipa_call_summary_t): New class.
27406 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27407 (ipa_call_summaries): New.
27408 (inline_edge_summary): Remove.
27409 (estimate_edge_growth): Update.
27410 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27411 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27412 * ipa-split.c (execute_split_functions): Update.
27413 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27415 2017-05-19 Richard Biener <rguenther@suse.de>
27417 PR middle-end/80764
27418 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
27420 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
27422 * config/rs6000/rs6000.c (struct machine_function): Add field
27423 fpr_is_wrapped_separately.
27424 (rs6000_get_separate_components): Use 64 components. Handle the
27425 new FPR components.
27426 (rs6000_components_for_bb): Handle the FPR components.
27427 (rs6000_emit_prologue_components): Handle the FPR components.
27428 (rs6000_emit_epilogue_components): Handle the FPR components.
27429 (rs6000_set_handled_components): Handle the FPR components.
27430 (rs6000_emit_prologue): Don't output prologue code for those FPRs
27431 that are already separately shrink-wrapped.
27432 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
27433 that are already separately shrink-wrapped.
27435 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27438 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
27441 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
27442 (define_peephole2 for Altivec d-form load): Add peepholes to catch
27443 cases where the register allocator uses a move and an offsettable
27444 memory operation to/from a FPR register on ISA 2.06/2.07.
27445 (define_peephole2 for Altivec d-form store): Likewise.
27447 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
27450 * config/i386/mmx.md (*mov<mode>_internal): Enable
27451 alternatives 11, 12, 13 and 14 also for 32bit targets.
27452 Remove alternatives 15, 16, 17 and 18.
27453 * config/i386/sse.md (vec_concatv2di): Change
27454 alternative (!x, *y) to (x, ?!*Yn).
27456 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
27458 * dumpfile.h (enum dump_kind): Remove stray comma.
27460 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27462 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
27463 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
27464 predicate::num_conditions
27465 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
27466 (CHANGED): turn into predicate::changed.
27467 (agg_position_info): Move to ipa-predicate.h
27468 (add_condition, predicate::add_clause, predicate::operator &=,
27469 predicate::or_with, predicate::evaluate, predicate::probability,
27470 dump_condition, dump_clause, predicate::dump,
27471 predicate::remap_after_duplication, predicate::remap_after_inlining,
27472 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
27473 (evaluate_conditions_for_known_args): Update.
27474 (set_cond_stmt_execution_predicate): Update.
27475 * ipa-inline.h: Include ipa-predicate.h
27476 (condition, inline_param_summary, conditions, agg_position_info,
27477 predicate): Move to ipa-predicate.h
27478 * ipa-predicate.c: New file.
27479 * ipa-predicate.h: New file.
27481 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
27483 * final.c (leaf_function_p): Check we are not in a sequence.
27485 2017-05-18 Martin Liska <mliska@suse.cz>
27487 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
27488 * dumpfile.c (dump_register): Use new enum dump_kind.
27489 (get_dump_file_name): Likewise.
27490 (dump_enable_all): Likewise.
27491 (dump_switch_p_1): Likewise.
27492 (enable_rtl_dump_file): Remove usage of TDF_RTL.
27493 * dumpfile.h (enum dump_kind): New enum type.
27494 (struct dump_file_info): Create constructor and
27495 format fields and comments.
27496 * passes.c (pass_manager::register_one_dump_file):
27498 * statistics.c (statistics_early_init): Likewise.
27499 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
27500 TDF_TREE with TDF_SLIM.
27501 (gather_memory_references_ref): Likewise.
27503 2017-05-18 Martin Liska <mliska@suse.cz>
27505 * vec.h (struct vnull): Use it.
27507 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
27509 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
27510 (true_predicate, false_predicate, true_predicate_p,
27511 false_predicate_p): Remove.
27512 (single_cond_predicate, not_inlined_predicate): Turn to member function
27514 (add_condition): Update.
27515 (add_clause): Turn to...
27516 (predicate::add_clause): ... this one; update; allow passing NULL
27518 (and_predicates): Turn to ...
27519 (predicate::operator &=): ... this one.
27520 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
27521 (or_predicates): Turn to ...
27522 (predicate::or_with): ... this one.
27523 (evaluate_predicate): Turn to ...
27524 (predicate::evaluate): ... this one.
27525 (predicate_probability): Turn to ...
27526 (predicate::probability): ... this one.
27527 (dump_condition): Update.
27528 (dump_predicate): Turn to ...
27529 (predicate::dump): ... this one.
27530 (account_size_time): Update.
27531 (edge_set_predicate): Update.
27532 (set_hint_predicate): UPdate.
27533 (evaluate_conditions_for_known_args): Update.
27534 (evaluate_properties_for_edge): Update.
27535 (remap_predicate_after_duplication): Turn to...
27536 (predicate::remap_after_duplication): ... this one.
27537 (remap_hint_predicate_after_duplication): Update.
27538 (inline_summary_t::duplicate): UPdate.
27539 (dump_inline_edge_summary): Update.
27540 (dump_inline_summary): Update.
27541 (set_cond_stmt_execution_predicate): Update.
27542 (set_switch_stmt_execution_predicate): Update.
27543 (compute_bb_predicates): Update.
27544 (will_be_nonconstant_expr_predicate): Update.
27545 (will_be_nonconstant_predicate): Update.
27546 (phi_result_unknown_predicate): Update.
27547 (predicate_for_phi_result): Update.
27548 (array_index_predicate): Update.
27549 (estimate_function_body_sizes): Update.
27550 (estimate_node_size_and_time): Update.
27551 (estimate_ipcp_clone_size_and_time): Update.
27552 (remap_predicate): Rename to ...
27553 (predicate::remap_after_inlining): ... this one.
27554 (remap_hint_predicate): Update.
27555 (inline_merge_summary): Update.
27556 (inline_update_overall_summary): Update.
27557 (estimate_size_after_inlining): Update.
27558 (read_predicate): Rename to ...
27559 (predicate::stream_in): ... this one.
27560 (read_inline_edge_summary): Update.
27561 (write_predicate): Rename to ...
27562 (predicate::stream_out): ... this one.
27563 (write_inline_edge_summary): Update.
27564 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
27565 (clause_t): Turn to uint32_t
27566 (predicate): Turn to class; implement constructor and operators
27568 (size_time_entry): Update.
27569 (inline_summary): Update.
27570 (inline_edge_summary): Update.
27572 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
27574 * fold-const.c (fold_binary_loc): Move transformation...
27575 * match.pd (C - X CMP X): ... here.
27577 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
27579 * config/sparc/sparc.c (sparc_option_override): Set function
27580 alignment for -mcpu=niagara7 to 64 to match the I$ line.
27581 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
27583 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
27585 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
27587 2017-05-18 Marek Polacek <polacek@redhat.com>
27590 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
27591 (pass_ubsan::execute): Call gimple_assign_single_p instead of
27592 gimple_assign_load_p.
27594 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
27596 PR middle-end/80692
27597 * real.c (do_compare): Give decimal_do_compare preference over
27598 comparing just the signs.
27600 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
27602 * doc/md.texi (Canonicalization of Instructions): Describe the
27603 canonical form of instructions that inherently set a condition
27606 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
27608 PR middle-end/80775
27609 * tree-cfg.c: Move deletion of unreachable case statements to after
27610 the merging of consecutive case labels.
27612 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
27614 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
27615 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
27616 restoring of callee-saved registers.
27618 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
27620 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
27621 * config/visium/visium.c (single_set_and_flags): Likewise.
27622 * config/visium/visium.md (Substitutions): Likewise.
27624 2017-05-17 Martin Liska <mliska@suse.cz>
27626 * cfg.c: Introduce dump_flags_t type and
27627 use it instead of int type.
27629 * cfghooks.c: Likewise.
27630 * cfghooks.h (struct cfg_hooks): Likewise.
27631 * cfgrtl.c: Likewise.
27632 * cfgrtl.h: Likewise.
27633 * cgraph.c (cgraph_node::get_body): Likewise.
27634 * coretypes.h: Likewise.
27635 * domwalk.c: Likewise.
27636 * domwalk.h: Likewise.
27637 * dumpfile.c (struct dump_option_value_info): Likewise.
27638 (dump_enable_all): Likewise.
27639 (dump_switch_p_1): Likewise.
27640 (opt_info_switch_p): Likewise.
27641 * dumpfile.h (enum tree_dump_index): Likewise.
27642 (struct dump_file_info): Likewise.
27643 * genemit.c: Likewise.
27644 * generic-match-head.c: Likewise.
27645 * gengtype.c (open_base_files): Likewise.
27646 * gimple-pretty-print.c: Likewise.
27647 * gimple-pretty-print.h: Likewise.
27648 * graph.c (print_graph_cfg): Likewise.
27649 * graphite-scop-detection.c (dot_all_sese): Likewise.
27650 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
27651 * loop-unroll.c (report_unroll): Likewise.
27652 * passes.c (pass_manager::register_one_dump_file): Likewise.
27653 * print-tree.c: Likewise.
27654 * statistics.c: Likewise.
27655 * tree-cfg.c: Likewise.
27656 * tree-cfg.h: Likewise.
27657 * tree-dfa.c: Likewise.
27658 * tree-dfa.h: Likewise.
27659 * tree-dump.c (dump_function): Likewise.
27660 * tree-dump.h (struct dump_info): Likewise.
27661 * tree-pretty-print.c: Likewise.
27662 * tree-pretty-print.h: Likewise.
27663 * tree-ssa-live.c: Likewise.
27664 * tree-ssa-live.h: Likewise.
27665 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
27666 * tree-vect-loop.c: Likewise.
27667 * tree-vect-slp.c: Likewise.
27669 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
27670 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27672 PR tree-optimization/80457
27673 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
27674 of all arguments to a statement as scalar_to_vec operations.
27675 (vectorizable_call): Adjust call to vect_model_simple_cost for
27677 (vectorizable_conversion): Likewise.
27678 (vectorizable_assignment): Likewise.
27679 (vectorizable_shift): Likewise.
27680 (vectorizable_operation): Likewise.
27681 (vectorizable_comparison): Likewise.
27682 (vect_is_simple_cond): Record the def types for operands.
27683 (vectorizable_condition): Likewise, call vect_model_simple_cost.
27684 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
27685 for statement argument count.
27687 2017-05-16 Carl Love <cel@us.ibm.com>
27689 * config/rs6000/rs6000-c: Add support for built-in functions
27690 vector unsigned long long vec_bperm (vector unsigned long long,
27691 vector unsigned char)
27692 vector signed long long vec_mule (vector signed int,
27694 vector unsigned long long vec_mule (vector unsigned int,
27695 vector unsigned int)
27696 vector signed long long vec_mulo (vector signed int,
27698 vector unsigned long long vec_mulo (vector unsigned int,
27699 vector unsigned int)
27700 vector signed char vec_sldw (vector signed char,
27701 vector signed char,
27703 vector unsigned char vec_sldw (vector unsigned char,
27704 vector unsigned char,
27706 vector signed short vec_sldw (vector signed short,
27707 vector signed short,
27709 vector unsigned short vec_sldw (vector unsigned short,
27710 vector unsigned short,
27712 vector signed int vec_sldw (vector signed int,
27715 vector unsigned int vec_sldw (vector unsigned int,
27716 vector unsigned int,
27718 vector signed long long vec_sldw (vector signed long long,
27719 vector signed long long,
27721 vector unsigned long long vec_sldw (vector unsigned long long,
27722 vector unsigned long long,
27724 * config/rs6000/rs6000-c: Add support for built-in functions
27725 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
27726 * config/rs6000/altivec.h: Add defintion for vec_sldw.
27727 * doc/extend.texi: Update the built-in documentation for the
27728 new built-in functions.
27730 2017-05-16 Marek Polacek <polacek@redhat.com>
27734 * tree.c (save_expr): Don't fold the expression.
27736 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
27738 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
27739 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
27740 and (?*y,m). Update insn attributes.
27742 2017-05-16 Martin Liska <mliska@suse.cz>
27744 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
27745 flags argument of print_gimple_stmt, print_gimple_expr,
27746 print_generic_stmt and print_generic_expr.
27747 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
27748 * coretypes.h: Likewise.
27749 * except.c (dump_eh_tree): Likewise.
27750 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
27751 * gimple-pretty-print.h: Likewise.
27752 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
27753 (backprop::push_to_worklist): Likewise.
27754 (backprop::pop_from_worklist): Likewise.
27755 (backprop::process_use): Likewise.
27756 (backprop::intersect_uses): Likewise.
27757 (note_replacement): Likewise.
27758 * gimple-ssa-store-merging.c
27759 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
27760 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
27761 (pass_store_merging::execute): Likewise.
27762 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
27763 (ssa_base_cand_dump_callback): Likewise.
27764 (dump_incr_vec): Likewise.
27765 (replace_refs): Likewise.
27766 (replace_mult_candidate): Likewise.
27767 (create_add_on_incoming_edge): Likewise.
27768 (create_phi_basis): Likewise.
27769 (insert_initializers): Likewise.
27770 (all_phi_incrs_profitable): Likewise.
27771 (introduce_cast_before_cand): Likewise.
27772 (replace_one_candidate): Likewise.
27773 * gimplify.c (gimplify_expr): Likewise.
27774 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
27775 (set_rename): Likewise.
27776 (rename_uses): Likewise.
27777 (copy_loop_phi_nodes): Likewise.
27778 (add_close_phis_to_merge_points): Likewise.
27779 (copy_loop_close_phi_args): Likewise.
27780 (copy_cond_phi_args): Likewise.
27781 (graphite_copy_stmts_from_block): Likewise.
27782 (translate_pending_phi_nodes): Likewise.
27783 * graphite-poly.c (print_pdr): Likewise.
27784 (dump_gbb_cases): Likewise.
27785 (dump_gbb_conditions): Likewise.
27786 (print_scop_params): Likewise.
27787 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
27788 (build_cross_bb_scalars_use): Likewise.
27789 (gather_bbs::before_dom_children): Likewise.
27790 * hsa-dump.c (dump_hsa_immed): Likewise.
27791 * ipa-cp.c (print_ipcp_constant_value): Likewise.
27792 (get_replacement_map): Likewise.
27793 * ipa-inline-analysis.c (dump_condition): Likewise.
27794 (estimate_function_body_sizes): Likewise.
27795 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
27796 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
27797 * ipa-prop.c (ipa_dump_param): Likewise.
27798 (ipa_print_node_jump_functions_for_edge): Likewise.
27799 (ipa_modify_call_arguments): Likewise.
27800 (ipa_modify_expr): Likewise.
27801 (ipa_dump_param_adjustments): Likewise.
27802 (ipa_dump_agg_replacement_values): Likewise.
27803 (ipcp_modif_dom_walker::before_dom_children): Likewise.
27804 * ipa-pure-const.c (check_stmt): Likewise.
27805 (pass_nothrow::execute): Likewise.
27806 * ipa-split.c (execute_split_functions): Likewise.
27807 * omp-offload.c (dump_oacc_loop_part): Likewise.
27808 (dump_oacc_loop): Likewise.
27809 * trans-mem.c (tm_log_emit): Likewise.
27810 (tm_memopt_accumulate_memops): Likewise.
27811 (dump_tm_memopt_set): Likewise.
27812 (dump_tm_memopt_transform): Likewise.
27813 * tree-cfg.c (gimple_verify_flow_info): Likewise.
27814 (print_loop): Likewise.
27815 * tree-chkp-opt.c (chkp_print_addr): Likewise.
27816 (chkp_gather_checks_info): Likewise.
27817 (chkp_get_check_result): Likewise.
27818 (chkp_remove_check_if_pass): Likewise.
27819 (chkp_use_outer_bounds_if_possible): Likewise.
27820 (chkp_reduce_bounds_lifetime): Likewise.
27821 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
27822 (chkp_mark_completed_bounds): Likewise.
27823 (chkp_register_incomplete_bounds): Likewise.
27824 (chkp_mark_invalid_bounds): Likewise.
27825 (chkp_maybe_copy_and_register_bounds): Likewise.
27826 (chkp_build_returned_bound): Likewise.
27827 (chkp_get_bound_for_parm): Likewise.
27828 (chkp_build_bndldx): Likewise.
27829 (chkp_get_bounds_by_definition): Likewise.
27830 (chkp_generate_extern_var_bounds): Likewise.
27831 (chkp_get_bounds_for_decl_addr): Likewise.
27832 * tree-chrec.c (chrec_apply): Likewise.
27833 * tree-data-ref.c (dump_data_reference): Likewise.
27834 (dump_subscript): Likewise.
27835 (dump_data_dependence_relation): Likewise.
27836 (analyze_overlapping_iterations): Likewise.
27837 * tree-inline.c (expand_call_inline): Likewise.
27838 (tree_function_versioning): Likewise.
27839 * tree-into-ssa.c (dump_defs_stack): Likewise.
27840 (dump_currdefs): Likewise.
27841 (dump_names_replaced_by): Likewise.
27842 (dump_update_ssa): Likewise.
27843 (update_ssa): Likewise.
27844 * tree-object-size.c (pass_object_sizes::execute): Likewise.
27845 * tree-parloops.c (build_new_reduction): Likewise.
27846 (try_create_reduction_list): Likewise.
27847 (ref_conflicts_with_region): Likewise.
27848 (oacc_entry_exit_ok_1): Likewise.
27849 (oacc_entry_exit_single_gang): Likewise.
27850 * tree-pretty-print.h: Likewise.
27851 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
27852 (get_scalar_evolution): Likewise.
27853 (add_to_evolution): Likewise.
27854 (get_loop_exit_condition): Likewise.
27855 (analyze_evolution_in_loop): Likewise.
27856 (analyze_initial_condition): Likewise.
27857 (analyze_scalar_evolution): Likewise.
27858 (instantiate_scev): Likewise.
27859 (number_of_latch_executions): Likewise.
27860 (gather_chrec_stats): Likewise.
27861 (final_value_replacement_loop): Likewise.
27862 (scev_const_prop): Likewise.
27863 * tree-sra.c (dump_access): Likewise.
27864 (disqualify_candidate): Likewise.
27865 (create_access): Likewise.
27866 (reject): Likewise.
27867 (maybe_add_sra_candidate): Likewise.
27868 (create_access_replacement): Likewise.
27869 (analyze_access_subtree): Likewise.
27870 (analyze_all_variable_accesses): Likewise.
27871 (sra_modify_assign): Likewise.
27872 (initialize_constant_pool_replacements): Likewise.
27873 (find_param_candidates): Likewise.
27874 (decide_one_param_reduction): Likewise.
27875 (replace_removed_params_ssa_names): Likewise.
27876 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
27877 * tree-ssa-copy.c (dump_copy_of): Likewise.
27878 (copy_prop_visit_cond_stmt): Likewise.
27879 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
27880 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
27881 (record_equivalences_from_stmt): Likewise.
27882 * tree-ssa-dse.c (compute_trims): Likewise.
27883 (delete_dead_call): Likewise.
27884 (delete_dead_assignment): Likewise.
27885 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
27886 (forward_propagate_into_cond): Likewise.
27887 (pass_forwprop::execute): Likewise.
27888 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27889 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27891 (move_computations_worker): Likewise.
27892 (execute_sm): Likewise.
27893 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27894 (remove_exits_and_undefined_stmts): Likewise.
27895 (remove_redundant_iv_tests): Likewise.
27896 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27897 (adjust_iv_update_pos): Likewise.
27898 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27899 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27900 (value_replacement): Likewise.
27901 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27902 * tree-ssa-pre.c (print_pre_expr): Likewise.
27903 (get_representative_for): Likewise.
27904 (create_expression_by_pieces): Likewise.
27905 (insert_into_preds_of_block): Likewise.
27906 (eliminate_insert): Likewise.
27907 (eliminate_dom_walker::before_dom_children): Likewise.
27908 (eliminate): Likewise.
27909 (remove_dead_inserted_code): Likewise.
27910 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27911 * tree-ssa-reassoc.c (get_rank): Likewise.
27912 (eliminate_duplicate_pair): Likewise.
27913 (eliminate_plus_minus_pair): Likewise.
27914 (eliminate_not_pairs): Likewise.
27915 (undistribute_ops_list): Likewise.
27916 (eliminate_redundant_comparison): Likewise.
27917 (update_range_test): Likewise.
27918 (optimize_range_tests_var_bound): Likewise.
27919 (optimize_vec_cond_expr): Likewise.
27920 (rewrite_expr_tree): Likewise.
27921 (rewrite_expr_tree_parallel): Likewise.
27922 (linearize_expr): Likewise.
27923 (break_up_subtract): Likewise.
27924 (linearize_expr_tree): Likewise.
27925 (attempt_builtin_powi): Likewise.
27926 (attempt_builtin_copysign): Likewise.
27927 (transform_stmt_to_copy): Likewise.
27928 (transform_stmt_to_multiply): Likewise.
27929 (dump_ops_vector): Likewise.
27930 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27931 (print_scc): Likewise.
27932 (set_ssa_val_to): Likewise.
27933 (visit_reference_op_store): Likewise.
27934 (visit_use): Likewise.
27935 (sccvn_dom_walker::before_dom_children): Likewise.
27936 (run_scc_vn): Likewise.
27937 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27939 (expr_hash_elt::print): Likewise.
27940 (const_and_copies::pop_to_marker): Likewise.
27941 (const_and_copies::record_const_or_copy_raw): Likewise.
27942 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27943 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27944 (dump_predicates): Likewise.
27945 (find_uninit_use): Likewise.
27946 (warn_uninitialized_phi): Likewise.
27947 (pass_late_warn_uninitialized::execute): Likewise.
27948 * tree-ssa.c (verify_vssa): Likewise.
27949 (verify_ssa): Likewise.
27950 (maybe_optimize_var): Likewise.
27951 * tree-vrp.c (dump_value_range): Likewise.
27952 (dump_all_value_ranges): Likewise.
27953 (dump_asserts_for): Likewise.
27954 (register_edge_assert_for_2): Likewise.
27955 (vrp_visit_cond_stmt): Likewise.
27956 (vrp_visit_switch_stmt): Likewise.
27957 (vrp_visit_stmt): Likewise.
27958 (vrp_visit_phi_node): Likewise.
27959 (simplify_cond_using_ranges_1): Likewise.
27960 (fold_predicate_in): Likewise.
27961 (evrp_dom_walker::before_dom_children): Likewise.
27962 (evrp_dom_walker::push_value_range): Likewise.
27963 (evrp_dom_walker::pop_value_range): Likewise.
27964 (execute_early_vrp): Likewise.
27966 2017-05-16 Richard Biener <rguenther@suse.de>
27968 * dwarf2out.c (loc_list_from_tree_1): Do not create
27969 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27971 2017-05-16 Richard Biener <rguenther@suse.de>
27973 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27975 (note_variable_value_in_expr): If we resolved the decl ref
27976 do not push to the stack.
27978 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27980 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27981 operations in fast-math mode.
27982 (vaddq_f16): Likewise.
27983 (vmul_f16): Likewise.
27984 (vmulq_f16): Likewise.
27985 (vsub_f16): Likewise.
27986 (vsubq_f16): Likewise.
27987 * config/arm/neon.md (add<mode>3): New.
27989 (fma:<VH:mode>3): New. Also remove outdated comment.
27992 2017-05-16 Martin Liska <mliska@suse.cz>
27996 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27997 (odr_types_equivalent_p): Likewise.
27999 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28001 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28003 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28006 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28007 non-interunit SSE move alternatives with '?'.
28008 (zero-extendsidi peephole2): New peephole to skip intermediate
28009 general register in SSE zero-extend sequence.
28011 2017-05-15 Jeff Law <law@redhat.com>
28013 * reorg.c (relax_delay_slots): Create a new variable to hold
28014 the temporary target rather than clobbering TARGET_LABEL.
28016 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28017 missing argument to extract_bit_field call.
28018 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28020 2017-05-15 Martin Liska <mliska@suse.cz>
28023 * gcc.c (process_command): Do not allow empty argument of -o option.
28025 2017-05-15 Renlin Li <renlin.li@arm.com>
28027 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28028 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28029 * config/aarch64/constraints.md (Usf): Add long call check.
28030 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28031 (call_value): Likewise.
28032 (sibcall): Likewise.
28033 (sibcall_value): Likewise.
28035 (call_value_insn): New.
28036 (sibcall_insn): Update rtx pattern.
28037 (sibcall_value_insn): Likewise.
28038 (call_internal): Remove.
28039 (call_value_internal): Likewise.
28040 (sibcall_internal): Likewise.
28041 (sibcall_value_internal): Likewise.
28042 (call_reg): Likewise.
28043 (call_symbol): Likewise.
28044 (call_value_reg): Likewise.
28045 (call_value_symbol): Likewise.
28047 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28050 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28052 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28054 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28055 compatible with CCGOCmode and with CCZmode.
28057 2017-05-14 Martin Sebor <msebor@redhat.com>
28059 PR middle-end/77671
28060 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28061 (gimple_fold_builtin_snprintf): Same.
28062 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28063 (gimple_fold_builtin_snprintf): Same.
28064 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28065 of character types.
28066 (is_call_safe): New function.
28067 (try_substitute_return_value): Call it.
28068 (try_simplify_call): New function.
28069 (pass_sprintf_length::handle_gimple_call): Call it.
28071 2017-05-14 Martin Sebor <msebor@redhat.com>
28073 PR middle-end/80669
28074 * builtins.c (expand_builtin_stpncpy): Simplify.
28076 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28078 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28079 * config/i386/i386.h
28080 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28081 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28082 (struct machine_function): Add new members call_ms2sysv,
28083 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28084 (struct machine_frame_state): New fields sp_realigned and
28085 sp_realigned_offset.
28086 * config/i386/i386.c
28087 (enum xlogue_stub): New enum.
28088 (enum xlogue_stub_sets): New enum.
28089 (class xlogue_layout): New class.
28090 (struct ix86_frame): New fields stack_realign_allocate_offset,
28091 stack_realign_offset and outlined_save_offset. Modify comments to
28092 detail stack layout when using out-of-line stubs.
28093 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28094 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28095 -mcall-ms2sysv-xlogues.
28096 (stub_managed_regs): New static variable.
28097 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28098 registers managed by out-of-line stub.
28099 (disable_call_ms2sysv_xlogues): New function.
28100 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28101 m->call_ms2sysv when appropriate and compute frame layout for
28103 (sp_valid_at, fp_valid_at): New inline functions.
28104 (choose_basereg): New function.
28105 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28107 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28108 Use align parameter of choose_baseaddr to generated aligned SSE movs
28110 (pro_epilogue_adjust_stack): Modify to track
28111 machine_frame_state::sp_realigned.
28112 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28113 (ix86_nsaved_sseregs): Likewise.
28114 (ix86_emit_save_regs): Likewise.
28115 (ix86_emit_save_regs_using_mov): Likewise.
28116 (ix86_emit_save_sse_regs_using_mov): Likewise.
28117 (get_scratch_register_on_entry): Likewise.
28118 (gen_frame_set): New function.
28119 (gen_frame_load): Likewise.
28120 (gen_frame_store): Likewise.
28121 (emit_outlined_ms2sysv_save): Likewise.
28122 (emit_outlined_ms2sysv_restore): Likewise.
28123 (ix86_expand_prologue): Modify stack re-alignment code and call
28124 emit_outlined_ms2sysv_save when appropriate.
28125 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28126 parameter rtx_insn *insn, which allows the function to be used to only
28127 generate the notes.
28128 (ix86_expand_epilogue): Modify validity checks of frame and stack
28129 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28130 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28131 * config/i386/predicates.md
28132 (save_multiple): New predicate.
28133 (restore_multiple): Likewise.
28134 * config/i386/sse.md
28135 (save_multiple<mode>): New pattern.
28136 (save_multiple_realign<mode>): Likewise.
28137 (restore_multiple<mode>): Likewise.
28138 (restore_multiple_and_return<mode>): Likewise.
28139 (restore_multiple_leave_return<mode>): Likewise.
28140 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28142 2017-05-14 Julia Koval <julia.koval@intel.com>
28144 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28145 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28146 (__builtin_ia32_xsetbv): New builtins.
28147 * config/i386/i386.c (ix86_expand_special_args_builtin):
28149 (ix86_expand_builtin): Special expand for new intrinsics.
28150 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28151 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28152 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28154 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28156 * cfganal.c (inverted_post_order_compute): Change argument type
28158 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28159 * df-core.c (rest_of_handle_df_initialize): Adjust.
28160 (rest_of_handle_df_finish): Likewise.
28161 (df_analyze_1): Likewise.
28162 (df_analyze): Likewise.
28163 (loop_inverted_post_order_compute): Change argument to be a vec *.
28164 (df_analyze_loop): Adjust.
28165 (df_get_n_blocks): Likewise.
28166 (df_get_postorder): Likewise.
28167 * df.h (struct df_d): Change field to be a vec.
28168 * lcm.c (compute_laterin): Adjust.
28169 (compute_available): Likewise.
28170 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28171 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28172 * tree-ssa-pre.c (compute_antic): Likewise.
28174 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28176 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28177 (depth_first_search::depth_first_search): Change structure init
28178 function to this constructor.
28179 (depth_first_search::add_bb): Rename function to this member.
28180 (depth_first_search::execute): Likewise.
28181 (flow_dfs_compute_reverse_finish): Adjust.
28183 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28185 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28186 (longest_simple_path): Likewise.
28187 * shrink-wrap.c (spread_components): Likewise.
28188 (disqualify_problematic_components): Likewise.
28189 (emit_common_heads_for_components): Likewise.
28190 (emit_common_tails_for_components): Likewise.
28191 (insert_prologue_epilogue_for_components): Likewise.
28193 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28195 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28198 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28200 * df-core.c (df_set_blocks): Start using auto_bitmap.
28201 (df_compact_blocks): Likewise.
28202 * df-problems.c (df_rd_confluence_n): Likewise.
28203 * df-scan.c (df_insn_rescan_all): Likewise.
28204 (df_process_deferred_rescans): Likewise.
28205 (df_update_entry_block_defs): Likewise.
28206 (df_update_exit_block_uses): Likewise.
28207 (df_entry_block_bitmap_verify): Likewise.
28208 (df_exit_block_bitmap_verify): Likewise.
28209 (df_scan_verify): Likewise.
28210 * lra-constraints.c (lra_constraints): Likewise.
28211 (undo_optional_reloads): Likewise.
28212 (lra_undo_inheritance): Likewise.
28213 * lra-remat.c (calculate_gen_cands): Likewise.
28214 (do_remat): Likewise.
28215 * lra-spills.c (assign_spill_hard_regs): Likewise.
28216 (spill_pseudos): Likewise.
28217 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28218 (bitmap_set_subtract_values): Likewise.
28220 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28222 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28223 management with auto_bitmap.
28224 (fix_inter_tick): Likewise.
28225 (fix_recovery_deps): Likewise.
28226 * ira.c (add_store_equivs): Likewise.
28227 (find_moveable_pseudos): Likewise.
28228 (split_live_ranges_for_shrink_wrap): Likewise.
28229 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28230 (rtx_reuse_manager::seen_def_p): Likewise.
28231 (rtx_reuse_manager::set_seen_def): Likewise.
28232 * print-rtl.h (class rtx_reuse_manager): Likewise.
28234 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28236 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28238 (migrate_btr_def): Likewise.
28239 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28240 * df-core.c (loop_post_order_compute): Likewise.
28241 (loop_inverted_post_order_compute): Likewise.
28242 * hsa-common.h: Likewise.
28243 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28244 * init-regs.c (initialize_uninitialized_regs): Likewise.
28245 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28246 (inline_small_functions): Likewise.
28247 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28248 * ira.c (combine_and_move_insns): Likewise.
28249 (build_insn_chain): Likewise.
28250 * loop-invariant.c (find_invariants): Likewise.
28251 * lower-subreg.c (propagate_pseudo_copies): Likewise.
28252 * predict.c (tree_predict_by_opcode): Likewise.
28253 (predict_paths_leading_to): Likewise.
28254 (predict_paths_leading_to_edge): Likewise.
28255 (estimate_loops_at_level): Likewise.
28256 (estimate_loops): Likewise.
28257 * shrink-wrap.c (try_shrink_wrapping): Likewise.
28258 (spread_components): Likewise.
28259 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
28260 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
28261 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
28262 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
28263 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
28264 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
28265 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
28266 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
28267 (mark_threaded_blocks): Likewise.
28268 (thread_through_all_blocks): Likewise.
28269 * tree-ssa.c (verify_ssa): Likewise.
28270 (execute_update_addresses_taken): Likewise.
28271 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
28273 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28275 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
28277 (post_order_compute): Likewise.
28278 (inverted_post_order_compute): Likewise.
28279 (pre_and_rev_post_order_compute_fn): Likewise.
28281 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28283 * genrecog.c (int_set::int_set): Explicitly construct our
28284 auto_vec base class.
28285 * vec.h (auto_vec::auto_vec): New constructor.
28287 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28289 * bitmap.h (class auto_bitmap): New constructor taking
28290 bitmap_obstack * argument.
28292 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28294 * bitmap.h (class auto_bitmap): Change type of m_bits to
28295 bitmap_head, and adjust ctor / dtor and member operators.
28297 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
28299 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
28300 when returned register mode doesn't match original mode.
28302 2017-05-12 Jeff Law <law@redhat.com>
28303 Jakub Jelinek <jakub@redhat.com>
28305 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
28306 we look for cc setter after the compare-elim changes.
28307 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
28308 within the vector to match what compare-elim now expects.
28309 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
28310 (xorsi3_flags, one_cmplsi2_flags): Likewise.
28312 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
28313 after the compare-elim changes.
28314 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
28315 the vector to match what compare-elim now expects.
28316 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
28317 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
28318 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
28319 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
28320 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
28322 * config/visium/visium.c (single_set_and_flags): Fix where
28323 we look for cc setter after the compare-elim changes.
28324 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
28325 with the vector to match what compare-elim now expects.
28326 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
28327 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
28328 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
28329 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
28330 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
28331 (neg<mode>2_insn_set_overflow): Likewise.
28333 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
28335 PR middle-end/79794
28336 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
28337 maybe_expand_insn call, set ops[0].target. If still set after call,
28338 set alt_rtl. Add extra arg to recursive calls.
28339 (extract_bit_field): Add alt_rtl argument. Pass to
28341 * expmed.h (extract_bit_field): Fix prototype.
28342 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
28343 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
28344 to extract_bit_field_calls.
28345 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
28346 Pass alt_rtl to extract_bit_field calls.
28347 * calls.c (store_unaligned_arguments_into_psuedos)
28348 load_register_parameters): Pass extra NULL to extract_bit_field calls.
28349 * optabs.c (maybe_legitimize_operand): Clear op->target when call
28351 * optabs.h (struct expand_operand): Add target bitfield.
28353 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28355 * compare-elim.c (try_eliminate_compare): Canonicalize
28356 operation with embedded compare to
28357 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
28358 (set (reg) (operation)].
28360 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
28362 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
28365 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
28366 cost of adding a carry flag for ADC instruction.
28367 [case MINUS]: Ignore the cost of subtracting a carry flag
28368 for SBB instruction.
28370 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
28372 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
28374 * config/rs6000/bmiintrin.h: New file.
28375 * config/rs6000/bmi2intrin.h: New file.
28376 * config/rs6000/x86intrin.h: New file.
28378 2017-05-12 Jeff Law <law@redhat.com>
28380 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
28383 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
28385 PR middle-end/80707
28386 * tree-cfg.c: Remove cfg edges of unreachable case statements.
28388 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28390 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28391 early expansion of vector divide builtins.
28392 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
28393 builtins identified as having unsigned arguments.
28395 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28397 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
28398 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
28399 expansion of vector logical operations (and, andc, or, xor,
28402 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28404 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
28405 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
28407 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
28409 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
28410 early GIMPLE expansion of vector multiplies.
28412 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28414 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
28415 TARGET_HAVE_MOVT conditional.
28416 (movt splitter): Likewise.
28418 2017-05-12 Richard Biener <rguenther@suse.de>
28420 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
28421 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28422 Fold all stmts not inplace.
28424 2017-05-12 Richard Biener <rguenther@suse.de>
28426 PR tree-optimization/80713
28427 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
28428 inserted_exprs bit for not removed stmts.
28430 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
28432 PR middle-end/69921
28433 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
28434 parallelized" attribute for parallelized OpenACC kernels.
28435 * omp-offload.c (execute_oacc_device_lower): Use it.
28437 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
28438 Set "oacc kernels" attribute.
28439 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
28440 parameter. Adjust all users.
28441 (oacc_fn_attrib_kernels_p): Remove function.
28442 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
28443 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
28444 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
28445 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
28446 assert "oacc kernels" attribute is set.
28448 2017-05-11 Carl Love <cel@us.ibm.com>
28450 * config/rs6000/rs6000-c: Add support for built-in functions
28451 vector unsigned char vec_popcnt (vector signed char)
28452 vector unsigned char vec_popcnt (vector unsigned char)
28453 vector unsigned short vec_popcnt (vector signed short)
28454 vector unsigned short vec_popcnt (vector unsigned short)
28455 vector unsigned int vec_popcnt (vector signed int)
28456 vector unsigned int vec_popcnt (vector unsigned int)
28457 vector unsigned long long vec_popcnt (vector signed long long)
28458 vector unsigned long long vec_popcnt (vector unsigned long long)
28459 vector signed long long vec_slo (vector signed long long,
28460 vector signed char)
28461 vector signed long long vec_slo (vector signed long long,
28462 vector unsigned char)
28463 vector unsigned long long vec_slo (vector unsigned long long,
28464 vector signed char)
28465 vector unsigned long long vec_slo (vector unsigned long long,
28466 vector unsigned char)
28467 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
28468 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
28469 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
28470 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
28471 * doc/extend.texi: Update the built-in documentation file for the
28472 new built-in functions.
28474 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28476 * attribs.h (sorted_attr_string): Move machine independent
28477 functions for target clone support from the i386 port to common
28478 code. Rename ix86_function_versions to common_function_versions.
28479 Rename make_name to make_unique_name.
28480 (common_function_versions): Likewise.
28481 (make_unique_name): Likewise.
28482 (make_dispatcher_decl): Likewise.
28483 (is_function_default_version): Likewise.
28484 * attribs.c (attr_strcmp): Likewise.
28485 (sorted_attr_string): Likewise.
28486 (common_function_versions): Likewise.
28487 (make_unique_name): Likewise.
28488 (make_dispatcher_decl): Likewise.
28489 (is_function_default_version): Likewise.
28490 * config/i386/i386.c (attr_strcmp): Likewise.
28491 (sorted_attr_string): Likewise.
28492 (ix86_function_versions): Likewise.
28493 (make_name): Likewise.
28494 (make_dispatcher_decl): Likewise.
28495 (is_function_default_version): Likewise.
28496 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
28498 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28501 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
28502 Account for direct move costs for vec_construct of integer
28505 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
28508 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
28509 (UNSPEC_STX_ATOMIC): Ditto.
28510 (loaddi_via_sse): New insn.
28511 (storedi_via_sse): Ditto.
28512 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
28513 Update corresponding peephole2 patterns.
28514 (atomic_storedi_fpu): Ditto.
28516 2017-05-11 Julia Koval <julia.koval@intel.com>
28518 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
28519 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
28521 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
28522 (__builtin_ia32_rsqrt14ss_mask): New builtins.
28523 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
28525 2017-05-11 Nathan Sidwell <nathan@acm.org>
28527 * graphite-poly.c: Include dumpfile.h.
28529 * dumpfle.h (dump_function): Declare here ...
28530 * tree-dump.h (dump_function): ... not here.
28531 * dumpfile.c: #include tree-cfg.h.
28532 (dump_function): Move here from ...
28533 * tree-dump.c (dump_function): ... here.
28534 * gimplify.c: #include splay-tree.h, not tree-dump.h.
28535 * graphite-poly.c: Don't include tree-dump.h.
28536 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
28537 * print-tree.c: Likewise.
28538 * stor-layout.c: Likewise.
28539 * tree-nested.c: Likewise.
28541 * dumpfile.c (dump_start): Use TDF_FLAGS.
28542 (dump_enable_all): Fix TDF_KIND check thinko.
28544 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
28546 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28547 array entries to represent two legal parameterizations of the
28548 overloaded __builtin_cmpb function, as represented by the
28549 P6_OV_BUILTIN_CMPB constant.
28550 (altivec_resolve_overloaded_builtin): Add special case handling
28551 for the __builtin_cmpb function, as represented by the
28552 P6_OV_BUILTIN_CMPB constant.
28553 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
28554 (BU_P6_64BIT_2): New macro.
28555 (BU_P6_OVERLOAD_2): New macro
28556 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
28557 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
28558 (CMPB): Add overload support to represent both 32-bit and 64-bit
28559 compare-bytes function.
28560 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
28561 support for TARGET_CMPB.
28562 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
28563 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28564 documentation of the __builtin_cmpb overloaded built-in function.
28566 2017-05-11 Richard Biener <rguenther@suse.de>
28568 PR tree-optimization/80705
28569 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
28570 bases are not vectorizable.
28572 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28574 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
28575 when counting register pressure.
28577 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28579 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
28580 (iv_ca_more_deps): Renamed to ...
28581 (iv_ca_compare_deps): ... this.
28582 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
28584 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28586 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
28588 (determine_group_iv_costs): ... here.
28589 (find_inv_vars_cb): Record inv var if it's not recorded before.
28591 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28593 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
28594 (get_shiftadd_cost): Ditto.
28596 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28598 * tree-ssa-address.c: Include header file.
28599 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
28601 (add_to_parts): Refactor.
28602 (addr_to_parts): New parameter. Update use of move_hint_to_base.
28603 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
28606 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28608 PR tree-optimization/53090
28609 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
28611 (extract_cond_operands): Detect condition with IV on both sides
28612 and return COMP_IV_EXPR_2.
28613 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
28614 (rewrite_use_compare): Simplify by removing call to function
28615 extract_cond_operands.
28617 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28619 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
28620 (extract_cond_operands): Detect condition comparing against non-
28621 invariant bound and return appropriate enum value.
28622 (find_interesting_uses_cond): Update use of extract_cond_operands.
28623 Handle its return value accordingly.
28624 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
28626 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28628 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
28629 nonlinear iv_use computation in loop invariant sensitive way.
28631 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28633 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
28634 (find_iv_candidates): Call relate_compare_use_with_all_cands.
28636 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28638 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
28639 (dump_cand): Support iv_cand.inv_exprs.
28640 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
28642 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
28645 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28647 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
28649 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
28650 as local function. Include necessary header files.
28651 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
28653 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28655 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
28657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28659 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
28660 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
28661 RSHIFT_EXPR and BIT_NOT_EXPR.
28663 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28665 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
28666 (adjust_setup_cost): New parameter supporting round up adjustment.
28667 (struct address_cost_data): Delete.
28668 (force_expr_to_var_cost): Don't bound cost with spill_cost.
28669 (split_address_cost, ptr_difference_cost): Delete.
28670 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
28671 (struct ainc_cost_data): New struct.
28672 (get_address_cost_ainc): New function.
28673 (get_address_cost, get_computation_cost): Reimplement.
28674 (determine_group_iv_cost_address): Record inv_expr for all uses of
28676 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
28677 (iv_ca_has_deps): Reimplemented to ...
28678 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
28680 (iv_ca_extend): Call iv_ca_more_deps.
28682 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28684 * tree-ssa-address.c (struct mem_address): Move to header file.
28685 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
28686 * tree-ssa-address.h (struct mem_address): Move from C file.
28687 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
28689 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28691 * tree-affine.h (aff_combination_type): New interface.
28692 (aff_combination_zero_p): Remove static.
28693 (aff_combination_const_p): New interface.
28694 (aff_combination_singleton_var_p): New interfaces.
28696 2017-05-11 Richard Biener <rguenther@suse.de>
28698 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
28699 Skip unreachable blocks and destinations.
28700 (eliminate): Move stmt removal and fixup ...
28701 (fini_eliminate): ... here. Skip inserted exprs.
28702 (pass_pre::execute): Move fini_pre after fini_eliminate.
28703 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
28704 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
28705 PRE to get rid of dead code that has invalid SSA form and
28706 split critical edges again.
28708 2017-05-11 Bin Cheng <bin.cheng@arm.com>
28710 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28712 2017-05-11 Richard Biener <rguenther@suse.de>
28714 * passes.c (execute_function_todo): Verify loops if they are
28715 said to be up-to-date.
28716 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
28717 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
28719 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
28722 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
28723 handle calling assemble_external ourself.
28726 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
28727 modes with zero size. Enhance comment.
28729 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28731 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
28732 built-ins for vec_xl and vec_xst with short and char pointer
28735 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28737 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
28738 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
28739 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
28740 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
28741 (_mm_maskz_min_round_ss): New intrinsics.
28742 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
28743 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
28744 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
28745 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
28746 (__builtin_ia32_minss_mask_round): New builtins.
28747 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28748 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28749 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
28751 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
28752 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
28754 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
28757 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
28759 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
28760 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
28761 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
28762 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
28763 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
28764 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
28765 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
28766 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28767 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28768 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
28769 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
28770 (__builtin_ia32_mulss_mask_round): New builtins.
28771 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28772 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28773 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
28775 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
28776 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28778 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28781 2017-05-10 Julia Koval <julia.koval@intel.com>
28783 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
28784 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
28785 (_mm256_setr_m128i): New intrinsics.
28787 2017-05-10 Julia Koval <julia.koval@intel.com>
28789 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
28790 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
28791 (_mm_maskz_rcp14_ss): New intrinsics.
28792 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
28793 (__builtin_ia32_rcp14ss_mask): New builtins.
28794 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
28796 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
28798 PR tree-optimization/51513
28799 * tree-cfg.c (gimple_seq_unreachable_p): New function.
28800 (assert_unreachable_fallthru_edge_p): Use it.
28801 (group_case_labels_stmt): Likewise.
28802 * tree-cfg.h: Prototype it.
28803 * stmt.c: Include cfghooks.h and tree-cfg.h.
28804 (emit_case_dispatch_table) <gap_label>: New local variable.
28805 Use it to fill dispatch table gaps.
28806 Test for default_label before updating probabilities.
28807 (expand_case) <default_label>: Remove unneeded initialization.
28808 Test for unreachable default case statement and remove its edge.
28809 Set default_label accordingly.
28810 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
28812 2017-05-10 Carl Love <cel@us.ibm.com>
28814 * config/rs6000/rs6000-c: Add support for built-in functions
28815 vector signed char vec_neg (vector signed char)
28816 vector signed short int vec_neg (vector short int)
28817 vector signed int vec_neg (vector signed int)
28818 vector signed long long vec_neg (vector signed long long)
28819 vector float vec_neg (vector float)
28820 vector double vec_neg (vector double)
28821 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
28823 * config/rs6000/altivec.h: Add define for vec_neg
28824 * doc/extend.texi: Update the built-in documentation for the
28825 new built-in functions.
28827 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28829 PR tree-optimization/77644
28830 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
28832 2017-05-10 Nathan Sidwell <nathan@acm.org>
28834 * dumpfile.h (TDI_lang_all): New.
28835 (TDF_KIND): New. Renumber others
28836 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
28838 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
28840 (get_dump_file_name): Adjust suffix generation.
28841 (dump_enable_all): Use TDF_KIND.
28842 * doc/invoke.texi (-fdump-lang-all): Document.
28844 * dumpfile.h: Tabify.
28846 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
28849 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
28850 Move member access before delete.
28852 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
28854 * tree-inline.c (expand_call_inline): Split block at stmt
28857 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
28860 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
28861 are now unused after splitting mov{sf,sd}_hardfloat.
28862 (f32_lr2): Likewise.
28863 (f32_lm): Likewise.
28864 (f32_lm2): Likewise.
28865 (f32_li): Likewise.
28866 (f32_li2): Likewise.
28867 (f32_lv): Likewise.
28868 (f32_sr): Likewise.
28869 (f32_sr2): Likewise.
28870 (f32_sm): Likewise.
28871 (f32_sm2): Likewise.
28872 (f32_si): Likewise.
28873 (f32_si2): Likewise.
28874 (f32_sv): Likewise.
28875 (f32_dm): Likewise.
28876 (f32_vsx): Likewise.
28877 (f32_av): Likewise.
28878 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
28879 For movsf, order stores so the VSX stores occur before the GPR
28880 store which encourages the register allocator to use a traditional
28881 FPR instead of a GPR. For movsd, order the stores so that the GPR
28882 store comes before the VSX stores to allow the power6 to work.
28883 This is due to the power6 not having a 32-bit integer store
28884 instruction from a FPR.
28885 (movsf_hardfloat): Likewise.
28886 (movsd_hardfloat): Likewise.
28888 2017-05-09 Martin Sebor <msebor@redhat.com>
28890 PR translation/80280
28891 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28894 PR translation/80280
28895 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28896 data member added in r247778.
28897 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28899 2017-05-09 Nathan Sidwell <nathan@acm.org>
28901 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28903 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28906 2017-05-09 Marek Polacek <polacek@redhat.com>
28908 * doc/invoke.texi: Fix typo.
28910 2017-05-09 Richard Biener <rguenther@suse.de>
28912 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28913 (vrp_val_is_min): Likewise.
28914 (set_value_range_to_value): Likewise.
28915 (set_value_range_to_nonnegative): Likewise.
28916 (gimple_assign_nonzero_p): Likewise.
28917 (gimple_stmt_nonzero_p): Likewise.
28918 (vrp_int_const_binop): Likewise. Remove unreachable case.
28919 (adjust_range_with_scev): Adjust comments.
28920 (compare_range_with_value): Likewise.
28921 (extract_range_from_phi_node): Likewise.
28922 (test_for_singularity): Likewise.
28924 2017-05-09 Richard Biener <rguenther@suse.de>
28926 * tree-vrp.c (get_single_symbol): Add assert that we don't
28927 get overflowed constants as invariant part.
28928 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28929 checks. Use wi::cmp instead of recursing for integer constants.
28930 (compare_values): Just ignore whether we assumed undefined
28931 overflow instead of failing the compare.
28932 (extract_range_for_var_from_comparison_expr): Add comment before the
28933 TREE_NO_WARNING sets.
28934 (test_for_singularity): Likewise.
28935 (extract_range_from_comparison): Do not disable optimization
28936 when we assumed undefined overflow.
28937 (extract_range_basic): Remove init of unused var.
28939 2017-05-09 Richard Biener <rguenther@suse.de>
28941 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28942 (extract_range_from_multiplicative_op_1): Adjust.
28943 (extract_range_from_binary_expr_1): Use int_const_binop.
28945 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28948 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28949 rs6000_store_data_bypass_p in seven define_bypass directives and
28950 in several comments.
28951 * config/rs6000/rs6000-protos.h: Add prototype for
28952 rs6000_store_data_bypass_p function.
28953 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28954 function implements slightly different (rs6000-specific) semantics
28955 than store_data_bypass_p, returning false rather than aborting
28956 with assertion error when arguments do not satisfy the
28957 requirements of store data bypass.
28958 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28959 rs6000_store_data_bypass_p.
28961 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28963 * config/xtensa/xtensa-protos.h
28964 (xtensa_initial_elimination_offset): New declaration.
28965 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28966 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28967 macro definition, add case for FRAME_POINTER_REGNUM when
28968 FRAME_GROWS_DOWNWARD.
28969 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28970 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28971 xtensa_initial_elimination_offset.
28973 2017-05-08 Nathan Sidwell <nathan@acm.org>
28975 * doc/invoke.texi: Alphabetize -fdump options.
28977 2017-05-08 Martin Sebor <msebor@redhat.com>
28979 PR translation/80280
28980 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28982 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28984 * target.def (compute_frame_layout): New optional target hook.
28985 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28986 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28987 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28989 * reload1.c (verify_initial_elim_offsets): Likewise.
28990 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28991 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28992 (arm_get_frame_offsets): Split up into this ...
28993 (arm_compute_frame_layout): ... and this function.
28995 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28997 * config/aarch64/constraints.md (Usa): New constraint.
28998 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29000 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29002 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29003 with_multilib_list after it has been checked.
29005 2017-05-08 Richard Biener <rguenther@suse.de>
29007 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29008 (bitmap_set_subtract_values): Likewise.
29010 2017-05-08 Richard Biener <rguenther@suse.de>
29012 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29013 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29015 (gimple_stmt_nonzero_warnv_p): Rename to ...
29016 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29018 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29019 (extract_range_basic): Adjust, do not disable propagation on
29020 strict overflow sensitive simplification.
29021 (vrp_visit_cond_stmt): Likewise.
29023 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29025 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29026 body size unconditionally.
29028 2017-05-07 Jeff Law <law@redhat.com>
29031 2017-05-06 Jeff Law <law@redhat.com>
29032 PR tree-optimization/78496
29033 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29036 PR tree-optimization/78496
29037 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29038 (simplify_stmt_using_ranges): Call it.
29039 (vrp_dom_walker::before_dom_children): Extract equivalences
29040 from an ASSERT_EXPR with an equality comparison against a
29043 2017-05-06 Jeff Law <law@redhat.com>
29045 PR tree-optimization/78496
29046 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29049 PR tree-optimization/78496
29050 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29051 (simplify_stmt_using_ranges): Call it.
29052 (vrp_dom_walker::before_dom_children): Extract equivalences
29053 from an ASSERT_EXPR with an equality comparison against a
29056 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29058 * lra-constraints.c (lra_copy_reg_equiv): New function.
29059 (split_reg): Use it to copy equivalence information from the
29060 original register to the spill register.
29062 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29064 PR rtl-optimization/75964
29065 * simplify-rtx.c (simplify_const_relational_operation): Remove
29066 invalid handling of comparisons of integer ABS.
29068 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29070 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29071 initialize to zero.
29072 (init_regs): Remove declaration.
29073 (function_arg_advance_32): Initialize error_p as boolean variable.
29075 2017-05-05 Nathan Sidwell <nathan@acm.org>
29077 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29078 lines. Use for (;;).
29080 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29082 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29083 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29084 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29085 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29086 VF=2 that require versioning.
29088 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29090 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29093 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29095 * diagnostic.h (diagnostic_override_option_index): Convert from
29096 macro to inline function.
29098 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29100 * diagnostic.c (last_module_changed_p): New function.
29101 (set_last_module): New function.
29102 (diagnostic_report_current_module): Convert macro usage to
29103 the above functions.
29104 * diagnostic.h (diagnostic_context::last_module): Strengthen
29105 from const line_map * to const line_map_ordinary *.
29106 (diagnostic_last_module_changed): Delete macro.
29107 (diagnostic_set_last_module): Delete macro.
29109 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29111 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29112 with diagnostic_report_diagnostic.
29113 (diagnostic_n_impl_richloc): Likewise.
29114 * diagnostic.h (report_diagnostic): Delete macro.
29115 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29116 with diagnostic_report_diagnostic.
29117 * substring-locations.c (format_warning_va): Likewise.
29119 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29121 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29122 save/restor of format_spec. Move option-printing code to...
29123 (print_option_information): ...this new function, and
29124 reimplement by simply printing to the pretty_printer,
29125 rather than appending to the format string.
29127 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29129 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29130 handling logic into...
29131 (update_effective_level_from_pragmas): ...this new function.
29133 2017-05-04 Andrew Waterman <andrew@sifive.com>
29135 * config/riscv/riscv.opt (mstrict-align): New option.
29136 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29137 (SLOW_UNALIGNED_ACCESS): Define.
29138 (riscv_slow_unaligned_access): Declare.
29139 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29141 (riscv_slow_unaligned_access): New variable.
29142 (rocket_tune_info): Set slow_unaligned_access to true.
29143 (optimize_size_tune_info): Set slow_unaligned_access to false.
29144 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29145 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29146 (riscv_option_override): Set riscv_slow_unaligned_access.
29147 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29149 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29151 * config/riscv/riscv.md: Unify indentation.
29153 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29158 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29160 (extendsi<mode>2): Add support for doing sign extension via
29161 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29162 don't have ISA 3.0 instructions.
29163 (extendsi<mode>2 splitter): Likewise.
29164 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29165 generate the normal insns since SImode can now go in vector
29166 registers. Disallow the special UNSPECs needed for previous
29167 machines to hide SImode being used. Add new insns
29168 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29169 (fix_trunc<mode>si2_stfiwx): Likewise.
29170 (fix_trunc<mode>si2_internal): Likewise.
29171 (fixuns_trunc<mode>si2): Likewise.
29172 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29173 (fctiw<u>z_<mode>_smallint): Likewise.
29174 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29175 of floating point to 32-bit integer from doing a direct move to
29176 the GPR registers to do a store.
29177 (fctiwz_<mode>): Break long line.
29179 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29181 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29182 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29183 (addr_list, addr_offset_valid_p): New.
29184 (split_address_groups): Check offset validity with above function.
29185 (gt-tree-ssa-loop-ivopts.h): Include header file.
29187 2017-05-05 Nathan Sidwell <nathan@acm.org>
29189 * config.gcc (arm*-*-*): Add missing 'fi'.
29191 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29193 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29194 included in -fopt-info does not matter.
29195 * doc/optinfo.texi (-fopt-info): Fix description of default
29196 behavour. Explicitly say order of options included in -fopt-info
29199 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29201 * config.gcc: Allow combinations of aprofile and rmprofile values for
29202 --with-multilib-list.
29203 * config/arm/t-multilib: New file.
29204 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29205 variables. Remove setting of ISA and floating-point ABI in
29206 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29207 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29208 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29209 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29211 * config/arm/t-rmprofile: Likewise except for the matches changes.
29212 * doc/install.texi (--with-multilib-list): Document the combination of
29213 aprofile and rmprofile values and warn about pitfalls in doing that.
29215 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29217 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29218 (movdi_aarch64): Likewise.
29220 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29222 PR tree-optimization/80632
29223 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29225 (build_arrays): Initialize it for virtual phis.
29226 (fix_phi_nodes): Use it for virtual phis.
29228 PR tree-optimization/80558
29229 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29230 [x, y] op z into [x op, y op z] for op & or | if conditions
29233 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29234 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29237 * config/arm/arm.md (use_literal_pool): Remove.
29238 (64-bit immediate split): No longer takes cost into consideration
29239 if arm_disable_literal_pool is enabled.
29240 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29241 used when arm_disable_literal_pool is enabled.
29242 (arm_max_const_double_inline_cost): Remove use of
29243 arm_disable_literal_pool.
29244 (push_minipool_fix): Add assert.
29245 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29246 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29247 (no_literal_pool_sf_immediate): New.
29249 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29251 PR tree-optimization/80613
29252 * tree-ssa-dce.c (propagate_necessity): Remove cases for
29253 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
29255 2017-05-05 Richard Biener <rguenther@suse.de>
29257 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
29259 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
29261 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
29262 of this flag from insn conditions due to removal from r247495.
29264 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29266 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
29268 (arm_early_store_addr_dep_ptr): Likewise.
29269 * config/arm/aarch-common-protos.h
29270 (arm_early_load_addr_dep_ptr): Add prototype.
29271 (arm_early_store_addr_dep_ptr): Likewise.
29272 * config/arm/cortex-a53.md: Add new bypasses.
29274 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29276 * tree.c (next_type_uid): Change type to unsigned.
29277 (type_hash_canon): Decrement back next_type_uid if
29278 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
29279 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
29282 2017-05-04 Martin Sebor <msebor@redhat.com>
29284 * builtins.c: Fix a trivial typo in a comment.
29286 PR middle-end/79234
29287 * builtins.c (check_sizes): Adjust to handle reading past the end.
29288 Avoid printing excessive upper bound of ranges. Use %E to print
29289 tree nodes instead of converting them to %wu.
29290 (expand_builtin_memchr): New function.
29291 (compute_dest_size): Rename...
29292 (compute_objsize): ...to this.
29293 (expand_builtin_memcpy): Adjust.
29294 (expand_builtin_mempcpy): Adjust.
29295 (expand_builtin_strcat): Adjust.
29296 (expand_builtin_strcpy): Adjust.
29297 (check_strncat_sizes): Adjust.
29298 (expand_builtin_strncat): Adjust.
29299 (expand_builtin_strncpy): Adjust and simplify.
29300 (expand_builtin_memset): Adjust.
29301 (expand_builtin_bzero): Adjust.
29302 (expand_builtin_memcmp): Adjust.
29303 (expand_builtin): Handle memcmp.
29304 (maybe_emit_chk_warning): Check strncat just once.
29306 2017-05-04 Martin Sebor <msebor@redhat.com>
29308 PR preprocessor/79214
29309 PR middle-end/79222
29310 PR middle-end/79223
29311 * builtins.c (check_sizes): Add inlining context and issue
29312 warnings even when -Wno-system-headers is set.
29313 (check_strncat_sizes): Same.
29314 (expand_builtin_strncat): Same.
29315 (expand_builtin_memmove): New function.
29316 (expand_builtin_stpncpy): Same.
29317 (expand_builtin): Handle memmove and stpncpy.
29319 2017-05-04 Bin Cheng <bin.cheng@arm.com>
29321 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
29322 which is not used any more.
29324 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29326 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
29328 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29330 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
29331 (cortexa53_tunings): Likewise.
29332 (cortexa57_tunings): Likewise.
29333 (cortexa72_tunings): Likewise.
29334 (cortexa73_tunings): Likewise.
29336 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29338 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
29339 Set loop alignment to 8.
29341 2017-05-04 Martin Sebor <msebor@redhat.com>
29343 PR translation/80280
29344 * builtins.c (expand_builtin_object_size): Add missing quoting to
29345 %D and like directives.
29346 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
29347 (hsa_type_for_tree_type): Same.
29348 (verify_function_arguments): Same.
29349 * symtab.c (symbol_table::change_decl_assembler_name): Same.
29350 * varasm.c (get_section): Same.
29353 2017-05-04 Martin Sebor <msebor@redhat.com>
29355 PR translation/80280
29356 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
29358 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
29360 * config/aarch64/aarch64.c (generic_addrcost_table):
29361 Change HI/TI mode setting.
29363 2017-05-04 Martin Jambor <mjambor@suse.cz>
29365 PR tree-optimization/80622
29366 * tree-sra.c (comes_initialized_p): New function.
29367 (build_accesses_from_assign): Only set write lazily when
29368 comes_initialized_p is false.
29369 (analyze_access_subtree): Use comes_initialized_p.
29370 (propagate_subaccesses_across_link): Assert !comes_initialized_p
29371 instead of testing for PARM_DECL.
29373 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29375 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
29376 constraint on operand 0 to allow more general addressing modes.
29377 Adjust output template.
29378 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
29380 * config/aarch64/aarch64-protos.h
29381 (aarch64_address_valid_for_prefetch_p): Declare prototype.
29382 * config/aarch64/constraints.md (Dp): New address constraint.
29383 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
29386 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
29388 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
29389 update use of estimate_ipcp_clone_size_and_time.
29390 (estimate_local_effects): Update use of
29391 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
29392 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
29393 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
29394 Return nonspecialized time.
29396 2017-05-04 Richard Biener <rguenther@suse.de>
29398 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
29399 for the last VUSE which def dominates the PHI. Directly call
29401 (get_continuation_for_phi_1): Remove.
29403 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
29405 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
29406 to explain the use of truncating division. Cap the number of
29407 iterations to the maximum given by nb_iterations_upper_bound,
29410 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
29412 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
29413 * configure: Regenerate.
29414 * config.in: Regenerate.
29415 * config/i386/driver-mingw32.c: new file.
29416 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
29417 * config.host: Link driver-mingw32.o on MinGW host.
29418 * doc/install.texi: Document new --enable-mingw-wildcard configure
29421 2017-05-04 Marek Polacek <polacek@redhat.com>
29423 PR tree-optimization/80612
29424 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
29426 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29427 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
29429 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
29430 (movt splitter): Likewise.
29431 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
29432 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
29433 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
29434 block for Thumb-1 with MOVT.
29435 (thumb2_legitimate_address_p): Move code block ...
29436 (can_avoid_literal_pool_for_label_p): ... into this new function.
29437 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
29439 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
29440 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
29441 "M-profile targets with the MOVT instruction".
29443 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29445 * config/arm/arm-builtins.c (arm_init_builtins): Rename
29446 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
29447 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
29449 2017-05-04 Martin Liska <mliska@suse.cz>
29451 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
29452 variable cond_code.
29454 2017-05-04 Richard Biener <rguenther@suse.de>
29456 * tree.c (array_at_struct_end_p): Handle arrays at struct
29457 end with flexarrays more conservatively. Refactor and treat
29458 arrays of arrays or aggregates more strict. Fix
29459 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
29460 * tree.c (array_at_struct_end_p): Adjust prototype.
29461 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
29462 * gimple-fold.c (get_range_strlen): Likewise.
29463 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
29465 2017-05-04 Richard Biener <rguenther@suse.de>
29467 PR tree-optimization/31130
29468 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
29470 (supports_overflow_infinity): Likewise.
29471 (is_negative_overflow_infinity): Likewise.
29472 (is_positive_overflow_infinity): Likewise.
29473 (is_overflow_infinity): Likewise.
29474 (stmt_overflow_infinity): Likewise.
29475 (overflow_infinity_range_p): Likewise.
29476 (usable_range_p): Remove as always returning true.
29477 (make_overflow_infinity): Remove.
29478 (negative_overflow_infinity): Likewise.
29479 (positive_overflow_infinity): Likewise.
29480 (avoid_overflow_infinity): Likewise.
29481 (set_value_range): Adjust accordingly.
29482 (set_value_range_to_nonnegative): Likewise, remove now unused
29483 overflow_infinity arg.
29484 (vrp_operand_equal_p): Adjust.
29485 (update_value_range): Likewise.
29486 (range_int_cst_singleton_p): Likewise.
29487 (operand_less_p): Likewise.
29488 (compare_values_warnv): Likewise.
29489 (extract_range_for_var_from_comparison_expr): Likewise.
29490 (vrp_int_const_binop): Likewise.
29491 (zero_nonzero_bits_from_vr): Likewise.
29492 (extract_range_from_multiplicative_op_1): Likewise.
29493 (extract_range_from_binary_expr_1): Likewise.
29494 (extract_range_from_unary_expr): Likewise.
29495 (extract_range_from_comparison): Likewise.
29496 (extract_range_basic): Likewise.
29497 (adjust_range_with_scev): Likewise.
29498 (compare_ranges): Likewise.
29499 (compare_range_with_value): Likewise.
29500 (dump_value_range): Likewise.
29501 (test_for_singularity): Likewise, remove strict_overflow_p parameter
29503 (simplify_cond_using_ranges): Adjust.
29505 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29507 * brig-builtins.def: Added a builtin for class_f64.
29508 * builtin-types.def: Added a builtin type needed by class_f64.
29510 2017-05-03 Jason Merrill <jason@redhat.com>
29512 * timevar.def: Add TV_CONSTEXPR.
29514 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29516 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
29518 2017-05-03 Martin Jambor <mjambor@suse.cz>
29520 * ipa-prop.c (ipa_update_after_lto_read): Removed.
29521 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
29522 * ipa-cp.c (ipcp_propagate_stage): Do not call
29523 ipa_update_after_lto_read.
29524 * ipa-inline.c (ipa_inline): Likewise.
29526 2017-05-03 Martin Jambor <mjambor@suse.cz>
29528 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
29529 tag. Added a default constructor and a destructor.
29530 (ipa_edge_args_sum_t): New class;
29531 (ipa_edge_args_sum): Declare.
29532 (ipa_edge_args_vector): Remove declaration.
29533 (IPA_EDGE_REF): Use ipa_edge_args_sum.
29534 (ipa_free_edge_args_substructures): Remove declaration.
29535 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
29536 (ipa_edge_args_info_available_for_edge_p): Likewise.
29537 * ipa-prop.c (ipa_edge_args_vector): Removed.
29538 (edge_removal_hook_holder): Likewise.
29539 (edge_duplication_hook_holder): Likewise.
29540 (ipa_edge_args_sum): New variable.
29541 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
29542 ipa_edge_args_vector.
29543 (ipa_free_edge_args_substructures): Likewise.
29544 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
29545 ipa_edge_args_vector.
29546 (ipa_edge_removal_hook): Turned into method
29547 ipa_edge_args_sum_t::remove.
29548 (ipa_edge_duplication_hook): Turned into method
29549 ipa_edge_args_sum_t::duplicate.
29550 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
29551 registering edge hooks.
29552 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
29553 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
29554 ipa_edge_args_sum instead of ipa_edge_args_vector.
29555 * ipa-profile.c (ipa_profile): Likewise.
29557 2017-05-03 Martin Jambor <mjambor@suse.cz>
29559 * symbol-summary.h (function_summary): New method exists.
29560 (function_summary::symtab_removal): Deallocate through release.
29561 (call_summary): New class.
29562 (gt_ggc_mx): New overload.
29563 (gt_pch_nx): Likewise.
29564 (gt_pch_nx): Likewise.
29566 2017-05-03 Jeff Law <law@redhat.com>
29568 PR tree-optimization/78496
29569 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
29570 from simplify_cond_using_ranges. Split off code to walk
29571 backwards through casts into ...
29572 (simplify_cond_using_ranges_2): New function.
29573 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
29574 (execute_vrp): After identifying jump threads, call
29575 simplify_cond_using_ranges_2.
29577 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
29580 * ipa-inline.h (inline_summary): Add ctor.
29581 (create_ggc): Do not use ggc_cleared_alloc.
29583 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
29584 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29586 * gcc.c (handle_braces): Support escaping in switch matching
29588 * doc/invoke.texi (Spec Files): Document it.
29589 Remove superfluous @code markup in items.
29591 2017-05-03 David Malcolm <dmalcolm@redhat.com>
29593 * diagnostic-show-locus.c (struct column_range): New struct.
29594 (get_affected_columns): New function.
29595 (get_printed_columns): New function.
29596 (struct correction): New struct.
29597 (correction::ensure_capacity): New function.
29598 (correction::ensure_terminated): New function.
29599 (struct line_corrections): New struct.
29600 (line_corrections::~line_corrections): New dtor.
29601 (line_corrections::add_hint): New function.
29602 (layout::print_trailing_fixits): Reimplement in terms of the new
29604 (selftest::test_overlapped_fixit_printing): New function.
29605 (selftest::diagnostic_show_locus_c_tests): Call it.
29607 2017-05-03 Nathan Sidwell <nathan@acm.org>
29609 Canonicalize canonical type hashing
29610 * tree.h (type_hash_canon_hash): Declare.
29611 * tree.c (type_hash_list, attribute_hash_list): Move into
29612 type_hash_canon_hash.
29613 (build_type_attribute_qual_variant): Break out hash code calc into
29614 type_hash_canon_hash.
29615 (type_hash_canon_hash): New. Generic type hash computation.
29616 (build_range_type_1, build_array_type_1, build_function_type,
29617 build_method_type_directly, build_offset_type, build_complex_type,
29618 make_vector_type): Call it.
29620 2017-05-03 Richard Biener <rguenther@suse.de>
29622 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
29623 When all DRs have unknown misaligned do not always peel
29624 when there is a store but apply the same costing model as if
29625 there were only loads.
29627 2017-05-03 Richard Biener <rguenther@suse.de>
29630 PR tree-optimization/80492
29631 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29632 compare_base_decls returning dont-know properly.
29634 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
29636 * config/arm/iterators.md (CCSI): New mode iterator.
29637 (arch): New mode attribute.
29638 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
29639 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
29640 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
29641 code iterator for success result mode.
29642 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
29643 the corresponding new insn generators.
29645 2017-05-03 Bin Cheng <bin.cheng@arm.com>
29648 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29649 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29651 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
29653 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
29654 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
29655 (DDR_A): Wrap DDR argument in brackets.
29656 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
29657 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
29658 (DDR_REVERSED_P): Likewise.
29660 2017-05-03 Jakub Jelinek <jakub@redhat.com>
29662 PR tree-optimization/79472
29663 * tree-switch-conversion.c (struct switch_conv_info): Add
29664 contiguous_range and default_case_nonstandard fields.
29665 (collect_switch_conv_info): Compute contiguous_range and
29666 default_case_nonstandard fields, don't clear final_bb if
29667 contiguous_range and only the default case doesn't have the required
29669 (check_all_empty_except_final): Set default_case_nonstandard instead
29670 of failing if contiguous_range and the default case doesn't have empty
29672 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
29673 and only the default case doesn't have the required constants. Skip
29675 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
29676 if default_case_nonstandard.
29677 (build_constructors): Build constant 1 just once. Assert that default
29678 values aren't inserted in between cases if contiguous_range. Skip
29680 (build_arrays): Skip virtual phis.
29681 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
29682 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
29683 Handle virtual phis.
29684 (gen_inbound_check): Handle default_case_nonstandard case.
29685 (process_switch): Adjust check_final_bb caller. Call
29686 gather_default_values with the first non-default case instead of
29687 default case if default_case_nonstandard.
29689 2017-05-02 Nathan Sidwell <nathan@acm.org>
29691 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
29692 check. Fix formatting.
29694 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
29696 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
29697 errors when comparing specialized and unspecialized times.
29699 2017-05-02 David Malcolm <dmalcolm@redhat.com>
29701 * diagnostic-show-locus.c
29702 (layout::should_print_annotation_line_p): Make private.
29703 (layout::print_annotation_line): Make private.
29704 (layout::annotation_line_showed_range_p): Make private.
29705 (layout::show_ruler): Make private.
29706 (layout::print_source_line): Make private. Pass in line and
29707 line_width, rather than calling location_get_source_line. Drop
29709 (layout::print_leading_fixits): New method.
29710 (layout::print_any_fixits): Rename to...
29711 (layout::print_trailing_fixits): ...this, and make private.
29712 Don't print newline fixits.
29713 (diagnostic_show_locus): Move logic for printing one row into...
29714 (layout::print_line): ...this new function. Move the
29715 location_get_source_line call and error-handling from
29716 print_source_line to here. Call print_leading_fixits, and rename
29717 print_any_fixits to print_trailing_fixits.
29718 (selftest::test_fixit_insert_containing_newline): Update now that
29719 newlines are partially supported.
29720 (selftest::test_fixit_insert_containing_newline_2): New test.
29721 (selftest::test_fixit_replace_containing_newline): Update comments.
29722 (selftest::diagnostic_show_locus_c_tests): Call the new test.
29723 * edit-context.c (class added_line): New class.
29724 (class edited_line): Describe newline handling in comment.
29725 (edited_line::actually_edited_p): New method.
29726 (edited_line::print_content): Delete redundant decl.
29727 (edited_line::m_predecessors): New field.
29728 (edited_file::print_content): Call edited_line::print_content.
29729 (edited_file::print_diff): Update to support newlines.
29730 (edited_file::print_diff_hunk): Likewise.
29731 (edited_file::print_run_of_changed_lines): New function.
29732 (edited_file::print_diff_line): Convert to...
29733 (print_diff_line): ...this.
29734 (edited_file::get_effective_line_count): New function.
29735 (edited_line::edited_line): Initialize new field m_predecessors.
29736 (edited_line::~edited_line): Clean up m_predecessors.
29737 (edited_line::apply_fixit): Handle newlines.
29738 (edited_line::get_effective_line_count): New function.
29739 (edited_line::print_content): New function.
29740 (edited_line::print_diff_lines): New function.
29741 (selftest::test_applying_fixits_insert_containing_newline): New
29743 (selftest::test_applying_fixits_replace_containing_newline): New
29745 (selftest::insert_line): New function.
29746 (selftest::test_applying_fixits_multiple_lines): Add example of
29748 (selftest::edit_context_c_tests): Call the new tests.
29750 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29752 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
29753 parameter cand. Update dump information.
29754 (get_computation_cost): Update uses.
29756 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29758 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
29759 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
29760 (get_computation_at, rewrite_use_address): Update use of
29761 get_computation_aff.
29763 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29765 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
29766 (get_computation): Delete.
29767 (get_computation_cost): Implement like get_computation_cost_at.
29768 Use get_computation_at.
29769 (get_computation_cost_at): Delete.
29770 (rewrite_use_nonlinear_expr): Use get_computation_at.
29771 (rewrite_use_compare, remove_unused_ivs): Ditto.
29773 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29775 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
29777 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29779 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
29780 (ivopts_global_cost_for_size): Rename parameter and update uses.
29781 (iv_ca_recount_cost): Update uses.
29782 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
29783 candidates seperately in n_invs and n_cands.
29784 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
29786 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29788 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
29789 (find_inv_vars_cb): New.
29790 (find_depends): Renamed to ...
29791 (find_inv_vars): ... this.
29792 (add_candidate_1, force_var_cost): Call find_inv_vars.
29793 (split_address_cost, determine_group_iv_cost_cond): Ditto.
29795 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29797 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
29798 inv_vars. Add inv_exprs.
29799 (struct iv_cand): Rename depends_on to inv_vars.
29800 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
29801 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
29802 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
29803 (dump_cand): Dump inv_vars.
29804 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
29805 (record_invariant, find_depends, add_candidate_1): Ditto.
29806 (set_group_iv_cost, force_var_cost): Ditto.
29807 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
29808 (get_computation_cost_at, get_computation_cost): Ditto.
29809 (determine_group_iv_cost_generic): Ditto.
29810 (determine_group_iv_cost_address): Ditto.
29811 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
29812 (determine_group_iv_costs): Ditto.
29813 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
29814 (iv_ca_set_remove_invariants): Renamed to ...
29815 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
29816 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
29817 (iv_ca_set_add_invariants): Renamed to ...
29818 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
29819 (iv_ca_set_cp): Use iv_ca_set_add_invs.
29820 (iv_ca_has_deps): Support inv_vars and inv_exprs.
29821 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
29822 (create_new_ivs): Remove useless dump.
29824 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29826 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
29828 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
29829 (iv_ca_set_no_cp, create_new_iv): Ditto.
29831 2017-05-02 Bin Cheng <bin.cheng@arm.com>
29833 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29835 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29837 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
29838 function tree_check2.
29840 2017-05-02 Martin Liska <mliska@suse.cz>
29842 * doc/gcov.texi: Add missing preposition.
29843 * gcov.c (function_info::function_info): Properly fill up
29844 all member variables.
29846 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29848 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
29850 2017-05-02 Tamar Christina <tamar.christina@arm.com>
29852 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
29854 2017-05-02 Martin Liska <mliska@suse.cz>
29857 * lto-streamer-in.c (lto_read_tree_1): Remove
29858 LTO_STREAMER_DEBUG.
29859 * lto-streamer.c (struct tree_hash_entry): Likewise.
29860 (struct tree_entry_hasher): Likewise.
29861 (tree_entry_hasher::hash): Likewise.
29862 (tree_entry_hasher::equal): Likewise.
29863 (lto_streamer_init): Likewise.
29864 (lto_orig_address_map): Likewise.
29865 (lto_orig_address_get): Likewise.
29866 (lto_orig_address_remove): Likewise.
29867 * lto-streamer.h: Likewise.
29868 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
29869 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
29871 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
29873 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
29874 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
29875 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
29876 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
29877 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
29878 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
29879 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
29880 (mm_maskz_sub_ss): New intrinsics.
29881 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29882 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29883 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
29884 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
29885 (__builtin_ia32_subss_mask_round): New builtins.
29886 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29887 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29888 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29890 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29891 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29893 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29896 2017-05-02 Martin Jambor <mjambor@suse.cz>
29898 PR tree-optimization/78687
29899 * tree-sra.c (access): New field parent.
29900 (process_subtree_disqualification): New function.
29901 (disqualify_candidate): Call it.
29902 (build_accesses_from_assign): Reset write flag if creating an
29904 (build_access_subtree): Fill in parent field and also prpagate
29905 down grp_write flag.
29906 (create_artificial_child_access): New parameter set_grp_write, set
29907 grp_write to its value.
29908 (propagate_subaccesses_across_link): Also propagate grp_write flag
29910 (propagate_all_subaccesses): Push the closest parent back to work
29911 queue if add_access_to_work_queue returned true.
29913 2017-05-02 Richard Biener <rguenther@suse.de>
29915 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29916 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29917 -fstrict-overflow documentation.
29918 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29919 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29920 flag_strict_overflow.
29921 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29922 * lto-opts.c (lto_write_options): Do not stream it.
29923 * lto-wrapper.c (merge_and_complain): Do not handle it.
29924 * opts.c (default_options_table): Do not set -fstrict-overflow.
29925 (finish_options): Likewise do not clear it when sanitizing.
29926 * simplify-rtx.c (simplify_const_relational_operation): Do not
29927 test flag_strict_overflow.
29929 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29931 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29932 using enabled attribute.
29933 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29934 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29935 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29936 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29937 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29938 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29939 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29940 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29941 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29942 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29944 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29946 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29948 2017-05-02 Richard Biener <rguenther@suse.de>
29950 PR tree-optimization/80591
29952 2017-04-10 Richard Biener <rguenther@suse.de>
29954 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29957 2017-05-02 Richard Biener <rguenther@suse.de>
29959 PR tree-optimization/80549
29960 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29961 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29962 headers if they do not have a preheader.
29964 2017-05-02 Martin Liska <mliska@suse.cz>
29967 * common.opt: Fix typo.
29968 * doc/invoke.texi: Likewise.
29970 2017-05-01 Jan Beulich <jbeulich@suse.com>
29972 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29973 swapping, add (x,x,m,x,n) alternative.
29975 2017-05-01 Nathan Sidwell <nathan@acm.org>
29977 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29978 unnecessary unadjusted_alignment check.
29980 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29983 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29985 * gimplify.c (gimplify_cilk_detach): New function.
29986 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29987 * tree-core.h: Document EXPR_CILK_SPAWN.
29988 * tree.h (EXPR_CILK_SPAWN): Define.
29990 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29992 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29993 to use new fixit_hint representation, using the "replace" logic.
29994 (get_line_span_for_fixit_hint): Likewise.
29995 (layout::print_any_fixits): Likewise.
29996 (selftest::test_one_liner_many_fixits): Rename to...
29997 (selftest::test_one_liner_many_fixits_1): ...this, and update
29998 comment and expected output to reflect that the multiple fix-it
29999 hints are now consolidated into one insertion.
30000 (selftest::test_one_liner_many_fixits_2): New test.
30001 (selftest::test_diagnostic_show_locus_one_liner): Update for
30003 (selftest::test_fixit_consolidation): Update for fix-it API
30005 * diagnostic.c (print_parseable_fixits): Likewise.
30006 * edit-context.c (edited_line::m_line_events): Convert from
30007 auto_vec <line_event *> to auto_vec <line_event>.
30008 (class line_event): Convert from abstract base class to a concrete
30009 class, taking over the role of replace_event.
30010 (class insert_event): Delete.
30011 (class replace_event): Rename to class line_event. Convert to
30013 (edit_context::add_fixits): Reimplement.
30014 (edit_context::apply_insert): Delete.
30015 (edit_context::apply_replace): Rename to...
30016 (edit_context::apply_fixit): ...this. Convert to half-open range.
30017 (edited_file::apply_insert): Delete.
30018 (edited_file::apply_replace): Rename to...
30019 (edited_file::apply_fixit): ...this.
30020 (edited_line::~edited_line): Drop deletion of events.
30021 (edited_line::apply_insert): Delete.
30022 (edited_line::apply_replace): Rename to...
30023 (edited_line::apply_fixit): ...this. Convert to half-open range.
30024 Update for change to type of m_line_events.
30025 * edit-context.h (edit_context::apply_insert): Delete.
30026 (edit_context::apply_replace): Rename to...
30027 (edit_context::apply_fixit): ...this.
30029 2017-05-01 Martin Sebor <msebor@redhat.com>
30031 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30034 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30037 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30038 __get_cpuid_max returns 0.
30039 (__get_cpuid_count): Ditto.
30041 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30043 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30044 replacement expression is another instance of one of its arguments.
30046 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30049 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30050 check for stack push/pop autoinc.
30051 * config/i386/i386.c (ix86_agi_dependent): Return false
30052 if the only reason why modified_in_p returned true is that
30053 addr is SP based and set_insn is a push or pop.
30055 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30057 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30060 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30063 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30064 (account_size_time): Use two predicates - exec_pred and
30066 (evaluate_conditions_for_known_args): Compute both clause and
30068 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30069 (inline_summary_t::duplicate): Update.
30070 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30072 (compute_inline_parameters): Likewise.
30073 (estimate_edge_size_and_time): Update caluclation of time.
30074 (estimate_node_size_and_time): Compute both time and nonspecialized
30076 (estimate_ipcp_clone_size_and_time): Update.
30077 (inline_merge_summary): Update.
30078 (do_estimate_edge_time): Update.
30079 (do_estimate_edge_size): Update.
30080 (do_estimate_edge_hints): Update.
30081 (inline_read_section, inline_write_summary): Stream both new predicates.
30082 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30084 (compute_inlined_call_time): Cleanup.
30085 (big_speedup_p): Update.
30086 (edge_badness): Update.
30087 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30088 (size_time_entry): Replace predicate by exec_predicate and
30089 nonconst_predicate.
30090 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30091 (estimate_edge_time): Return also nonspec_time.
30092 (reset_edge_growth_cache): Update.
30094 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30096 PR rtl-optimization/80491
30097 * ifcvt.c (noce_process_if_block): When looking for x setter
30098 with missing else_bb, don't check only the insn right before
30099 cond_earliest, but look for the last insn that x is modified in
30100 within the same bb.
30102 PR rtl-optimization/80491
30103 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30105 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30107 PR tree-optimization/80487
30108 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30110 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30112 PR tree-optimization/79697
30113 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30114 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30115 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30117 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30118 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30120 2017-04-28 Martin Sebor <msebor@redhat.com>
30122 PR tree-optimization/80523
30123 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30124 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30126 (maybe_warn, format_directive, parse_directive): Use new functions.
30127 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30129 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30131 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30133 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30135 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30136 target_header_dir): Set correctly.
30137 * configure: Regenerated.
30138 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30139 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30140 instead of SYSTEM_HEADER_DIR.
30142 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30144 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30145 (estimate_local_effects): Likewise.
30146 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30147 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30148 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30149 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30150 do_estimate_edge_time, estimate_edge_time): Likewise.
30151 * ipa-inline-analysis.c (estimate_node_size_and_time,
30152 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30153 (estimate_time_after_inlining): Remove.
30155 2017-04-28 Martin Liska <mliska@suse.cz>
30157 * doc/gcov.texi: Enhance documentation of gcov.
30159 2017-04-28 Martin Liska <mliska@suse.cz>
30161 * doc/gcov.texi: Sort options in alphabetic order.
30162 * doc/gcov-dump.texi: Likewise.
30163 * doc/gcov-tool.texi: Likewise.
30164 * gcov.c (print_usage): Likewise.
30165 * gcov-dump.c (print_usage): Likewise.
30166 * gcov-tool.c (print_merge_usage_message): Likewise.
30167 (print_rewrite_usage_message): Likewise.
30168 (print_overlap_usage_message): Likewise.
30170 2017-04-28 Martin Liska <mliska@suse.cz>
30172 PR gcov-profile/53915
30173 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30175 2017-04-28 Martin Liska <mliska@suse.cz>
30177 PR gcov-profile/79891
30178 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30179 is marked by compiler as living on a line.
30180 (get_cycles_count): Remove usage of the union.
30181 (output_intermediate_file): Likewise.
30182 (find_source): Fix GNU coding style.
30183 (accumulate_line_counts): Remove old non-all block mode.
30184 (output_lines): Remove usage of the union.
30185 * profile.c (output_location): Include all BBs, even if
30186 belonging to a same line (and file) as a previous BB.
30188 2017-04-28 Martin Liska <mliska@suse.cz>
30190 * gcov.c (process_args): Handle new argument 'w'.
30191 (read_graph_file): Assign ID to BBs.
30192 (output_branch_count): Display BB # if verbose flag is set.
30193 (output_lines): Likewise for arcs.
30194 (print_usage): Add '--verbose' option help.
30195 * doc/gcov.texi: Document --verbose (-w) option.
30197 2017-04-28 Martin Liska <mliska@suse.cz>
30199 * gcov.c (struct block_location_info): New struct.
30200 (process_file): Fill up the new structure.
30201 (read_graph_file): Replace usage of encoding by the newly added
30203 (add_line_counts): Likewise.
30204 (accumulate_line_counts): Remove usage of the union.
30205 (function_info::function_info): New function.
30206 (function_info::~function_info): Likewise.
30207 (process_file): Call delete instead of release_function.
30208 (release_function): Release the function.
30209 (release_structures): Call delete instead of release_function.
30210 (solve_flow_graph): Replace usage of num_blocks.
30211 (find_exception_blocks): Likewise.
30212 (output_lines): Fix GNU coding style.
30214 2017-04-28 Martin Liska <mliska@suse.cz>
30217 * coverage.c (coverage_remove_note_file): New function.
30218 * coverage.h: Declare the function.
30219 * toplev.c (finalize): Clean if an error has been seen.
30221 2017-04-28 Martin Liska <mliska@suse.cz>
30223 PR gcov-profile/80031
30224 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30225 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30226 * gcov.c (read_graph_file): Read just number of blocks.
30227 * profile.c (branch_prob): Do not stream 0 flags per a basic
30230 2017-04-28 Martin Liska <mliska@suse.cz>
30232 * gcov-dump.c (tag_*): Add new argument to declarations.
30233 (dump_gcov_file): Likewise.
30234 (tag_blocks): Add and use new argument depth.
30235 (tag_arcs): Likewise.
30236 (tag_lines): Likewise.
30237 (tag_counters): Likewise.
30238 (tag_summary): Likewise.
30239 (dump_working_sets): Use depth to do a proper indentation.
30241 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30244 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30245 * symtab.c (symtab_node::debug_symtab): Move definition here.
30247 2017-04-28 Richard Biener <rguenther@suse.de>
30249 * lto-streamer.h (LTO_major_version): Bump to 7.
30251 2017-04-28 Richard Biener <rguenther@suse.de>
30253 * tree-vrp.c (assert_info): New struct.
30254 (add_assert_info): New helper.
30255 (register_edge_assert_for_2): Refactor to add asserts to a vector
30257 (register_edge_assert_for_1): Likewise.
30258 (register_edge_assert_for): Likewise.
30259 (finish_register_edge_assert_for): New helper actually registering
30260 asserts where live on edge.
30261 (find_conditional_asserts): Adjust.
30262 (find_switch_asserts): Likewise.
30263 (evrp_dom_walker::try_find_new_range): Generalize.
30264 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
30266 2017-04-27 Marek Polacek <polacek@redhat.com>
30269 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
30270 arg10 and arg11 to itype.
30272 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
30274 * doc/extend.texi (Object Size Checking): Improve grammar.
30276 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
30279 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
30280 that the logic for permitting reciprocal estimates matches that
30283 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30286 * tree.c (type_cache_hasher::equal): Only compare
30287 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
30288 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
30289 non-aggregate element types.
30290 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
30291 about the flag on ARRAY_TYPEs in the comment, formatting fix.
30293 2017-04-27 Richard Biener <rguenther@suse.de>
30295 PR middle-end/80533
30296 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
30297 stripping ARRAY_REFs from MEM_EXPR make sure we're not
30298 keeping a reference to a trailing array.
30300 2017-04-27 Richard Biener <rguenther@suse.de>
30302 PR middle-end/80539
30303 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
30304 being in loop-closed SSA form conservatively.
30305 (chrec_fold_multiply_poly_poly): Likewise.
30307 2017-04-27 Tamar Christina <tamar.christina@arm.com>
30309 PR middle-end/79665
30310 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
30311 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
30313 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30316 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
30317 (aarch64_function_arg_alignment): Return unsigned int again, but still
30318 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
30319 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30320 Don't emit -Wpsabi note.
30321 (aarch64_function_arg_boundary): Likewise.
30322 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30325 2017-04-26 Nathan Sidwell <nathan@acm.org>
30327 * tree.h (crc32_unsigned_n): Declare.
30328 (crc32_unsigned, crc32_unsigned): Make inline.
30329 * tree.c (crc32_unsigned_bits): Replace with ...
30330 (crc32_unsigned_n): ... this.
30331 (crc32_unsigned, crc32_byte): Remove.
30332 (crc32_string): Remove unnecessary braces.
30334 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30336 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
30337 * ipa-inline-analysis.c (MAX_TIME): Remove.
30338 (account_size_time): Use sreal for time.
30339 (dump_inline_summary): Update.
30340 (estimate_function_body_sizes): Update.
30341 (estimate_edge_size_and_time): Update.
30342 (estimate_calls_size_and_time): Update.
30343 (estimate_node_size_and_time): Update.
30344 (inline_merge_summary): Update.
30345 (inline_update_overall_summary): Update.
30346 (estimate_time_after_inlining): Update.
30347 (inline_read_section): Update.
30348 (inline_write_summary): Update.
30349 * ipa-inline.c (compute_uninlined_call_time): Update.
30350 (compute_inlined_call_time): Update.
30351 (recursive_inlining): Update.
30352 (inline_small_functions): Update.
30353 (dump_overall_stats): Update.
30354 * ipa-inline.h: Include sreal.h.
30355 (size_time_entry): Turn time to sreal.
30356 (inline_summary): Turn self_time nad time to sreal.
30358 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
30360 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
30362 (sreal::stream_out, sreal::stream_in): New.
30363 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
30365 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30367 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
30370 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
30373 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
30375 (dimode_scalar_chain::compute_convert_gain): Ditto.
30376 (dimode_scalar_chain::make_vector_copies): Ditto.
30377 (dimode_scalar_chain::convert_reg): Ditto.
30378 (dimode_scalar_chain::convert_insn): Ditto.
30379 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
30380 (VI248_AVX512BW_1): New mode iterator.
30381 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
30382 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
30385 2017-04-25 Martin Sebor <msebor@redhat.com>
30387 PR tree-optimization/80497
30388 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
30389 constants are representable in HOST_WIDE_INT.
30390 (parse_directive): Ditto.
30392 2017-04-25 Martin Sebor <msebor@redhat.com>
30395 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
30396 (new_zero_array): Adjust signature.
30397 (dom_info::dom_init): Used unsigned rather that size_t.
30398 (dom_info::dom_info): Same.
30400 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30401 Jakub Jelinek <jakub@redhat.com>
30404 * config/arm/arm.c: Include gimple.h.
30405 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30406 returns negative, increment ncrn only if it returned positive.
30407 (arm_needs_doubleword_align): Return int instead of bool,
30408 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
30409 members, but if there is any such non-FIELD_DECL
30410 > PARM_BOUNDARY aligned decl, return -1 instead of false.
30411 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
30412 returns negative, increment nregs only if it returned positive.
30413 (arm_setup_incoming_varargs): Likewise.
30414 (arm_function_arg_boundary): Emit -Wpsabi note if
30415 arm_needs_doubleword_align returns negative, return
30416 DOUBLEWORD_ALIGNMENT only if it returned positive.
30418 2017-04-25 Marek Polacek <polacek@redhat.com>
30421 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30422 first argument to type.
30424 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
30427 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
30428 type checks to test for compatibility instead of equality.
30430 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
30431 Jakub Jelinek <jakub@redhat.com>
30434 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
30436 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
30437 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
30438 the alignment computation, but return their maximum in warn_alignment.
30439 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
30440 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
30442 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
30443 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
30446 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30448 * config/arc/simdext.md (dmpyh): Fix typo.
30450 2017-04-25 Richard Biener <rguenther@suse.de>
30452 PR tree-optimization/80492
30453 * alias.c (compare_base_decls): Handle registers with asm
30454 specification conservatively.
30455 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30456 compare_base_decls returning dont-know properly.
30458 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30460 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
30461 (legitimate_offset_address_p): New function.
30462 (arc_legitimate_address_p): Use above function.
30464 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30466 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
30468 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30470 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
30471 ACCH registers whenever they are available.
30473 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30475 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
30476 double regs fix when not used.
30478 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30480 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
30482 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
30483 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
30485 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30487 * config/arc/arc.c (arc_output_addsi): Check for h-register class
30488 when emitting short ADD instructions.
30490 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
30492 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
30494 (cmpsi_cc_c_insn): Likewise.
30495 (cbranchsi4_scratch): Compute proper instruction length using
30496 compact_hreg_operand.
30497 * config/arc/predicates.md (compact_hreg_operand): New predicate.
30499 2017-04-25 Richard Biener <rguenther@suse.de>
30501 PR middle-end/80509
30502 * passes.c (pass_manager::pass_manager): Initialize
30503 m_name_to_pass_map.
30505 2017-04-25 Richard Biener <rguenther@suse.de>
30507 PR tree-optimization/79201
30508 * tree-ssa-sink.c (statement_sink_location): Handle calls.
30510 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30513 * config/s390/vector.md: Split MEM->GPR vector moves for
30514 non-s_operand addresses.
30516 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30519 * config/s390/predicates.md (reload_const_wide_int_operand): New
30521 * config/s390/s390.md ("movti"): Remove d/P alternative.
30522 ("movti_bigconst"): New pattern definition.
30524 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30527 * s390-protos.h (s390_expand_cs_hqi): Removed.
30528 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
30529 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
30530 modes as well as CCZ1mode and CCZmode.
30531 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
30532 signature of s390_emit_compare_and_swap.
30533 (s390_expand_cs_hqi): Likewise, make static.
30534 (s390_expand_cs_tdsi): Generate an explicit compare before trying
30535 compare-and-swap, in some cases.
30536 (s390_expand_cs): Wrapper function.
30537 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
30539 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
30540 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
30541 patterns for small and large integers. Forbid symref memory operands.
30542 Move expander to s390.c. Require cc register.
30543 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
30544 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
30545 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
30546 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
30547 symref memory operands. Remove CC mode and call s390_match_ccmode
30549 ("atomic_exchange<mode>"): Allow and implement all integer modes.
30551 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30553 * config/s390/s390.md (define_peephole2): New peephole to help
30554 combining the load-and-test pattern with volatile memory.
30556 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
30558 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
30559 with CCZmode for TARGET_Z196.
30561 2017-04-25 Jakub Jelinek <jakub@redhat.com>
30563 PR rtl-optimization/80501
30564 * combine.c (make_compound_operation_int): Set subreg_code to SET
30565 even for AND with mask of the sign bit of mode.
30567 PR rtl-optimization/80500
30568 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
30569 sum's initial value.
30571 2017-04-25 Julian Brown <julian@codesourcery.com>
30572 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30574 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
30576 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
30578 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
30580 2017-04-25 Julian Brown <julian@codesourcery.com>
30581 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30583 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
30584 (thunderx2t99_sha): New Reservation.
30586 2017-04-25 Julian Brown <julian@codesourcery.com>
30587 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30589 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
30590 type for 1-element load.
30592 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
30594 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
30596 2017-04-24 Martin Jambor <mjambor@suse.cz>
30598 PR tree-optimization/80293
30599 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
30600 char arrays not totally scalarizable if it is false.
30601 (analyze_all_variable_accesses): Pass correct value in the new
30602 parameter. Add a statistics counter.
30604 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
30606 PR middle-end/79931
30607 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
30609 2017-04-24 Richard Biener <rguenther@suse.de>
30611 PR tree-optimization/80494
30612 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
30613 out for complex types.
30615 2017-04-24 Richard Biener <rguenther@suse.de>
30617 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
30618 * tree-ssa-sccvn.c (print_scc): Print SCC size.
30619 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
30620 (DFS): Adjust and never fail.
30621 (sccvn_dom_walker::fail): Remove.
30622 (sccvn_dom_walker::before_dom_children): Adjust.
30623 (run_scc_vn): Likewise and never fail.
30624 * tree-ssa-pre.c (pass_pre::execute): Adjust.
30625 (pass_fre::execute): Likewise.
30627 2017-04-24 Richard Biener <rguenther@suse.de>
30629 PR tree-optimization/79725
30630 * tree-ssa-sink.c (statement_sink_location): Return whether
30631 failure reason was zero uses. Move that check later.
30632 (sink_code_in_bb): Deal with zero uses by removing the stmt
30635 2017-04-24 Richard Biener <rguenther@suse.de>
30638 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
30639 pointer-based references.
30641 2017-04-24 Richard Biener <rguenther@suse.de>
30644 * pass_manager.h (pass_manager::operator new): Remove.
30645 (pass_manager::operator delete): Likewise.
30646 * passes.c (pass_manager::operator new): Remove.
30647 (pass_manager::operator delete): Likewise.
30648 (pass_manager::pass_manager): Zero individual pass members.
30650 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
30653 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
30654 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
30655 Check "XEXP (src, 1)" operand here.
30656 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
30657 Check "XEXP (src, 1)" operand here.
30658 (dimode_scalar_chain::make_vector_copies): Detect count register
30659 of a shift instruction. Zero extend count register from QImode
30660 to DImode to satisfy vector shift pattern count operand predicate.
30661 Substitute vector shift count operand with a DImode copy.
30662 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
30665 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
30667 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
30668 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30669 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
30670 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
30671 (UNSPEC_NOREX_MEM): Remove definition.
30673 2017-04-21 Richard Biener <rguenther@suse.de>
30675 PR tree-optimization/79547
30676 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
30677 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
30678 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
30679 without any constraints.
30681 2017-04-21 Richard Biener <rguenther@suse.de>
30683 PR tree-optimization/78847
30684 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
30686 2017-04-21 Richard Biener <rguenther@suse.de>
30688 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
30689 (build_distinct_type_copy): Likewise.
30690 (build_variant_type_copy): Likewise.
30691 * tree.c (build_qualified_type): Pass down mem-stat info.
30692 (build_distinct_type_copy): Likewise.
30693 (build_variant_type_copy): Likewise.
30695 2017-04-21 Richard Biener <rguenther@suse.de>
30697 PR tree-optimization/80237
30698 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
30700 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
30701 for a simplified result.
30703 2016-04-21 Richard Biener <rguenther@suse.de>
30705 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
30706 sth as strict as a simple_iv but a chrec without symbols and an
30707 operand defined in the loop we are peeling (and not some subloop).
30708 (propagate_constants_for_unrolling): Propagate all constants.
30710 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30713 * config/i386/i386.c (print_reg): Remove assert for disalowed
30714 regno values, call output_operand_lossage instead.
30716 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
30719 * config/i386/constraints.md (Yc): New register constraint.
30720 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
30721 Use Yc constraint for alternative 2 of operand 0. Remove
30722 preferred_for_speed attribute.
30724 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30726 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
30727 lastprivate clauses in SIMT case.
30729 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
30731 * doc/invoke.texi (-Wextra-semi): Document new warning option.
30733 2017-04-20 Richard Biener <rguenther@suse.de>
30735 PR tree-optimization/57796
30736 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
30737 as N scalar stores.
30738 (vect_model_load_cost): Cost gathers as N scalar loads.
30740 2017-04-20 Richard Biener <rguenther@suse.de>
30742 * ggc-page.c (ggc_allocated_p): Rename to ...
30743 (safe_lookup_page_table_entry): ... this and return the lookup
30745 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
30747 2017-04-20 Richard Biener <rguenther@suse.de>
30749 PR tree-optimization/80453
30750 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
30751 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
30752 from the conditions.
30753 (vn_phi_eq): Pass them down.
30754 (vn_phi_lookup): Record them.
30755 (vn_phi_insert): Likewise.
30757 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
30759 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
30760 uninitialized variable warning to avoid buffer overrun.
30762 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
30765 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
30766 is suppressed for '{ 0 }' in C.
30768 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30770 * BASE-VER: Set to 8.0.0.
30772 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
30774 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
30775 priority .init_array and .fini_array section with SECTION_NOTYPE
30778 2017-04-20 Jakub Jelinek <jakub@redhat.com>
30780 PR middle-end/80423
30781 * tree.h (build_array_type): Add typeless_storage default argument.
30782 * tree.c (type_cache_hasher::equal): Also compare
30783 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
30784 (build_array_type): Add typeless_storage argument, set
30785 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
30787 (build_nonshared_array_type): Adjust build_array_type_1 caller.
30788 (build_array_type): Likewise. Add typeless_storage argument.
30790 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30791 Jakub Jelinek <jakub@redhat.com>
30793 PR tree-optimization/80426
30794 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
30795 operation on symbolic operands, also compute the overflow for the
30796 invariant part when the operation degenerates into a negation.
30798 2017-04-19 Jakub Jelinek <jakub@redhat.com>
30801 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
30802 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
30805 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
30807 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
30810 * config/avr/avr.c (tree.h): Include it.
30811 (cgraph.h): Include it.
30812 (avr_encode_section_info): Don't warn for uninitialized progmem
30813 variable if it's just an alias.
30815 2017-04-19 Richard Biener <rguenther@suse.de>
30818 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
30819 when needed by AutoPGO.
30821 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
30824 * doc/lto.texi: Remove an extra 'that'.
30826 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
30828 PR rtl-optimization/80429
30829 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
30830 are only used in debug insns.
30832 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
30833 Vladimir Makarov <vmakarov@redhat.com>
30835 * config/sparc/predicates.md (input_operand): Add comment. Return
30836 true for any memory operand when LRA is in progress.
30837 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
30839 2017-04-18 Jeff Law <law@redhat.com>
30842 * mips.md ({return,simple_return}_internal): Do not overwrite
30845 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30847 PR tree-optimization/80443
30848 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
30849 instead of adding 1, subtract -1 and similarly instead of subtracting
30852 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
30854 PR rtl-optimization/80357
30855 * haifa-sched.c (tmp_bitmap): New variable.
30856 (model_recompute): Handle duplicate use records.
30857 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
30858 (free_global_sched_pressure_data): Free it.
30860 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
30863 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30864 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30865 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30866 instead of SYSTEM_HEADER_DIR.
30868 2017-04-18 Jeff Law <law@redhat.com>
30870 PR middle-end/80422
30871 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
30872 predecessors after walking up the insn chain.
30874 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30877 * dwarf2out.c (modified_type_die): Try harder not to emit internal
30878 sizetype type into debug info.
30880 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30883 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
30884 unneeded test for TARGET_UPPER_REGS_SF.
30885 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30887 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30890 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30891 instead of gsi_after_labels.
30893 2017-04-18 Jeff Law <law@redhat.com>
30895 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30899 2017-04-13 Jeff Law <law@redhat.com>
30900 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30901 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30903 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30906 * config/avr/avr.c (intl.h): Include it.
30907 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30909 2017-04-18 Martin Liska <mliska@suse.cz>
30911 PR gcov-profile/78783
30912 * gcov-tool.c (gcov_output_files): Validate that destination
30913 file is either removed by the tool or by a user.
30915 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30916 Guy Benyei <guybe@mellanox.com>
30918 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30919 block, and do not negate it, the stored id is already negative.
30921 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30923 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30925 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30928 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30929 masks of options that should be turned off if the VSX vector
30930 options are turned off.
30931 (OTHER_P8_VECTOR_MASKS): Likewise.
30932 (OTHER_VSX_VECTOR_MASKS): Likewise.
30933 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30934 rs6000_disable_incompatible_switches to validate no type switches
30936 (rs6000_incompatible_switch): New function to disallow turning on
30937 other vector options if -mno-vsx, -mno-power8-vector, or
30938 -mno-power9-vector are specified.
30940 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30942 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30944 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30946 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30947 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30948 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30949 (ARG_POINTER_CFA_OFFSET): Likewise.
30951 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30953 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30954 conditions to take advantage of various optimizations.
30956 2017-04-13 Jeff Law <law@redhat.com>
30958 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30959 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30960 (zero_extendsidi2_dext): Likewise.
30962 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30965 * fold-const.c (fold_ternary_loc): Revert
30966 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30969 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30971 PR rtl-optimization/80343
30972 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30973 new scratch pseudo.
30975 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30978 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30979 to ubsan_encode_value.
30981 2017-04-13 Jeff Law <law@redhat.com>
30983 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30984 appearing in DEBUG_INSNs.
30986 2017-04-13 Martin Liska <mliska@suse.cz>
30988 PR gcov-profile/80413
30989 * gcov-io.c (gcov_write_string): Copy to buffer just when
30990 allocated size is greater than zero.
30992 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30995 * dwarf2out.c (decls_for_scope): Ignore declarations of
30996 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30998 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31001 * ipa-visibility.c (non_local_p): Fix typos.
31002 (localize_node): When localizing symbol in same comdat group,
31003 dissolve the group only when we know external symbols are going
31005 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31007 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31009 PR tree-optimization/79390
31010 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31011 order does not result in usable sequence, retry with reversed operand
31017 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31018 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31019 op0 instead of fold_convert_loc (loc, type, arg0).
31021 2017-04-12 Jeff Law <law@redhat.com>
31023 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31024 has a delay slot in the generated code.
31026 * config/cris/cris.md (cris_preferred_reload_class): Return
31027 GENNONACR_REGS rather than GENERAL_REGS.
31029 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31032 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31033 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31034 signedness of the result type.
31036 2017-04-12 Richard Biener <rguenther@suse.de>
31037 Jeff Law <law@redhat.com>
31039 PR tree-optimization/80359
31040 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31041 trim stores to TARGET_MEM_REFs.
31043 2017-04-12 Richard Biener <rguenther@suse.de>
31045 PR tree-optimization/79390
31046 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31047 threading case even more.
31049 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31052 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31053 for quad_address_p for TImode, instead of just not indexed_address.
31055 2017-04-12 Richard Biener <rguenther@suse.de>
31056 Bernd Edlinger <bernd.edlinger@hotmail.de>
31058 PR middle-end/79671
31059 * alias.c (component_uses_parent_alias_set_from): Handle
31060 TYPE_TYPELESS_STORAGE.
31061 (get_alias_set): Likewise.
31062 * tree-core.h (tree_type_common): Add typeless_storage flag.
31063 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31064 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31065 for types containing members with TYPE_TYPELESS_STORAGE.
31066 (place_field): Likewise.
31067 (layout_type): Likewise for ARRAY_TYPE.
31068 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31069 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31070 TYPE_TYPELESS_STORAGE.
31071 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31073 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31076 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31077 first argument to type.
31079 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31083 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31084 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31085 (rs6000_expand_binop_builtin): Likewise.
31086 (rs6000_expand_ternop_builtin): Likewise; also add missing
31087 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31088 vshasigma built-ins.
31089 * doc/extend.texi: Document that vec_xxpermdi's third argument
31090 must be a constant.
31092 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31094 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31095 Use shift_const cost parameter when calculating gain of STV shifts.
31097 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31099 PR rtl-optimization/70478
31100 * lra-constraints.c (process_alt_operands): Check memory for
31101 disfavoring memory insn operand.
31103 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31105 PR middle-end/80100
31106 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31107 left shift in unsigned HOST_WIDE_INT type.
31109 PR rtl-optimization/80385
31110 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31111 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31114 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31115 if they have any depend clauses.
31117 2017-04-11 Martin Liska <mliska@suse.cz>
31120 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31121 * ipa-split.c (split_function): Create a local comdat symbol
31122 if caller is in a comdat group.
31124 2017-04-11 Martin Liska <mliska@suse.cz>
31127 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31130 2017-04-11 Martin Sebor <msebor@redhat.com>
31132 PR middle-end/80364
31133 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31134 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31136 (directive::set_width, directive::set_precision, format_character):
31138 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31141 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31144 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31145 conflict, set target->arch_name instead of target->cpu_name.
31147 2017-04-11 Richard Biener <rguenther@suse.de>
31149 PR tree-optimization/80374
31150 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31151 build_zero_cst, remove fold_convertible_p check again.
31153 2017-04-11 Martin Liska <mliska@suse.cz>
31156 * ubsan.c (instrument_object_size): Do not instrument register
31159 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31162 * config/i386/i386-builtin-types.def
31163 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31164 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31165 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31166 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31167 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31168 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31169 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31170 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31171 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31172 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31173 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31174 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31175 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31176 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31177 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31178 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31179 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31180 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31181 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31182 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31183 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31184 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31185 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31186 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31187 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31188 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31189 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31190 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31191 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31192 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31193 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31194 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31195 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31196 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31197 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31198 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31199 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31200 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31201 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31202 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31203 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31204 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31205 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31206 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31207 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31208 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31210 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31211 flag to second_arg_count, handle 4 argument function type _COUNT
31212 aliases, handle second_arg_count on second argument rather than last.
31214 2017-04-10 Jeff Law <law@redhat.com>
31216 PR tree-optimization/80374
31217 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31218 record anything if we can not convert integer_zero_node to the
31221 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31224 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31225 Enhance special handling given to the TARGET_P9_MINMAX option in
31226 relation to certain other options.
31228 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31230 PR tree-optimization/80153
31231 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31232 remove POINTER_PLUS_EXPR's base part directly, rather than through
31235 2017-04-10 Richard Biener <rguenther@suse.de>
31236 Bin Cheng <bin.cheng@arm.com>
31238 PR tree-optimization/80153
31239 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31240 the first element of pointer type aff_tree. Build result expr in
31242 (add_elt_to_tree): Convert to type unconditionally. Remove other
31243 fold_convert calls.
31244 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31245 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31247 2017-04-10 Richard Biener <rguenther@suse.de>
31249 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31252 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
31254 PR rtl-optimization/70478
31255 * lra-constraints.c (curr_small_class_check): New.
31256 (update_and_check_small_class_inputs): New.
31257 (process_alt_operands): Update curr_small_class_check. Disfavor
31258 alternative insn memory operands. Check available regs for small
31261 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31264 * config/mips/mips.opt (-mvirt): Update description.
31265 * doc/invoke.texi (-mvirt): Likewise.
31267 2017-04-10 Richard Biener <rguenther@suse.de>
31269 PR middle-end/80362
31270 * fold-const.c (fold_binary_loc): Look at unstripped ops when
31271 looking for NEGATE_EXPR in -A / -B to A / B folding.
31273 2017-04-10 Martin Liska <mliska@suse.cz>
31275 PR gcov-profile/80224
31276 * gcov.c (print_usage): Fix usage string.
31277 (get_gcov_intermediate_filename): Remove.
31278 (output_gcov_file): Use both for normal and intermediate format.
31279 (generate_results): Do not initialize special file for
31280 intermediate format.
31282 2017-04-10 Richard Biener <rguenther@suse.de>
31284 PR tree-optimization/80304
31285 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
31288 2017-04-10 Nathan Sidwell <nathan@acm.org>
31291 * config/rs6000/rs6000.c (rs6000_vector_type): New.
31292 (rs6000_init_builtins): Use it.
31294 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31296 * config/arm/arm.md (<mrc>): Add mode to SET source.
31297 (<mrrc>): Likewise.
31299 2017-04-10 Richard Biener <rguenther@suse.de>
31301 PR middle-end/80344
31302 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
31304 2017-04-10 Jakub Jelinek <jakub@redhat.com>
31307 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
31308 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
31309 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
31310 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
31311 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
31312 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
31313 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
31314 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
31315 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
31316 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
31317 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
31318 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
31319 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
31320 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
31321 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
31322 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
31323 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
31324 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
31325 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
31326 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
31327 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
31328 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
31329 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
31331 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
31333 PR rtl-optimization/70478
31334 * lra-constraints.c: Reverse the last patch.
31336 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
31338 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
31339 Add comment for WCHAR_T.
31341 2017-04-08 Martin Liska <mliska@suse.cz>
31344 2017-04-07 Martin Liska <mliska@suse.cz>
31347 * ipa-split.c (split_function): Add function part to a same comdat
31350 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31353 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
31355 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
31357 * rs6000/rs6000.c (vec_load_pendulum): Rename...
31358 (vec_pairing): ...to this.
31359 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
31360 (rs6000_sched_init): Adjust for name change.
31361 (struct rs6000_sched_context): Likewise.
31362 (rs6000_init_sched_context): Likewise.
31363 (rs6000_set_sched_context): Likewise.
31365 2017-04-07 Jakub Jelinek <jakub@redhat.com>
31371 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
31373 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
31374 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
31375 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
31377 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
31379 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
31381 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31383 PR rtl-optimization/70703
31384 * ira-color.c (update_conflict_hard_regno_costs): Use
31385 int64_t instead of HOST_WIDE_INT.
31387 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
31389 PR rtl-optimization/70478
31390 * lra-constraints.c (process_alt_operands): Disfavor alternative
31391 insn memory operands.
31393 2017-04-07 Jeff Law <law@redhat.com>
31395 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
31396 CALL and NOTE_INSN_CALL_ARG_LOCATION.
31398 2017-04-07 Martin Liska <mliska@suse.cz>
31401 * config/aarch64/aarch64.c (aarch64_process_target_attr):
31402 Show error message instead of an ICE.
31404 2017-04-07 Martin Liska <mliska@suse.cz>
31407 * ipa-split.c (split_function): Add function part to a same comdat
31410 2017-04-07 Richard Biener <rguenther@suse.de>
31412 PR middle-end/80341
31413 * tree.c (get_unwidened): Also handle ! for_type case for
31415 * convert.c (do_narrow): Split out from ...
31416 (convert_to_integer_1): ... here. Do not pass final truncation
31417 type to get_unwidened for TRUNC_DIV_EXPR.
31419 2017-04-07 Richard Biener <rguenther@suse.de>
31421 * tree-affine.c (wide_int_ext_for_comb): Take type rather
31423 (aff_combination_const): Adjust.
31424 (aff_combination_scale): Likewise.
31425 (aff_combination_add_elt): Likewise.
31426 (aff_combination_add_cst): Likewise.
31427 (aff_combination_convert): Likewise.
31428 (add_elt_to_tree): Likewise. Remove unused argument.
31429 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
31431 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
31433 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
31435 * config/arm/arm.c (arm_default_short_enums): Use
31436 ARM_DEFAULT_SHORT_ENUMS.
31437 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
31439 2017-04-06 Jakub Jelinek <jakub@redhat.com>
31442 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
31443 members with redundant out-of-class redeclaration.
31445 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31448 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
31449 * config/i386/i386.md (*zero_extendsidi2):
31450 Add (?*x,*x) and (?*v,*v) alternatives.
31452 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
31455 * config/i386/i386.c (ix86_expand_builtin)
31456 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
31457 mode from insn data. Convert operands to insn operand mode.
31458 Copy operands that don't satisfy insn predicate to a register.
31460 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
31462 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
31465 2017-04-06 Richard Biener <rguenther@suse.de>
31467 PR tree-optimization/80334
31468 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
31469 preserve alignment of accesses.
31471 2017-04-06 Richard Biener <rguenther@suse.de>
31473 PR tree-optimization/80262
31474 * tree-sra.c (build_ref_for_offset): Preserve address-space
31476 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
31477 Drop useless address-space information on MEM_REF offsets.
31479 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
31481 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
31483 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31485 PR rtl-optimization/70703
31486 * ira-color.c (update_conflict_hard_regno_costs): Use
31487 HOST_WIDE_INT instead of long.
31489 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
31492 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
31493 not defined for x86_64 target. Add -mmmx target option when __SSE2__
31495 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
31496 for x86_64 target. Handle -m3dnowa option.
31498 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
31500 PR rtl-optimization/70703
31501 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
31502 (update_conflict_hard_regno_costs): Use long instead of unsigned
31503 arithmetic for cost calculation.
31505 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31506 Bernd Edlinger <bernd.edlinger@hotmail.de>
31509 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
31512 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
31515 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
31516 ptr_mode with Pmode throughout.
31517 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
31518 into probe_stack_range and use DImode.
31520 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
31523 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
31524 call_eh_return is true.
31526 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31528 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
31529 Initialize last_match_fntype_index.
31531 2017-04-05 Jakub Jelinek <jakub@redhat.com>
31534 * tree-nvr.c: Include internal-fn.h.
31535 (pass_return_slot::execute): Ignore internal calls without
31538 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31539 Richard Biener <rguenther@suse.de>
31542 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
31543 captures used multiple times, except for the last use.
31544 * generic-match-head.c: Include gimplify.h.
31546 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31548 PR tree-optimization/79390
31549 * target.h (struct noce_if_info): Declare.
31550 * targhooks.h (default_noce_conversion_profitable_p): Declare.
31551 * target.def (noce_conversion_profitable_p): New target hook.
31552 * ifcvt.h (struct noce_if_info): New type, moved from ...
31553 * ifcvt.c (struct noce_if_info): ... here.
31554 (noce_conversion_profitable_p): Renamed to ...
31555 (default_noce_conversion_profitable_p): ... this. No longer
31557 (noce_try_store_flag_constants, noce_try_addcc,
31558 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
31559 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
31560 instead of noce_conversion_profitable_p.
31561 * config/i386/i386.c: Include ifcvt.h.
31562 (ix86_option_override_internal): Don't override
31563 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
31564 (ix86_noce_conversion_profitable_p): New function.
31565 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
31566 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
31567 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
31568 * doc/tm.texi: Regenerated.
31570 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31572 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
31575 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31578 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
31579 instructions for small multiply cores.
31581 2017-04-04 Jeff Law <law@redhat.com>
31583 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
31585 (mips_expand_vec_perm_const): Initialize elements in orig_perm
31586 that are not set by the loop over the elements.
31588 2017-04-04 Jakub Jelinek <jakub@redhat.com>
31591 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
31592 int mode, convert_modes it to mode as unsigned, otherwise use
31593 lowpart_subreg to mode rather than SImode.
31594 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
31595 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
31596 Use DImode instead of SImode for the shift count operand.
31597 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
31600 2017-04-04 Richard Biener <rguenther@suse.de>
31602 PR middle-end/80281
31603 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
31604 arithmetic done for the negate or the plus. Simplify.
31605 (A - (-B) -> A + B): Likewise.
31606 * fold-const.c (split_tree): Make sure to not negate pointers.
31608 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
31610 PR rtl-optimization/60818
31611 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
31612 a compare of comparisons with the thing compared if this results
31613 in a different machine mode.
31615 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
31617 * alias.c (base_alias_check): Fix typo in comment.
31618 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
31619 * cgraphunit.c (symbol_table::compile): Likewise.
31620 * collect2.c (maybe_run_lto_and_relink): Likewise.
31621 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
31622 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
31623 * config/avr/avr.c (avr_map_op_t): Likewise.
31624 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
31625 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
31626 * config/epiphany/epiphany.md (movcc): Likewise.
31627 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
31628 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
31630 * config/mips/mips.c (mips_save_restore_reg): Likewise.
31631 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
31632 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
31633 * config/sh/sh.c (sh_rtx_costs): Likewise.
31634 * fold-const.c (fold_truth_andor): Likewise.
31635 * genautomata.c (collapse_flag): Likewise.
31636 * gengtype.h (struct type::u::s): Likewise.
31637 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
31638 * input.c (FORMAT_AMOUNT): Likewise.
31639 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
31640 (known_aggs_to_agg_replacement_list): Likewise.
31641 * ipa-inline-analysis.c: Likewise.
31642 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
31643 * ipa-polymorphic-call.c
31644 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
31645 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
31646 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
31648 * modulo-sched.c (apply_reg_moves): Likewise.
31649 * omp-expand.c (build_omp_regions_1): Likewise.
31650 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
31651 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
31652 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
31653 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31654 * value-prof.c: Likewise.
31655 * var-tracking.c (val_reset): Likewise.
31657 2017-04-03 Richard Biener <rguenther@suse.de>
31659 PR tree-optimization/80275
31660 * fold-const.c (split_address_to_core_and_offset): Handle
31663 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
31665 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
31666 descriptors is at least equal to that of functions.
31668 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31670 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
31672 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
31675 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
31676 (mov<IMOD4:mode>): New expander.
31677 (*mov<IMOD4:mode>_internal): New insn and split pattern.
31679 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
31681 PR rtl-optimization/79405
31682 * fwprop.c (propagations_left): New variable.
31683 (forward_propagate_into): Decrement it.
31684 (fwprop_init): Initialize it.
31685 (fw_prop): If the variable has reached zero, stop propagating.
31686 (fwprop_addr): Ditto.
31688 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31691 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
31692 a FUNCTION_DECL, pass it as decl instead of origin to
31695 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
31697 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
31700 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
31703 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
31704 TARGET_VSX_SMALL_INTEGER.
31706 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31708 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31709 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
31711 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
31713 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
31714 extraction from odd-numbered MSA register.
31716 2017-03-31 Jakub Jelinek <jakub@redhat.com>
31718 PR middle-end/80173
31719 * expmed.c (store_bit_field_1): Don't attempt to create
31720 a word subreg out of hard registers wider than word if they
31721 have HARD_REGNO_NREGS of 1 for their mode.
31723 PR middle-end/80163
31724 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
31725 conversions to integer types wider than word and pointer.
31728 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
31729 (rtx_equal_for_cselib_p): Pass 0 to it.
31730 * cselib.c (cselib_hasher::equal): Likewise.
31731 (rtx_equal_for_cselib_1): Add depth argument. If depth
31732 is 128, don't look up VALUE locs and punt. Increment
31733 depth in recursive calls when walking VALUE locs.
31735 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
31737 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
31738 (make_gcov_file_name): Use the canonical path name for generating
31740 (read_line): Fix handling of files with ascii null bytes.
31742 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
31744 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
31745 to initialise a vector register instead
31746 of using a const_int.
31748 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31750 PR translation/80189
31751 * gimplify.c (omp_default_clause): Use %qs instead of %s in
31752 diagnostic messages.
31754 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
31757 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
31758 (dfp_diex_<mode>): Update mode of operand 1.
31759 * doc/extend.texi (dxex, dxexq): Document change to return type.
31760 (diex, diexq): Document change to argument type.
31762 2017-03-30 Martin Jambor <mjambor@suse.cz>
31765 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
31766 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
31767 it reflects the signature changes performed at the callee side.
31768 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
31769 to cgraph_build_function_type_skip_args.
31770 (build_function_decl_skip_args): Adjust call to the above function.
31772 2017-03-30 Jakub Jelinek <jakub@redhat.com>
31775 * config/i386/sse.md
31776 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
31777 register as dest whenever it is a MEM not rtx_equal_p to the
31778 corresponding dup operand, and when forcing into reg move the
31779 reg into the memory afterwards.
31780 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
31781 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
31782 for the force_reg mode.
31783 (avx512vl_vextractf128<mode>): Use register as dest either
31784 always when a MEM, or when it is a MEM not rtx_equal_p to the
31785 corresponding dup operand, or even not when it is a CONST_VECTOR
31786 depending on the mode and lo vs. hi.
31787 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
31789 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
31790 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
31791 Likewise. Require that operands[2] is even.
31792 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
31793 Remove extraneous parens. Require that operands[2] is a multiple
31795 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
31796 operands[0] is a MEM if <mask_applied>, the predicates/constraints
31797 disallow memory then.
31799 2017-03-30 Richard Biener <rguenther@suse.de>
31801 PR tree-optimization/77498
31802 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
31803 to non-constants over backedges.
31805 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
31807 PR rtl-optimization/80233
31808 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
31809 as last_combined_insn. Do not test for BARRIER_P separately.
31811 2017-03-29 Andreas Schwab <schwab@suse.de>
31814 * calls.c (prepare_call_address): Convert funexp to Pmode before
31815 copying to temp reg.
31817 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31819 PR tree-optimization/80158
31820 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
31821 Handle possible future case of more than one alternate
31823 (replace_rhs_if_not_dup): Likewise.
31824 (replace_one_candidate): Likewise.
31826 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
31828 PR rtl-optimization/80193
31829 * ira.c (ira): Do not check allocation for LRA.
31831 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
31833 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
31834 (nvptx_output_simt_exit): Declare.
31835 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
31836 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
31837 (init_softstack_frame): Move initialization of crtl->is_leaf to...
31838 (nvptx_declare_function_name): ...here. Emit declaration of local
31839 memory space buffer for omp_simt_enter insn.
31840 (nvptx_output_unisimt_switch): New.
31841 (nvptx_output_softstack_switch): New.
31842 (nvptx_output_simt_enter): New.
31843 (nvptx_output_simt_exit): New.
31844 * config/nvptx/nvptx.h (struct machine_function): New fields
31845 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
31846 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31847 (UNSPECV_SIMT_EXIT): Ditto.
31848 (omp_simt_enter_insn): New insn.
31849 (omp_simt_enter): New expansion.
31850 (omp_simt_exit): New insn.
31851 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
31853 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
31854 (expand_GOMP_SIMT_ENTER_ALLOC): New.
31855 (expand_GOMP_SIMT_EXIT): New.
31856 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
31857 (GOMP_SIMT_ENTER_ALLOC): Ditto.
31858 (GOMP_SIMT_EXIT): Ditto.
31859 * target-insns.def (omp_simt_enter): New insn.
31860 (omp_simt_exit): Ditto.
31861 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
31863 (lower_rec_simd_input_clauses): Implement SIMT privatization.
31864 (lower_rec_input_clauses): Likewise.
31865 (lower_lastprivate_clauses): Handle SIMT privatization.
31867 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
31868 (ompdevlow_adjust_simt_enter): New.
31869 (find_simtpriv_var_op): New.
31870 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
31871 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
31873 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
31874 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
31875 (copy_decl_for_dup_finish): Ditto.
31877 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
31879 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
31882 * config/i386/i386.c (ix86_option_override_internal): Always
31883 allow -mpreferred-stack-boundary=3 for 64-bit targets.
31885 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31887 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31889 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31891 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31892 mark new edge's irreducible flag accordign to it.
31893 (vect_do_peeling): Check loop preheader edge's irreducible flag
31894 and pass it to function slpeel_add_loop_guard.
31896 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31898 PR tree-optimization/80218
31899 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31900 Update block frequencies and counts.
31902 2017-03-28 Richard Biener <rguenther@suse.de>
31904 PR tree-optimization/78644
31905 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31906 of a simplification result we may not use it at all.
31908 2017-03-28 Richard Biener <rguenther@suse.de>
31911 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31912 without arguments, generate default definition of a SSA name.
31914 2017-03-28 Richard Biener <rguenther@suse.de>
31916 PR middle-end/80222
31917 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31918 TYPE_REF_CAN_ALIAS_ALL references.
31919 * fold-const.c (fold_indirect_ref_1): Likewise.
31921 2017-03-28 Martin Liska <mliska@suse.cz>
31924 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31925 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31927 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31928 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31930 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31931 (EXTRA_SPECS): Define.
31932 (SUBTARGET_EXTRA_SPECS): Likewise.
31933 (SUBTARGET_CPP_SPEC): Likewise.
31934 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31935 SUBTARGET_EXTRA_SPECS.
31936 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31938 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31940 * config/arc/simdext.md (vst64_insn): Update pattern.
31941 (vld32wh_insn): Likewise.
31942 (vld32wl_insn): Likewise.
31943 (vld64_insn): Likewise.
31944 (vld32_insn): Likewise.
31946 2017-03-28 Marek Polacek <polacek@redhat.com>
31949 * fold-const.c (fold_comparison): Use protected_set_expr_location
31950 instead of SET_EXPR_LOCATION.
31952 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31954 * tree.c (add_expr): Avoid name lookup warning.
31956 2017-03-27 Jeff Law <law@redhat.com>
31958 PR tree-optimization/80216
31959 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31960 function name. Limit recursion depth.
31961 (record_temporary_equivalences): Corresponding changes.
31963 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31965 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31968 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31971 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31973 * cfgcleanup.c (reg_note_cfa_p): New array.
31974 (insns_have_identical_cfa_notes): New function.
31975 (old_insns_match_p): Don't cross-jump in between /f
31976 and non-/f instructions. If both i1 and i2 are frame related,
31977 verify all CFA notes, their order and content.
31979 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31982 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31983 HImode and SImode with zero extend to DImode to one insn.
31984 (bswap<mode>2_extenddi): Likewise.
31985 (bswapsi2_extenddi): Likewise.
31986 (bswaphi2_extendsi): Likewise.
31987 (bswaphi2): Combine bswap HImode and SImode into one insn.
31988 Separate memory insns from swapping register.
31989 (bswapsi2): Likewise.
31990 (bswap<mode>2): Likewise.
31991 (bswaphi2_internal): Delete, no longer used.
31992 (bswapsi2_internal): Likewise.
31993 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31994 store, and gpr<-gpr swap insns.
31995 (bswap<mode>2_store): Likewise.
31996 (bswaphi2_reg): Register only splitter, combine with the splitter.
31997 (bswaphi2 splitter): Likewise.
31998 (bswapsi2_reg): Likewise.
31999 (bswapsi2 splitter): Likewise.
32000 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32001 the insns into load, store, and register/register insns.
32002 (bswapdi2_ldbrx): Likewise.
32003 (bswapdi2_load): Likewise.
32004 (bswapdi2_store): Likewise.
32005 (bswapdi2_reg): Likewise.
32007 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32009 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32010 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32012 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32015 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32017 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32018 special handling for target option conflicts between dform
32019 options (-mpower9-dform, -mpower9-dform-vector,
32020 -mpower9-dform-scalar) and -mno-direct-move.
32022 2017-03-27 Richard Biener <rguenther@suse.de>
32024 PR tree-optimization/80181
32025 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32027 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32029 * config/arc/predicates.md (move_double_src_operand): Replace the
32030 call to move_double_src_operand with a call to address_operand.
32032 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32034 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32035 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32036 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32038 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32040 * config/arc/predicates.md (long_immediate_loadstore_operand):
32041 Consider scaled addresses cases.
32043 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32045 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32046 restored when in interrupt.
32047 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32048 doesn't have delay slot.
32050 2017-03-27 Richard Biener <rguenther@suse.de>
32053 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32054 inlined thunk clones.
32056 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32059 * asan.c (instrument_derefs): Copy over last operand from
32060 original COMPONENT_REF to the new COMPONENT_REF with
32061 DECL_BIT_FIELD_REPRESENTATIVE.
32062 * ubsan.c (instrument_object_size): Likewise.
32064 2017-03-27 Richard Biener <rguenther@suse.de>
32066 PR tree-optimization/80170
32067 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32068 sure DR/SCEV didnt fold in constants we do not see when looking
32069 at the reference base alignment.
32071 2017-03-27 Richard Biener <rguenther@suse.de>
32073 PR middle-end/80171
32074 * gimple-fold.c (fold_ctor_reference): Properly guard against
32075 NULL return value from canonicalize_constructor_val.
32077 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32080 * config/i386/i386.c (ix86_expand_builtin)
32081 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32082 flags reg setting and flags reg using instructions.
32083 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32084 clobbering instructions to zero extend op2.
32086 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32088 * doc/install.texi (Configuration) <--with-aix-soname>:
32089 Update link to AIX ld.
32091 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32093 PR rtl-optimization/80160
32094 PR rtl-optimization/80159
32095 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32096 reg_alternate_class into account.
32098 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32101 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32102 to consider in curr_insn_transform.
32104 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32106 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32107 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32108 and emit_mode_inner.
32110 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32112 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32113 argument to the overloaded builtin variants. Use the new flag to
32114 deprecate certain builtin variants.
32115 * config/s390/s390-builtin-types.def: Add new builtin types.
32116 * config/s390/s390-builtins.h: Support new flags field for
32117 overloaded builtins.
32118 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32119 (s390_macro_to_expand): Enable vector float data type.
32120 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32121 builtins by incrementing the __VEC__ version number.
32122 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32124 (s390_resolve_overloaded_builtin): Emit error messages depending
32125 on the builtin flags.
32126 * config/s390/s390.c (s390_expand_builtin): Support additional
32127 flags argument. Change error message to match the messages
32128 emitted in s390-c.c.
32129 * config/s390/s390.md: New UNSPEC_* constants.
32130 (op_type): Add new instruction types.
32131 * config/s390/vecintrin.h: Add new builtins and test data class
32133 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32134 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32135 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32136 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32137 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32138 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32140 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32141 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32142 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32143 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32145 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32146 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32147 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32148 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32150 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32151 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32152 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32153 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32154 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32155 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32156 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32158 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32159 ("vec_scatter_element<V_HW_4:mode>_DI")
32160 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32161 ("vec_fpint<mode>", "vflls")
32162 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32163 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32164 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32165 ("*vec_cmphe<mode>_cc"): ... these.
32167 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32168 mode constant instead of magic value.
32170 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32172 * config/s390/s390.c (s390_expand_vec_compare): Support other
32173 vector floating point modes than just V2DF.
32174 (s390_expand_vcond): Likewise.
32175 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32176 (s390_cannot_change_mode_class): Prevent mode changes between TF
32177 and V1TF in vector registers.
32178 * config/s390/s390.md (DF, SF): New mode attributes.
32179 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32180 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32181 SFmode support for VRs.
32182 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32184 (VFT, VF_HW): New mode iterators.
32185 (vw, sdx): New mode attributes.
32186 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32187 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32188 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32189 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32190 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32191 also the new vector floating point modes. Renaming to ...
32193 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32194 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32195 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32196 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32197 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32198 ("vec_unordered<mode>"): ... these.
32200 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32201 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32202 ("*vec_extendv2df"): New insn definitions.
32204 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32206 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32207 ("mulditi3_2", "*muldi3_sign"): New patterns.
32208 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32209 rename the pattern definition.
32211 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32213 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32215 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32217 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32219 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32220 instruction if possible.
32221 * config/s390/vector.md (vec_halfnumelts): New mode
32223 ("*vec_vllezlf<mode>"): New pattern.
32225 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32227 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32228 ("popcountv4si2", "popcountv2di2"): Rename to ...
32229 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32230 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32232 ("popcount<mode>2_vxe"): New pattern.
32234 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32236 * common/config/s390/s390-common.c (processor_flags_table): Add
32238 * config.gcc: Add arch12.
32239 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32240 Default to arch12 for unknown CPU model numbers.
32241 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32242 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32243 PROCESSOR_max sanity check.
32244 * config/s390/s390-opts.h (enum processor_type): Add
32246 * config/s390/s390.c (processor_table): Add arch12.
32247 (s390_expand_builtin): Add check for B_VXE flag.
32248 (s390_issue_rate): Add PROCESSOR_ARCH12.
32249 (s390_get_sched_attrmask): Likewise.
32250 (s390_get_unit_mask): Likewise.
32251 (s390_sched_score): Enable z13 scheduling for arch12.
32252 (s390_sched_reorder): Likewise.
32253 (s390_sched_variable_issue): Likewise.
32254 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
32256 (s390_tune_attr): Use z13 scheduling also for arch12.
32257 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
32258 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
32259 (TARGET_VXE_P): New macros.
32260 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
32261 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
32262 * config/s390/s390.opt: Add arch12 as processor_type.
32264 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32266 * config/s390/s390.md
32267 ("fixuns_truncdddi2", "fixuns_trunctddi2")
32268 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
32269 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
32271 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
32272 Rename expanders to ...
32274 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
32275 ("fixuns_truncdddi2_emu"): ... these.
32277 ("fixuns_trunc<mode>si2_emu"): New expander.
32279 ("*fixuns_truncdfdi2_z13"): Rename to ...
32280 ("*fixuns_truncdfdi2_vx"): ... this.
32282 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32284 * config/s390/2964.md: Remove the single element vector compare
32285 instructions which are no longer used.
32286 * config/s390/s390.c (s390_select_ccmode): Remove handling of
32288 (s390_canonicalize_comparison): Remove handling of DFmode
32290 (s390_expand_vec_compare_scalar): Remove function.
32291 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
32292 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
32294 ("*cmp<mode>_ccs"): Add wfcdb instruction.
32296 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32298 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
32300 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
32301 will anyway by matched by mov<mode>_64dfp.
32303 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32305 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
32306 vlef/vstef. Add missing operand to vleif.
32308 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32310 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
32311 pair for all vector types with 64 bit elements.
32312 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
32313 * config/s390/vector.md (V_HW_64): ... here.
32314 (V_128_NOSINGLE): New mode iterator.
32315 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
32316 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
32317 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
32318 ("*vec_load_pairv2di"): Change to ...
32319 ("*vec_load_pair<mode>"): ... this one.
32321 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32323 * config/s390/constraints.md: Add comments.
32324 (jKK): Reject element sizes > 8 bytes.
32325 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
32327 * config/s390/s390.md: Add the s_operand checks formerly in
32328 s390_split_ok_p to various splitters where they are still
32330 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
32331 for 128 bit vectors. Plus two splitters.
32333 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32335 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
32338 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32341 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
32342 error if the boundary argument is not constant.
32344 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32346 PR rtl-optimization/80112
32347 * loop-doloop.c (doloop_condition_get): Don't check condition
32348 if cmp isn't SET with IF_THEN_ELSE src.
32350 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32352 PR tree-optimization/80158
32353 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
32354 replacing a candidate statement, also replace it for the
32355 candidate's alternate interpretation.
32356 (replace_rhs_if_not_dup): Likewise.
32357 (replace_one_candidate): Likewise.
32359 2017-03-24 Richard Biener <rguenther@suse.de>
32361 PR tree-optimization/80167
32362 * graphite-isl-ast-to-gimple.c
32363 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
32365 (translate_isl_ast_to_gimple::get_rename): Likewise.
32367 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32369 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
32370 handling of certain combinations of target options, including the
32371 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
32372 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
32374 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32377 * config/arm/arm.md (*load_multiple): Add reload_completed to
32378 matching condition.
32380 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32381 Richard Biener <rguenth@suse.de>
32383 PR tree-optimization/79908
32384 PR tree-optimization/80136
32385 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32386 been cast away, gimplify_and_add suffices.
32388 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
32390 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
32392 2017-03-23 Richard Biener <rguenther@suse.de>
32394 PR tree-optimization/80032
32395 * gimplify.c (gimple_push_cleanup): Forced unconditional
32396 cleanups still have to go to the conditional_cleanups
32399 2017-03-22 Jakub Jelinek <jakub@redhat.com>
32401 PR tree-optimization/80072
32402 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
32404 (next_operand_entry_id): Change type to unsigned int.
32405 (sort_by_operand_rank): Make sure to return the right return value
32406 even if unsigned fields are bigger than INT_MAX.
32407 (struct oecount): Change cnt and id type to unsigned int.
32408 (oecount_hasher::equal): Formatting fix.
32409 (oecount_cmp): Make sure to return the right return value
32410 even if unsigned fields are bigger than INT_MAX.
32411 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
32414 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
32415 TREE_READONLY on result if writing it more than once.
32418 * doc/invoke.texi (-fsanitize=thread): Document that with
32419 -fnon-call-exceptions atomics are not able to throw
32423 * tsan.c: Include tree-eh.h.
32424 (instrument_builtin_call): Call maybe_clean_eh_stmt or
32425 maybe_clean_or_replace_eh_stmt where needed.
32426 (instrument_memory_accesses): Add cfg_changed argument.
32427 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
32428 if it returned true.
32429 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
32431 PR rtl-optimization/63191
32432 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
32433 wrapper function, moved the whole old content into ...
32434 (ix86_delegitimize_address_1): ... this. New inline function.
32435 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
32436 true as last argument instead of ix86_delegitimize_address.
32438 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32440 * config/aarch64/aarch64.c (generic_branch_cost): Copy
32441 cortexa57_branch_cost.
32443 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
32445 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
32447 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32450 * doc/md.texi (Constraints): Document wA constraint.
32451 * config/rs6000/constraints.md (wA): New.
32452 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
32453 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
32454 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
32455 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
32457 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
32460 * gimplify.c (is_oacc_declared): New function.
32461 (oacc_default_clause): Use it to set default flags for acc declared
32462 variables inside parallel regions.
32463 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
32464 declared variables.
32465 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
32466 declare attribute to any decl as necessary.
32468 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
32471 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
32472 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
32473 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
32474 (arm_arch_lpae): This.
32475 * config/arm/arm.c (arm_arch7ve): Rename into ...
32476 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
32477 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
32480 2017-03-22 Martin Liska <mliska@suse.cz>
32483 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
32484 error message instead of an ICE.
32486 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32488 * doc/extend.texi (6.11 Additional Floating Types): Revise.
32490 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
32492 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32494 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32497 2017-03-21 Martin Sebor <msebor@redhat.com>
32499 * doc/extend.texi: Use "cannot" instead of "can't."
32500 * doc/hostconfig.texi: Same.
32501 * doc/install.texi: Same.
32502 * doc/invoke.texi: Same.
32503 * doc/loop.texi: Same.
32504 * doc/md.texi: Same.
32505 * doc/objc.texi: Same.
32506 * doc/rtl.texi: Same.
32507 * doc/tm.texi: Same.
32508 * doc/tm.texi.in: Same.
32509 * doc/trouble.texi: Same.
32511 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
32514 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
32515 (collect_checksum_attributes): Set it.
32516 (die_checksum_ordered): Use it.
32518 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32520 PR tree-optimization/79908
32521 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
32522 change: For a VA_ARG whose LHS has been cast away, use
32523 force_gimple_operand to construct the side effects.
32525 2017-03-21 David Malcolm <dmalcolm@redhat.com>
32527 PR translation/80001
32528 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
32529 more amenable to translation.
32530 (oacc_loop_auto_partitions): Likewise.
32532 2017-03-21 Marek Polacek <polacek@redhat.com>
32533 Martin Sebor <msebor@redhat.com>
32535 PR tree-optimization/80109
32536 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
32537 on INTEGRAL_TYPE_P.
32539 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32540 Segher Boessenkool <segher@kernel.crashing.org>
32543 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
32544 check reg_used_between_p between insn and one of succ or succ2
32545 depending on if succ is artificial insn not inserted into insn
32548 2017-03-21 Martin Liska <mliska@suse.cz>
32550 PR gcov-profile/80081
32551 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
32552 * doc/gcc.texi: Include gcov-dump stuff.
32553 * doc/gcov-dump.texi: New file.
32555 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
32557 PR rtl-optimization/79150
32558 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
32559 conditional jump, if the jump is the last insn of the loop.
32561 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32562 Richard Biener <rguenth@suse.de>
32564 PR tree-optimization/79908
32565 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
32566 been cast away, use force_gimple_operand to construct the side
32569 2017-03-21 Martin Liska <mliska@suse.cz>
32571 PR libfortran/79956
32572 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
32575 2017-03-21 Brad Spengler <spender@grsecurity.net>
32578 * plugin.c (htab_hash_plugin): New function.
32579 (add_new_plugin): Use it and adjust.
32580 (parse_plugin_arg_opt): Adjust.
32581 (init_one_plugin): Likewise.
32583 2017-03-21 Richard Biener <rguenther@suse.de>
32585 PR tree-optimization/80032
32586 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
32587 if set force the cleanup to happen unconditionally.
32588 (gimplify_target_expr): Push inserted clobbers with force_uncond
32589 to avoid them being removed by control-dependent DCE.
32591 2017-03-21 Richard Biener <rguenther@suse.de>
32593 PR tree-optimization/80122
32594 * tree-inline.c (copy_bb): Do not expans va-arg packs or
32595 va_arg_pack_len when the inlined call stmt requires pack
32597 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
32599 2017-03-21 Jakub Jelinek <jakub@redhat.com>
32602 * tsan.c (instrument_builtin_call): If the memory model argument
32603 is not a constant, assume it is valid.
32606 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
32609 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32611 PR rtl-optimization/79910
32612 * combine.c (can_combine_p): Do not allow combining an I0 or I1
32613 if its dest is used by an insn before I2 (other than the combined
32614 insns themselves, which are properly handled already).
32616 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
32619 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32621 * combine.c (record_used_regs): New static function.
32622 (try_combine): Handle situations where there is an additional
32623 instruction between I2 and I3 which needs to have a LOG_LINK
32627 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32629 * combine.c (try_combine): Delete redundant i1 test. Call
32630 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32632 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32635 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
32636 alternatives 13/14.
32638 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32640 PR tree-optimization/80054
32641 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
32642 the optimization if a PHI or any of its arguments is not dominated
32643 by the candidate's basis. Use gphi* rather than gimple* as
32645 (replace_profitable_candidates): Clean up a gimple* variable that
32646 should be a gphi* variable.
32648 2017-03-20 Martin Sebor <msebor@redhat.com>
32651 * doc/extend.texi (attribute constructor): Document present limitation.
32653 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
32656 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
32657 __POWER9_VECTOR__ #ifdef control, change template definition to
32658 use Power9-specific built-in function.
32659 (vec_any_eq): Likewise.
32660 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
32661 to control outcomes from this test.
32662 (vector_ae_<mode>p): For VEC_F modes, likewise.
32664 2017-03-20 Ian Lance Taylor <iant@google.com>
32666 * config/i386/i386.c (ix86_function_regparm): Save an extra
32667 register for -fsplit-stack with DECL_STATIC_CHAIN.
32669 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32672 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
32673 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
32675 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
32677 * config/riscv/riscv.c (riscv_print_operand): Use "fence
32679 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
32682 2017-03-20 Marek Polacek <polacek@redhat.com>
32685 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
32687 2017-03-20 Richard Biener <rguenther@suse.de>
32689 PR tree-optimization/80113
32690 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
32691 allocate extra SSA name for PHI def.
32692 (add_close_phis_to_outer_loops): Likewise.
32693 (add_close_phis_to_merge_points): Likewise.
32694 (copy_loop_close_phi_args): Likewise.
32695 (copy_cond_phi_nodes): Likewise.
32697 2017-03-20 Martin Liska <mliska@suse.cz>
32699 PR middle-end/79753
32700 * tree-chkp.c (chkp_build_returned_bound): Do not build
32701 returned bounds for a LHS that's not a BOUNDED_P type.
32703 2017-03-20 Martin Liska <mliska@suse.cz>
32707 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
32708 COMPLEX_CST and VECTOR_CST.
32710 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32713 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
32714 target operand. A new splitter adds the clobber statement in case
32715 the target operand is dead anyway.
32717 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
32719 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
32720 to age-old versions of binutils and glibc.
32722 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
32724 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
32726 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32728 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
32730 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
32732 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
32733 requirement for binutils 2.13.
32735 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
32737 * combine.c (try_combine): Delete redundant i1 test. Call
32738 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
32740 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
32742 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
32743 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
32745 <riscv64-*-elf>: Re-arrange section
32746 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
32747 <riscv32-*-linux>: Likewise.
32748 <riscv64-*-elf>: Likewise
32749 <riscv64-*-linux>: Likewise.
32751 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
32754 * aarch64.opt(verbose-cost-dump): Fix typo.
32756 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
32759 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
32760 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
32762 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
32764 * reload.c (find_reloads): When reloading a nonoffsettable address,
32765 use RELOAD_OTHER for it and its address reloads.
32767 PR rtl-optimization/79910
32768 * combine.c (record_used_regs): New static function.
32769 (try_combine): Handle situations where there is an additional
32770 instruction between I2 and I3 which needs to have a LOG_LINK
32773 2017-03-17 Jeff Law <law@redhat.com>
32775 PR tree-optimization/71437
32776 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
32777 conditional in the hash table first.
32778 (vrp_dom_walker::before_dom_children): Extract condition from
32779 ASSERT_EXPR. Record condition, its inverion and any implied
32780 conditions as well.
32782 2017-03-17 Marek Polacek <polacek@redhat.com>
32783 Markus Trippelsdorf <markus@trippelsdorf.de>
32785 PR tree-optimization/80079
32786 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
32789 2017-03-17 Richard Biener <rguenther@suse.de>
32791 PR middle-end/80075
32792 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
32793 Properly verify the LHS before the RHS possibly claims to be
32795 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
32798 2017-03-17 Martin Jambor <mjambor@suse.cz>
32800 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
32801 (List of -O2 options): Likewise.
32802 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
32805 2017-03-17 Tom de Vries <tom@codesourcery.com>
32807 * gcov-dump.c (print_usage): Print bug_report_url.
32809 2017-03-17 Richard Biener <rguenther@suse.de>
32811 PR middle-end/80050
32812 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
32813 (parser::peek): Likewise.
32815 2017-03-17 Richard Biener <rguenther@suse.de>
32817 PR tree-optimization/80048
32818 * sese.c (free_sese_info): Properly release rename_map and
32819 copied_bb_map elements.
32821 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
32823 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
32824 Add linked-list forward and backlinks. Insert on
32825 construction, remove on destruction.
32826 (class pass_store_merging): Add m_stores_head field.
32827 (pass_store_merging::terminate_and_process_all_chains):
32828 Iterate over m_stores_head list.
32829 (pass_store_merging::terminate_all_aliasing_chains):
32831 (pass_store_merging::execute): Check for debug stmts first.
32832 Push new chains onto the m_stores_head stack.
32834 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
32837 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
32838 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
32839 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
32841 2017-03-16 Jeff Law <law@redhat.com>
32843 PR tree-optimization/71437
32844 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
32845 member function. Implementation moved into after_dom_children
32846 member function and into the threader's thread_outgoing_edges
32848 (dom_opt_dom_walker::after_dom_children): Simplify by moving
32849 some code into new thread_outgoing_edges.
32850 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
32851 definition. Simplify marker handling (do it here). Assume we always
32852 have the available expression and the const/copies tables.
32853 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
32855 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
32856 * tree-vrp.c (equiv_stack): No longer file scoped.
32857 (vrp_dom_walker): New class.
32858 (vrp_dom_walker::before_dom_children): New member function.
32859 (vrp_dom_walker::after_dom_children): Likewise.
32860 (identify_jump_threads): Setup domwalker. Use it rather than
32861 walking edges in a random order by hand. Simplify setup/finalization.
32862 (finalize_jump_threads): Remove.
32863 (vrp_finalize): Do not call identify_jump_threads here.
32864 (execute_vrp): Do it here instead and call thread_through_all_blocks
32867 PR tree-optimization/71437
32868 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
32870 (simplify_stmt_for_jump_threading): Add basic_block argument. All
32872 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
32873 (dom_opt_dom_walker::thread_across_edge): Remove
32874 handle_dominating_asserts argument. All callers changed.
32875 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
32876 changes. Remove calls to lhs_of_dominating_assert. Other
32877 uses of handle_dominating_asserts turn into unconditional code
32878 (simplify_control_stmt_condition_1): Likewise.
32879 (simplify_control_stmt_condition): Likewise.
32880 (thread_through_normal_block, thread_across_edge): Likewise.
32881 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
32882 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
32883 object if it is not an SSA_NAME.
32884 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
32885 before calling into the VRP specific simplifiers.
32886 (identify_jump_threads): Remove handle_dominating_asserts
32889 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32892 * tree-diagnostic.c (default_tree_printer): No longer static.
32893 * tree-diagnostic.h (default_tree_printer): New prototype.
32895 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32897 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32898 Change ins into fmov.
32900 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32902 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32903 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32904 Use h_con constraint for operand 1.
32905 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32906 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32908 2017-03-15 Jeff Law <law@redhat.com>
32910 PR tree-optimization/71437
32911 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32912 (record_temporary_equivalences): Use it.
32914 PR tree-optimization/71437
32915 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32916 tree-ssa-scopedtables.
32917 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32918 (record_conditions, record_cond, vuse_eq): Likewise.
32919 (record_edge_info): Adjust to API tweak of record_conditions.
32920 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32921 (record_temporary_equivalences, optimize_stmt): Likewise.
32922 (eliminate_redundant_computations): Likewise.
32923 (record_equivalences_from_stmt): Likewise.
32924 * tree-ssa-scopedtables.c: Include options.h and params.h.
32925 (vuse_eq): New function, moved from tree-ssa-dom.c
32926 (build_and_record_new_cond): Likewise.
32927 (record_conditions): Likewise. Accept vector of conditions rather
32928 than edge_equivalence structure for first argument.
32929 for the first argument.
32930 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32931 from tree-ssa-dom.c.
32932 (avail_exprs_stack::record_cond): Likewise.
32933 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32934 from tree-ssa-dom.c.
32935 (avail_exprs_stack): Add new member functions lookup_avail_expr
32937 (record_conditions): Declare.
32939 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32942 * lra-constraints.c (process_alt_operands): Increase reject for
32943 reloading an input/output operand.
32945 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32948 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32949 insns to convert from signed/unsigned char/short to IEEE 128-bit
32951 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32953 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32956 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32957 subreg of inner mode for values already in registers.
32959 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32961 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32962 iteration reg is used after the loop.
32964 2017-03-14 Martin Sebor <msebor@redhat.com>
32966 PR tree-optimization/79800
32967 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32968 precision in negative-positive range.
32969 (format_floating): Call non-const overload with adjusted precision.
32971 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32974 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32977 2017-03-14 Martin Sebor <msebor@redhat.com>
32979 PR middle-end/80020
32980 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32981 * builtins.def (aligned_alloc): Use it.
32984 * Makefile.in (GTFILES): Add calls.c.
32985 * calls.c: Include "gt-calls.h".
32987 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32989 PR rtl-optimization/79728
32990 * regs.h (struct target_regs): New field
32991 x_contains_allocatable_regs_of_mode.
32992 (contains_allocatable_regs_of_mode): New macro.
32993 * reginfo.c (init_reg_sets_1): Initialize it, and change
32994 contains_reg_of_mode so it includes global regs as well.
32995 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32996 rather than contains_regs_of_mode.
32998 2017-03-14 Martin Liska <mliska@suse.cz>
33000 * doc/invoke.texi: Document options that can't be combined with
33001 -fcheck-pointer-bounds.
33003 2017-03-14 Martin Liska <mliska@suse.cz>
33005 PR middle-end/79831
33006 * doc/invoke.texi (-Wchkp): Document the option.
33008 2017-03-14 Martin Liska <mliska@suse.cz>
33010 * Makefile.in: Install gcov-dump.
33012 2017-03-14 Martin Liska <mliska@suse.cz>
33014 * multiple_target.c (expand_target_clones): Bail out for
33015 an invalid attribute.
33017 2017-03-14 Richard Biener <rguenther@suse.de>
33019 * alias.c (struct alias_set_entry): Pack properly.
33020 * cfgloop.h (struct loop): Likewise.
33021 * cse.c (struct set): Likewise.
33022 * ipa-utils.c (struct searchc_env): Likewise.
33023 * loop-invariant.c (struct invariant): Likewise.
33024 * lra-remat.c (struct cand): Likewise.
33025 * recog.c (struct change_t): Likewise.
33026 * rtl.h (struct address_info): Likewise.
33027 * symbol-summary.h (function_summary): Likewise.
33028 * tree-loop-distribution.c (struct partition): Likewise.
33029 * tree-object-size.c (struct object_size_info): Likewise.
33030 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33031 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33032 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33033 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33034 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33035 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33036 (struct _stmt_vec_info): Likewise.
33038 2017-03-14 Martin Liska <mliska@suse.cz>
33041 * multiple_target.c (create_dispatcher_calls): Check that
33042 a target can create a function dispatcher.
33044 2017-03-14 Martin Liska <mliska@suse.cz>
33047 * multiple_target.c (expand_target_clones): Drop local.local
33048 flag for default implementation.
33050 2017-03-14 Richard Biener <rguenther@suse.de>
33052 PR tree-optimization/80030
33053 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33055 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33057 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33058 gcc_fallthrough() instead of __attribute__((fallthrough));
33060 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33062 * doc/gcc.texi: Remove "up" link to (DIR).
33063 * doc/gccint.texi: Ditto.
33065 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33067 * doc/install.texi (Specific) <avr>: Remove reference to
33070 2017-03-13 Jeff Law <law@redhat.com>
33072 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33073 attribute rather than comments.
33075 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33076 match_scratch operand is highest.
33078 2017-03-13 Martin Liska <mliska@suse.cz>
33080 PR middle-end/78339
33081 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33082 is a CHKP clone, use original declaration.
33084 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33086 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33087 (arc_conditional_register_usage): Use a different allocation order
33088 when optimizing for size.
33089 * common/config/arc/arc-common.c (arc_option_optimization_table):
33090 Section anchors default on when optimizing for size.
33092 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33094 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33096 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33098 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33099 * config/arc/arc.md (cpu_facility): Add cd variant.
33100 (*movqi_insn): Add code density variant.
33101 (*movhi_insn): Likewise.
33102 (*movqi_insn): Likewise.
33103 (*addsi3_mixed): Likewise.
33104 (subsi3_insn): Likewise.
33106 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33108 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33110 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33112 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33113 expressions with MINUS and UNARY ops.
33115 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33118 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33120 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33121 between vec_select and vector argument.
33122 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33123 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33124 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33125 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33126 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33127 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33129 2017-03-13 Richard Biener <rguenther@suse.de>
33132 * params.def (vect-max-peeling-for-alignment): Fix typo.
33134 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33136 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33137 issue that only occurred with binutils below 2.18.
33139 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33141 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33142 refer to binutils 2.11/2.12 minimum.
33144 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33146 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33147 ftp.kernel.org and simplify binutils requirement.
33149 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33151 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33153 (Optimize Options): Ditto. Also remove redundancy.
33155 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33157 PR translation/79848
33158 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33160 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33161 to G_ to avoid double translation.
33163 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33165 PR translation/79923
33166 * auto-profile.c (get_combined_location): Convert leading
33167 character of diagnostics to lower case and remove trailing period.
33168 (read_profile): Likewise for various diagnostics.
33169 * config/arm/arm.c (arm_option_override): Remove trailing period
33170 from various diagnostics.
33171 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33172 (msp430_expand_delay_cycles): Likewise.
33174 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33177 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33178 full command-line argument, rather than just "str".
33179 (aarch64_validate_march): Likewise.
33180 (aarch64_validate_mtune): Likewise.
33182 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33184 PR rtl-optimization/78911
33185 * lra-assigns.c (must_not_spill_p): New function.
33186 (spill_for): Use it.
33188 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33190 PR tree-optimization/79981
33191 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33192 ATOMIC_COMPARE_EXCHANGE ifn result.
33193 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33194 IFN_ATOMIC_COMPARE_EXCHANGE.
33196 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33199 * opts.c (parse_sanitizer_options): Add missing question mark to
33200 "did you mean" message.
33202 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33204 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33206 (VMULEUH_UNS): Likewise.
33207 (VMULOUB_UNS): Likewise.
33208 (VMULOUH_UNS): Likewise.
33209 * config/rs6000/rs6000.c (builtin_function_type): Remove
33210 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33212 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33215 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33216 x with result of extra_parsing_for_operand_code_0.
33217 (function_reader::extra_parsing_for_operand_code_0): Convert
33218 return type from void to rtx, returning x. When reading
33219 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33220 larger size containing struct block_symbol.
33222 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33224 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33225 -mfloat128-hardware without -m64.
33227 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33230 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33231 entries to the case statement that marks unsigned arguments to
33232 overloaded functions.
33234 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33236 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33237 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33239 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33242 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33243 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33245 2017-03-10 Martin Liska <mliska@suse.cz>
33249 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
33250 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
33253 2017-03-10 Olivier Hainque <hainque@adacore.com>
33255 * tree-switch-conversion (array_value_type): Start by resetting
33256 candidate type to it's main variant.
33258 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33260 PR rtl-optimization/79909
33261 * combine.c (try_combine): Use simplify_replace_rtx on individual
33262 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
33263 of the whole CALL_INSN_FUNCTION_USAGE.
33265 PR tree-optimization/79972
33266 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
33267 get_range_info on SSA_NAMEs. Formatting fixes.
33269 2017-03-10 Richard Biener <rguenther@suse.de>
33270 Jakub Jelinek <jakub@redhat.com>
33272 PR tree-optimization/77975
33273 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
33274 edge to be constant.
33275 (get_val_for): For constant x return it. Formatting fix.
33276 (loop_niter_by_eval): Avoid pointless looping if the next iteration
33277 would use the same bases as the current one.
33279 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33281 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
33282 instead of vec_select for V1TImode.
33283 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
33285 (VSX_LE_128): Add V1TI to this mode iterator.
33286 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
33287 (*vsx_le_perm_store_<mode>): Likewise.
33288 (pre-reload splitter for VSX stores): Likewise.
33289 (post-reload splitter for VSX stores): Likewise.
33290 (*vsx_xxpermdi2_le_<mode>): Likewise.
33291 (*vsx_lxvd2x2_le_<mode>): Likewise.
33292 (*vsx_stxvd2x2_le_<mode>): Likewise.
33294 2017-03-09 Michael Eager <eager@eagercon.com>
33296 Correct failures with --enable-checking=yes,rtl.
33298 * config/microblaze/microblaze.c (microblaze_expand_shift):
33299 Replace GET_CODE test with CONST_INT_P and INTVAL test with
33300 test for const0_rtx.
33301 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
33302 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
33304 2017-03-09 Richard Biener <rguenther@suse.de>
33306 PR tree-optimization/79977
33307 * graphite-scop-detection.c (scop_detection::merge_sese):
33308 Handle the case of extra exits to blocks dominating the entry.
33310 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
33312 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
33315 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
33317 PR rtl-optimization/79949
33318 * lra-constraints.c (process_alt_operands): Check memory when
33319 trying to predict a cycle. Print about the overall increase.
33321 2017-03-09 Richard Biener <rguenther@suse.de>
33323 PR middle-end/79971
33324 * gimple-expr.c (useless_type_conversion_p): Preserve
33325 TYPE_SATURATING for fixed-point types.
33327 2017-03-09 Richard Biener <rguenther@suse.de>
33330 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
33331 alignment of BLKmode params.
33333 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33336 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
33337 (VALL_NO_V2Q): Likewise.
33339 * config/aarch64/aarch64-simd.md
33340 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
33342 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
33343 VALL_NO_V2Q mode iterator.
33344 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
33346 2017-03-09 Martin Liska <mliska@suse.cz>
33348 PR tree-optimization/79631
33349 * tree-chkp-opt.c (chkp_is_constant_addr): Call
33350 tree_int_cst_sign_bit just for INTEGER constants.
33352 2017-03-09 Martin Liska <mliska@suse.cz>
33356 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
33359 2017-03-09 Marek Polacek <polacek@redhat.com>
33362 * tree.c (inchash::add_expr): Handle TREE_VEC.
33364 2017-03-09 Martin Liska <mliska@suse.cz>
33367 (chkp_narrow_size_and_offset): New function.
33368 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
33369 (void chkp_parse_bit_field_ref): New function.
33370 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
33371 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
33373 2017-03-09 Martin Liska <mliska@suse.cz>
33376 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
33377 (chkp_find_bounds_1): Remove gcc_unreachable.
33379 2017-03-09 Jakub Jelinek <jakub@redhat.com>
33382 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
33383 BUILT_IN_SYNC*, determine the access type from the size suffix and
33384 always build a MEM_REF with that type. Handle forgotten
33385 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
33388 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
33389 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
33390 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
33391 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
33392 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
33393 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
33394 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
33395 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
33396 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
33397 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
33398 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
33399 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
33400 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
33401 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
33402 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
33403 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
33404 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
33405 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
33406 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
33407 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
33408 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
33409 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
33410 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
33411 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
33412 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
33413 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
33414 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
33415 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
33416 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
33417 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
33418 definitions outside of __OPTIMIZE__ guarded section.
33421 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
33422 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
33423 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
33424 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
33427 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33429 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
33430 ("vfenez<mode>"): Add missing constraints.
33432 2017-03-08 Martin Sebor <msebor@redhat.com>
33435 * config/nds32/nds32.c (nds32_option_override):
33436 Fix misspelled diagnostic.
33438 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33441 * gimplify.c (gimplify_omp_for): Replace index var in outer
33442 taskloop statement with an artificial variable and add
33443 OMP_CLAUSE_PRIVATE clause for it.
33445 2017-03-08 Richard Biener <rguenther@suse.de>
33447 PR tree-optimization/79955
33448 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
33449 for accesses that are completely outside of the variable.
33451 2017-03-08 Andrew Haley <aph@redhat.com>
33453 PR tree-optimization/79943
33454 * tree-ssa-loop-split.c (compute_new_first_bound): When
33455 calculating the new upper bound, (END-BEG) should be added, not
33458 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33460 * config/avr/avr.md (setmemhi): Make sure match_dup
33461 operand number comes before match_scratch.
33463 2017-03-08 Richard Biener <rguenther@suse.de>
33465 PR tree-optimization/79920
33466 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
33467 with ncopies == 1 to ...
33468 (vect_transform_slp_perm_load): ... here. Properly compute
33469 all element loads by iterating VF times over the group. Do
33470 not handle ncopies (computed in a broken way) in
33471 vect_create_mask_and_perm.
33473 2017-03-08 Jakub Jelinek <jakub@redhat.com>
33476 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
33477 is a uniform vector, use uniform_vector_p return value instead of
33478 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
33480 2017-03-07 Marek Polacek <polacek@redhat.com>
33482 PR middle-end/79809
33483 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
33484 (alloca_call_type): Likewise.
33486 2017-03-07 Martin Liska <mliska@suse.cz>
33488 * gcov.c (process_args): Put comment to correct location.
33490 2017-03-07 Martin Liska <mliska@suse.cz>
33492 PR middle-end/68270
33493 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
33494 Use array_at_struct_end_p instead of DECL_CHAIN (field).
33495 (chkp_narrow_bounds_for_field): Likewise.
33496 (chkp_parse_array_and_component_ref): Pass one more argument to
33499 2017-03-07 Richard Biener <rguenther@suse.de>
33501 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
33504 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
33506 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
33507 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
33509 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33512 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
33513 to end of description.
33514 (PARAM_MAX_STORES_TO_MERGE): Likewise.
33516 2017-03-07 Jakub Jelinek <jakub@redhat.com>
33518 PR rtl-optimization/79901
33519 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
33521 (*avx512f_<code><mode>3<mask_name>): ... this.
33522 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
33523 iterator instead of VI8_AVX2_AVX512BW.
33525 PR rtl-optimization/79901
33526 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
33527 min/max expander, expand it using expand_vec_cond_expr.
33530 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
33533 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33536 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
33537 to void * for PCH reasons.
33538 * dwarf2out.c (output_loc_operands, output_die): Cast
33539 v.val_vec.array to unsigned char *.
33541 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
33544 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
33547 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
33549 PR rtl-optimization/79571
33550 * lra-constraints.c (process_alt_operands): Calculate static
33551 reject and subtract it from overall when only addresses will be
33554 2017-03-06 Julia Koval <julia.koval@intel.com>
33557 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
33558 incoming stack boundary to 128 for 64-bit targets.
33560 2017-03-06 Richard Biener <rguenther@suse.de>
33562 PR tree-optimization/79894
33563 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
33564 to NULL after folding it.
33566 2017-03-06 Richard Biener <rguenther@suse.de>
33568 PR tree-optimization/79824
33569 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
33570 check disabling peeling for gaps.
33572 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
33574 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
33575 attributes): Document gettimeofday.
33577 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33579 * config/s390/s390.c (s390_option_override_internal): Set
33580 PARAM_MIN_VECT_LOOP_BOUND
33582 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
33584 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
33585 * config/s390/s390.md: Likewise.
33587 2017-03-06 Jakub Jelinek <jakub@redhat.com>
33590 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
33591 (<avx2_avx512>_perm<mode>): Rename to ...
33592 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
33594 (<avx512>_perm<mode>_mask): Rename to ...
33595 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
33597 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
33598 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
33599 instead of VI8F_256_512.
33600 (avx512f_perm<mode>): New define_expand.
33601 (avx512f_perm<mode>_mask): Likewise.
33602 (avx512f_perm<mode>_1<mask_name>): New define_insn.
33603 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
33605 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33607 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
33608 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
33610 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
33612 2017-03-06 Martin Liska <mliska@suse.cz>
33615 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
33616 when having a SSA NAME w/o VAR_DECL assigned to it.
33618 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33620 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
33621 msa_dpsub_<su>_d): Fix MODE for vec_select.
33623 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
33625 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
33627 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
33629 2017-03-06 Richard Biener <rguenther@suse.de>
33631 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
33632 * plugin.c (register_plugin_info): Likewise.
33633 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
33635 2017-03-05 Jakub Jelinek <jakub@redhat.com>
33637 * config/i386/sse.md (sse_storehps, sse_storelps,
33638 avx_<castmode><avxsizesuffix>_<castmode>,
33639 avx512f_<castmode><avxsizesuffix>_<castmode>,
33640 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
33641 in condition that at least one operand is not a MEM.
33643 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33645 PR middle-end/79805
33646 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
33647 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
33649 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
33650 gimple_call_nothrow_p flag based on whether original builtin can throw.
33651 If it can, emit following stmts on the fallthrough edge.
33652 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
33653 don't create new bb if inserting just debug stmts on the edge, try to
33654 insert them on the fallthru bb or just reset debug stmts.
33656 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
33659 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
33660 restore recog_data (including the operand rtxes inside it) around
33661 the call to get_insn_template.
33663 2017-03-03 Martin Sebor <msebor@redhat.com>
33665 PR tree-optimization/79699
33666 * context.c (context::~context): Free MPFR caches to avoid
33667 a memory leak on program exit.
33669 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33671 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
33672 Use wide_int::ulow () instead of .elt (0).
33674 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33676 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
33677 (*pushxf): Limit oF constraint to 32bit targets and add oC
33678 constraint for 64bit targets.
33679 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
33680 (*pushdf): Change rmF constraint to rmC.
33682 2017-03-03 Martin Liska <mliska@suse.cz>
33684 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
33685 Remove unused variable.
33687 2017-03-03 Jakub Jelinek <jakub@redhat.com>
33690 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
33691 is a memory operand, increase num_memory.
33692 (ix86_expand_args_builtin): Likewise.
33694 2017-03-03 Jan Hubicka <jh@suse.cz>
33697 * ipa-devirt.c (maybe_record_node): Properly handle
33698 __cxa_pure_virtual visibility.
33700 2017-03-03 Martin Liska <mliska@suse.cz>
33702 PR tree-optimization/79803
33703 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
33705 (pass_loop_prefetch::execute): Disabled optimization if an
33706 assumption about L1 cache size is not met.
33708 2017-03-03 Martin Liska <mliska@suse.cz>
33710 PR rtl-optimization/79574
33711 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
33712 (hash_scan_set): Likewise.
33713 (dump_hash_table): Likewise.
33714 (hoist_code): Likewise.
33716 2017-03-03 Richard Biener <rguenther@suse.de>
33718 * fixed-value.c (fixed_from_string): Restore use of elt (1)
33719 in place of uhigh ().
33720 (fixed_convert_from_real): Likewise.
33722 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
33725 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
33727 2017-03-03 Richard Biener <rguenther@suse.de>
33729 PR middle-end/79818
33730 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
33731 TYPE_OVERFLOW_UNDEFINED check.
33733 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33735 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
33737 (vector_ae_<mode>_p): Likewise.
33738 (vector_nez_<mode>_p): Likewise.
33739 (vector_ne_v2di_p): Likewise.
33740 (vector_ae_v2di_p): Likewise.
33741 (vector_ne_<mode>_p): Likewise.
33742 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
33744 (vsx_tsqrt<mode>2_fe): Likewise.
33746 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
33749 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
33751 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33753 PR rtl-optimization/79780
33754 * cprop.c (one_cprop_pass): When second and further conditional trap
33755 in a single basic block is turned into an unconditional trap, turn it
33756 into a deleted note to avoid RTL verification failures.
33758 2017-03-02 Richard Biener <rguenther@suse.de>
33760 * fold-const.c (const_binop): Use ulow () instead of elt (0).
33762 2017-03-02 Richard Biener <rguenther@suse.de>
33764 PR tree-optimization/79345
33766 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
33767 param and abort the walk, returning -1 if it is hit.
33768 (walk_aliased_vdefs): Take a limit param and pass it on.
33769 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
33770 defaulting to 0 and return a signed int.
33771 * tree-ssa-uninit.c (struct check_defs_data): New struct.
33772 (check_defs): New helper.
33773 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
33774 about uninitialized memory.
33775 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
33776 bogus uninitialized warning.
33777 (fixed_convert_from_real): Likewise.
33779 2017-03-02 Bin Cheng <bin.cheng@arm.com>
33781 PR tree-optimization/66768
33782 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
33783 iv_use if base object can't be determined.
33785 2017-03-02 Jakub Jelinek <jakub@redhat.com>
33787 PR tree-optimization/79345
33788 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
33789 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
33790 (get_pattern_stats): Initialize it.
33791 * genemit.c (gen_expand): Verify match_scratch numbers come after
33792 match_operand/match_dup numbers.
33793 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
33794 match_scratch numbers.
33795 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
33797 * config/s390/s390.md (trunctdsd2): Likewise.
33799 2017-03-02 Richard Biener <rguenther@suse.de>
33801 * wide-int.h (wide_int_storage::operator=): Implement in terms
33804 2017-03-02 Richard Biener <rguenther@suse.de>
33806 PR tree-optimization/79777
33807 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
33808 the to insert expression to sth existing.
33810 2017-03-01 Martin Sebor <msebor@redhat.com>
33812 PR middle-end/79692
33813 * gimple-ssa-sprintf.c
33814 (directive::known_width_and_precision): New function.
33815 (format_integer): Use it.
33816 (get_mpfr_format_length): Consider the full range of precision
33817 when computing %g output with the # flag. Set the likely byte
33818 count to 3 rather than 1 when precision is indeterminate.
33819 (format_floating): Correct the lower bound of precision.
33821 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33823 * doc/invoke.texi: Document default code model for 64-bit Linux.
33825 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33828 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
33829 udiv rather than div since input pattern is unsigned.
33831 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
33833 * config/i386/i386.c (print_reg): Warn for values of
33834 unsupported size in integer register.
33836 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
33839 * config/rs6000/predicates.md (current_file_function_operand): Do
33840 not allow self calls to be local if the function is replaceable.
33842 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
33845 * config/rs6000/altivec.h (vec_ctz and others): Change the
33846 preprocessor macro that controls conditional compilation from
33847 _ARCH_PWR9 to __POWER9_VECTOR__.
33848 (vec_all_ne): Change parameterization of __altivec_scalar_pred
33849 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
33850 control (instead of _ARCH_PWR9 control) so that template
33851 definition uses power9-specific function.
33852 (vec_any_eq): Likewise.
33853 (vec_all_ne): Change macro definition to use a power9-specific
33854 expansion under #ifdef __POWER9_VECTOR__ control (instead of
33855 _ARCH_PWR9 control).
33856 (vec_any_eq) Likewise.
33857 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
33858 expansion for CMPNEF to remove support for xvcmpnesp instruction.
33859 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
33860 support for xvcmpnedp instruction.
33861 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
33862 macro expansion so that Power9 implementation of vec_all_ne does
33863 not use the AltiVec predicate framework.
33864 (VCMPNEH_P): Likewise.
33865 (VCMPNEW_P): Likewise.
33866 (VCMPNED_P): Likewise.
33867 (VCMPNEFP_P): Likewise.
33868 (VCMPNEDP_P): Likewise.
33869 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
33870 implementation of vec_any_eq to not use AltiVec predicate
33872 (VCMPAEH_P): Likewise.
33873 (VCMPAEW_P): Likewise.
33874 (VCMPAED_P): Likewise.
33875 (VCMPAEFP_P): Likewise.
33876 (VCMPAEDP_P): Likewise.
33877 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
33878 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
33879 not use the AltiVec predicate framework.
33880 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
33881 of vec_any_eq to not use AltiVec predicate framework.
33882 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33883 support for predefined __POWER9_VECTOR__ macro to indicate that
33884 Power9 instruction selection is enabled.
33885 (altivec_overloaded_builtins): Remove extraneous
33886 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33887 function argument types RS6000_BTI_bool_V16QI and
33888 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33889 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33890 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33891 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33892 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33893 Power9 for implementations of vec_cmpne. Change the signature for
33894 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33895 (representing vec_all_ne) to remove the previously described first
33896 argument of type RS6000_BTI_INTSI, as this was an artifact of
33897 reliance on the AltiVec predicate framework, which is no longer
33898 used in the implementation of these functions. Add
33899 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33900 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33901 since, unlike the AltiVec predicate framework implementation, we
33902 do not share function descriptors between vec_alle and vec_anyeq.
33903 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33904 set of modes that receive special treatment even when
33905 TARGET_P9_VECTOR is true. The special treatment emits code that
33906 does not depend on Power9 instructions.
33907 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33908 define_expand to not rely on AltiVec predicate framework.
33909 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33911 (vector_ne_v2di_p): Change this define_expand to not rely on
33912 AltiVec predicate framework.
33913 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33915 (vector_ne_<mode>_p): Change this define_expand to not rely on
33916 AltiVec predicate framework.
33917 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33919 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33920 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33921 define_insn pattern.
33922 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33923 define_insn pattern because the xvcmpne<VSs>. instruction is not
33925 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33926 instruction is not supported.
33928 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33930 * config/nvptx/nvptx.c: Include intl.h.
33932 2017-03-01 Martin Jambor <mjambor@suse.cz>
33935 * ipa-prop.h (ipa_bits): Removed field known.
33936 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33937 to pointers. Adjusted their comments to warn about their sharing.
33938 (ipcp_transformation_summary): Change bits to a vector of pointers.
33939 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33940 (ipa_get_ipa_bits_for_value): Declare.
33941 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33942 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33943 (ipa_bits_hash_table): Likewise.
33944 (ipa_vr_ggc_hash_traits): Likewise.
33945 (ipa_vr_hash_table): Likewise.
33946 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33947 being pointers and vr_known being removed.
33948 (ipa_set_jf_unknown): Likewise.
33949 (ipa_get_ipa_bits_for_value): New function.
33950 (ipa_set_jfunc_bits): Likewise.
33951 (ipa_get_value_range): New overloaded functions.
33952 (ipa_set_jfunc_vr): Likewise.
33953 (ipa_compute_jump_functions_for_edge): Use the above functions to
33954 construct bits and vr parts of jump functions.
33955 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33956 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33958 (ipcp_grow_transformations_if_necessary): Also allocate
33959 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33961 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33962 them. Fix too long lines.
33963 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33964 vr_known being removed.
33965 (ipa_read_jump_function): Use new setter functions to construct bits
33966 and vr parts of jump functions or set them to NULL.
33967 (write_ipcp_transformation_info): Adjust for bits being pointers.
33968 (read_ipcp_transformation_info): Likewise.
33969 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33971 Include gt-ipa-prop.h.
33972 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33974 (ipcp_store_bits_results): Likewise.
33975 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33976 Do not write to existing jump functions but use a temporary instead.
33978 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33981 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33982 attempt to use its first operand as BIT_FIELD_REF base.
33984 2017-03-01 Richard Biener <rguenther@suse.de>
33986 PR middle-end/79721
33987 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33988 interpolating formula in wrapping arithmetic.
33989 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33991 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33993 PR tree-optimization/79734
33994 * tree-vect-generic.c (expand_vector_condition): Optimize
33995 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33996 Handle VEC_COND_EXPR where comparison has different inner width from
33997 type's inner width.
33999 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34001 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34002 markup, and similar issues. Remove @opindex entries for things
34003 that aren't options. Add missing -mmpy-option entries.
34005 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34007 PR tree-optimization/79737
34008 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34009 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34010 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34011 instead of byte_size. Formatting fix.
34012 (shift_bytes_in_array_right): Formatting fix.
34014 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34017 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34018 condition on optimize for the leaf function test.
34020 2017-02-28 Martin Liska <mliska@suse.cz>
34023 * read-rtl-function.c (function_reader::handle_unknown_directive):
34024 Bail out when one uses -flto.
34026 2017-02-28 Martin Liska <mliska@suse.cz>
34028 * common.opt: Replace space with tabular for options of <number>
34030 * config/i386/i386.opt: Show <number> value for
34031 -mlarge-data-threshold.
34032 * opts.c (print_filtered_help): Do not display number in hexadecimal
34035 2017-02-28 Martin Liska <mliska@suse.cz>
34037 * common.opt: Fix --help=option -Q for options which are of
34040 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34042 * config/i386/i386.c (print_reg): Error out for values
34043 of 8-bit size in invalid integer register.
34045 2017-02-28 Martin Sebor <msebor@redhat.com>
34047 PR tree-optimization/79691
34048 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34050 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34053 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34054 gcc_unreachable with output_operand_lossage.
34056 2017-02-28 Richard Biener <rguenther@suse.de>
34058 PR tree-optimization/79740
34059 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34061 (visit_nary_op): Insert the nary into the hashtable if we
34062 pattern-matched sth.
34063 * tree-ssa-pre.c (eliminate_insert): Robustify.
34065 2017-02-28 Richard Biener <rguenther@suse.de>
34067 PR middle-end/79731
34068 * fold-const.c (decode_field_reference): Reject out-of-bound
34071 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34073 * config/i386/i386.c: Include intl.h.
34074 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34075 instead of just cond ? "..." : "...".
34076 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34077 * coverage.c (read_counts_file): Likewise.
34078 * omp-offload.c: Include intl.h.
34079 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34080 of just cond ? "..." : "...".
34081 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34082 of just cond ? "..." : "...".
34084 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34087 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34089 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34090 'tune for' CPU name.
34091 * config/arm/arm-cpu-data.h: Regenerated.
34093 2017-02-28 Richard Biener <rguenther@suse.de>
34095 PR tree-optimization/79732
34096 * tree-inline.c (expand_call_inline): Do not shadow var.
34098 2017-02-28 Richard Biener <rguenther@suse.de>
34100 PR tree-optimization/79723
34101 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34102 address-space properly.
34104 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34106 * doc/optinfo.texi (Optimization groups): Fix option used for
34108 * doc/invoke.texi (-fopt-info): Document "omp".
34109 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34110 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34111 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34112 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34113 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34115 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34117 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34118 OPTGROUP_OMP with "omp".
34120 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34123 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34124 for arithmetic shift of unsigned V2DI.
34126 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34128 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34129 arc/linux.h headers.
34130 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34131 (LINK_SPEC): Likewise.
34132 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34133 (EXTRA_SPECS): Likewise.
34134 (STARTFILE_SPEC): Likewise.
34135 (ENDFILE_SPEC): Likewise.
34136 (LIB_SPEC): Likewise.
34137 (TARGET_SDATA_DEFAULT): Likewise.
34138 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34139 (MULTILIB_DEFAULTS): Likewise.
34140 (DWARF2_UNWIND_INFO): Likewise.
34141 * config/arc/big.h: New file.
34142 * config/arc/elf.h: Likewise.
34143 * config/arc/linux.h: Likewise.
34144 * config/arc/t-uClibc: Remove.
34146 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34148 PR tree-optimization/77536
34149 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34150 (tree_transform_and_unroll_loop): Use above function to compute the
34151 estimated niter of unrolled loop and use it when scaling profile.
34152 Also use count info rather than frequency if it's non-zero.
34153 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34154 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34155 (vect_transform_loop): Call above function.
34157 2017-02-27 Richard Biener <rguenther@suse.de>
34159 PR tree-optimization/45397
34160 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34161 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34162 (visit_nary_op): Add pattern matching for CSEing sign-changed
34163 or truncated operations with wider ones.
34165 2017-02-27 Richard Biener <rguenther@suse.de>
34167 PR tree-optimization/79690
34168 * tree-vect-stmts.c (vectorizable_store): Use vector type
34169 built from the DR with address-space.
34171 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34173 * doc/invoke.texi (Optimize Options): Refine the description
34174 of asan-use-after-return.
34176 2017-02-25 Alan Modra <amodra@gmail.com>
34178 PR rtl-optimization/79584
34179 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34180 base, not ad->base_term, the reg within base. Remove assertion
34181 that ad->base == ad->base_term. Replace gen_int_mode using
34182 bogus mode with const0_rtx.
34184 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34186 PR middle-end/79396
34187 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34188 FMA_EXPR like tcc_binary or tcc_unary.
34190 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34193 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34195 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34196 (output_loc_operands): Handle DW_OP_call_ref and
34197 DW_OP_GNU_variable_value.
34198 (struct variable_value_struct): New type.
34199 (struct variable_value_hasher): Likewise.
34200 (variable_value_hash): New variable.
34201 (string_types): Remove.
34202 (copy_loc_descr): New function.
34203 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34204 (prepend_loc_descr_to_each): New function.
34205 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34206 instead of add_loc_descr_to_each if the first argument is single
34207 location list and the second has multiple.
34208 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34209 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34210 when looking for variable value which doesn't have other location info.
34211 (loc_list_from_tree): Formatting fix.
34212 (gen_array_type_die): Simplify DW_AT_string_length handling.
34213 (adjust_string_types): Remove.
34214 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34215 string_types. Call resolve_variable_values.
34216 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34217 (resolve_addr_in_expr): Likewise. Add A argument.
34218 (copy_deref_exprloc): Remove deref argument. Adjust for the
34219 original expression being DW_OP_GNU_variable_value with optionally
34220 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34221 optionally after it.
34222 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34223 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34224 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34225 (variable_value_hasher::hash, variable_value_hasher::equal): New
34227 (resolve_variable_value_in_expr, resolve_variable_value,
34228 resolve_variable_values, note_variable_value_in_expr,
34229 note_variable_value): New functions.
34230 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34233 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34236 * opts.h (handle_generated_option): Add GENERATED_P argument.
34237 * opts-common.c (handle_option): Adjust function comment.
34238 (handle_generated_option): Add GENERATED_P argument, pass it to
34240 (control_warning_option): Pass false to handle_generated_option
34242 * opts.c (maybe_default_option): Pass true to handle_generated_option
34244 * optc-gen.awk: Likewise.
34246 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34248 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34249 a REG, look at the REG it is a SUBREG of.
34250 (splitter for cmpeqsi_t): Ditto.
34252 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34254 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
34255 the special USEs with the pattern of the insn, not the insn itself.
34257 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
34260 * doc/invoke.texi: Document -mload-store-pairs.
34262 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34263 Sandra Loosemore <sandra@codesourcery.com>
34265 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
34266 argument isn't a CONST_INT.
34267 (nios2_alternate_compare_const): Assert op is a CONST_INT.
34268 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
34269 (nios2_validate_compare): Bypass alternate compare logic if *op2
34270 is not a CONST_INT.
34271 (ldstwm_operation_p): Return false if first_base is not a REG or
34272 if first_offset is not a CONST_INT.
34274 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34276 * config/cris/cris.md: Use correct operand in a define_peephole2.
34278 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34280 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
34282 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34284 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
34285 this_insn if it is an INSN or JUMP_INSN.
34286 (force_offsettable): Look at base, not at addr.
34287 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
34288 on things that aren't necessarily CONST_INTs.
34290 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
34292 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
34293 -mfpmath=sse is the default also for x86-32 targets with SSE2
34294 instruction set when @option{-ffast-math} is enabled
34296 2017-02-24 Jeff Law <law@redhat.com>
34298 PR rtl-optimizatoin/79286
34299 * ira.c (update_equiv_regs): Drop may_trap_p exception to
34302 2017-02-24 Richard Biener <rguenther@suse.de>
34304 PR tree-optimization/79389
34305 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
34308 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
34310 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
34311 function comment to reflect reality.
34312 (loop_exits_before_overflow): Fix typo in function description.
34314 2017-02-24 Richard Biener <rguenther@suse.de>
34316 PR tree-optimization/79389
34317 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
34318 properly that a threading opportunity exists. Detect conditional
34319 copy/constant propagation opportunities.
34321 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
34323 * config/visium/visium.md (type): Add trap.
34324 (b): New mode attribute.
34325 (*btst): Rename into...
34326 (*btst<mode>): ...this and adjust.
34327 (*cbranchsi4_btst_insn): Rename into...
34328 (*cbranch<mode>4_btst_insn): ...this and adjust.
34329 (trap): New define_insn.
34331 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34333 PR tree-optimization/79389
34334 * ifcvt.c (struct noce_if_info): Add rev_cond field.
34335 (noce_reversed_cond_code): New function.
34336 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
34337 reversed_comparison_code. Formatting fix.
34338 (noce_try_store_flag): Test rev_cond != NULL in addition to
34339 reversed_comparison_code.
34340 (noce_try_store_flag_constants): Likewise.
34341 (noce_try_store_flag_mask): Likewise.
34342 (noce_try_addcc): Use rev_cond if non-NULL instead of
34343 reversed_comparison_code.
34344 (noce_try_cmove_arith): Likewise. Formatting fixes.
34345 (noce_try_minmax, noce_try_abs): Clear rev_cond.
34346 (noce_find_if_block): Initialize rev_cond.
34347 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
34348 instead of false as last argument never attempt to reverse it
34351 2017-02-23 Bin Cheng <bin.cheng@arm.com>
34353 PR tree-optimization/79663
34354 * tree-predcom.c (combine_chains): Process refs in reverse order
34355 only for ZERO length chains, and add explaining comment.
34357 2017-02-23 Jeff Law <law@redhat.com>
34359 PR tree-optimization/79578
34360 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
34361 in call to operand_equal_p.
34363 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
34366 * config/i386/cpuid.h: Fix another undefined behavior.
34368 2017-02-23 Richard Biener <rguenther@suse.de>
34370 PR tree-optimization/79683
34371 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
34372 vector types for data-refs.
34374 2017-02-23 Martin Liska <mliska@suse.cz>
34376 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
34378 2017-02-23 Jakub Jelinek <jakub@redhat.com>
34380 PR middle-end/79665
34381 * internal-fn.c (get_range_pos_neg): Moved to ...
34382 * tree.c (get_range_pos_neg): ... here. No longer static.
34383 * tree.h (get_range_pos_neg): New prototype.
34384 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
34385 are known to be in between 0 and signed maximum inclusive, try to
34386 expand both unsigned and signed divmod and use the cheaper one from
34389 2017-02-22 Jeff Law <law@redhat.com>
34391 PR tree-optimization/79578
34392 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
34393 to compare base operands.
34395 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
34398 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
34399 gpc_reg_operand instead of fpr_reg_operand.
34401 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
34403 * config/mips/mips.c (mips_return_in_memory): Force FP
34404 vector types to be returned in memory for o32 ABI.
34406 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34408 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
34409 instead of DW_TAG_member for static data member declarations and don't
34410 set no_linkage_name for static inline data members.
34411 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
34414 2017-02-22 Martin Liska <mliska@suse.cz>
34416 * doc/invoke.texi: Replace inequality signs with square brackets
34419 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
34422 * lra-constraints.c (simplify_operand_subreg): Handle
34423 WORD_REGISTER_OPERATIONS targets.
34425 2017-02-22 Jakub Jelinek <jakub@redhat.com>
34428 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
34429 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
34430 elimination by swapping fld*.
34432 2017-02-22 Richard Biener <rguenther@suse.de>
34434 PR tree-optimization/79673
34435 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
34436 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
34437 irrelevant address-space qualifiers and avoiding a
34438 ADDR_SPACE_CONVERT_EXPR from fold_convert.
34440 2017-02-22 Richard Biener <rguenther@suse.de>
34442 PR tree-optimization/79666
34443 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
34444 to not symbolically negate if that may introduce undefined
34447 2017-02-22 Martin Liska <mliska@suse.cz>
34450 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
34451 * data-streamer-out.c (streamer_write_gcov_count_stream):
34453 * value-prof.c (stream_out_histogram_value): Make assert more
34454 precise based on type of counter.
34456 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
34459 * config/i386/i386.md (standard_x87sse_constant_load splitter):
34460 Use nonimmediate_operand instead of memory_operand for operand 1.
34461 (float-extend standard_x87sse_constant_load splitter): Ditto.
34463 2017-02-21 Jeff Law <law@redhat.com>
34465 PR tree-optimization/79621
34466 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
34467 blocks with edges to themselves.
34469 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34472 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
34473 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
34474 Use gimple_call_builtin_p.
34477 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
34478 on temporarily removed DEBUG_INSNs.
34480 PR tree-optimization/79649
34481 * tree-loop-distribution.c (classify_partition): Give up on
34482 non-generic address space loads/stores.
34484 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
34486 * doc/loop.texi (Loop manipulation): Remove nonexistent
34487 tree_ssa_loop_version from the documentation.
34488 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
34490 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34493 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
34494 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
34495 * config/rs6000/rs6000.c: Include except.h.
34496 (rs6000_expand_split_stack_prologue): Call
34497 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
34499 2017-02-21 Martin Jambor <mjambor@suse.cz>
34502 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
34503 have been analyzed.
34505 2017-02-21 Martin Jambor <mjambor@suse.cz>
34507 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
34508 for backward compatibility only.
34509 * doc/invoke.texi (Option Summary): Remove all references to
34510 -fipa-cp-alignment.
34512 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
34516 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34518 * lra-constraints.c (curr_insn_transform): Handle
34519 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34521 2017-02-21 Martin Liska <mliska@suse.cz>
34523 * config/i386/i386.opt: Replace -masm-dialect with -masm.
34525 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
34527 PR translation/79638
34528 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
34530 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
34533 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
34534 (arm_function_ok_for_sibcall): Return false for an indirect call by
34535 descriptor if all the argument registers are used.
34536 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
34537 alignment of the function.
34539 2017-02-21 Jakub Jelinek <jakub@redhat.com>
34541 PR tree-optimization/61441
34542 * simplify-rtx.c (simplify_const_unary_operation): For
34543 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
34544 the sNaN unmodified.
34546 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34548 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34549 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34550 instead of SYSTEM_HEADER_DIR.
34552 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
34553 Martin Liška <mliska@suse.cz>
34555 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
34556 Fix typos and grammar, use active voice, and clarify.
34558 2017-02-20 Marek Polacek <polacek@redhat.com>
34560 PR middle-end/79537
34561 * gimplify.c (gimplify_expr): Handle unused *&&L;.
34564 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
34566 2017-02-20 Jakub Jelinek <jakub@redhat.com>
34569 * config/i386/i386.c (ix86_expand_builtin): Handle
34570 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
34571 ix86_builtins_isa[fcode].isa as a requirement of those
34572 flags and any other flag in the bitmask.
34573 (ix86_init_mmx_sse_builtins): Use 0 instead of
34574 ~OPTION_MASK_ISA_64BIT as mask.
34575 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
34576 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
34577 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
34578 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
34580 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34583 * lra-constraints.c (split_reg): Check requested split mode
34584 is supported by the register.
34586 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34588 * lra-constraints.c (simplify_operand_subreg): Remove early
34591 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34594 * lra-constraints.c (curr_insn_transform): Tighten condition
34595 for converting SUBREG reloads from OP_OUT to OP_INOUT.
34597 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
34600 * lra-constraints.c (curr_insn_transform): Handle
34601 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
34603 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
34606 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
34608 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
34610 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
34613 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
34616 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34618 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
34621 2017-02-18 Jakub Jelinek <jakub@redhat.com>
34623 * final.c (last_columnnum, override_columnnum): New variables.
34624 (final_start_function): Set last_columnnum, pass it to begin_prologue
34625 hook and pass 0 to dwarf2out_begin_prologue.
34626 (final_scan_insn): Update override_columnnum. Pass last_columnnum
34627 to source_line debug hook.
34628 (notice_source_line): Compute last_columnnum and for debug_column_info
34629 return true on column changes.
34630 * debug.h (struct gcc_debug_hooks): Add column argument to
34631 source_line and begin_prologue hooks.
34632 (debug_nothing_int_charstar_int_bool): Remove prototype.
34633 (debug_nothing_int_int_charstar,
34634 debug_nothing_int_int_charstar_int_bool): New prototypes.
34635 (dwarf2out_begin_prologue): Add column argument.
34636 * debug.c (do_nothing_debug_hooks): Adjust source_line and
34637 begin_prologue hooks.
34638 (debug_nothing_int_charstar_int_bool): Remove.
34639 (debug_nothing_int_int_charstar,
34640 debug_nothing_int_int_charstar_int_bool): New functions.
34641 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
34642 through to dwarf2out_source_line.
34643 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
34644 (dwarf2out_source_line): Add column argument, emit it if requested.
34645 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
34647 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34648 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
34649 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
34650 through to dwarf2out_begin_prologue.
34651 (vmsdbgout_source_line): Add column argument, pass it through to
34652 dwarf2out_source_line.
34653 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
34654 dbxout_source_line caller.
34655 (dbxout_source_line): Add column argument.
34657 * common.opt (gno-column-info, gcolumn-info): New options.
34658 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
34659 (check_die): Also test for multiple DW_AT_decl_column attributes.
34660 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
34661 DW_AT_decl_column if requested.
34662 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
34664 (gen_variable_die): Likewise.
34665 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
34666 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
34669 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
34670 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
34671 (ix86_handle_option): Handle OPT_m3dnowa.
34672 * doc/invoke.texi (-m3dnowa): Document.
34673 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
34674 -m3dnowa instead of -m3dnow -march=athlon.
34677 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
34678 instead of gcc_assert for K, r and R code checks. Formatting fixes.
34680 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34683 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
34684 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
34685 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
34686 generator for vsx_xxpermdi_<mode>_be.
34687 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
34688 force big-endian semantics.
34689 (vsx_xxpermdi_<mode>_be): New define_expand with same
34690 implementation as previous version of vsx_xxpermdi_<mode>.
34692 2017-02-17 Jakub Jelinek <jakub@redhat.com>
34694 PR tree-optimization/79327
34695 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
34696 variable, its initialization and use.
34698 2017-02-17 Julia Koval <julia.koval@intel.com>
34700 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
34701 (OPTION_MASK_ISA_PKU_UNSET): New.
34702 (ix86_handle_option): Handle -mrdpid.
34703 * config/i386/cpuid.h (bit_RDPID): New.
34704 * config/i386/driver-i386.c (host_detect_local_cpu):
34705 Detect RDPID feature.
34706 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
34707 * config/i386/i386-c.c (ix86_target_macros_internal):
34709 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
34710 (ix86_valid_target_attribute_inner_p): Add "rdpid".
34711 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
34712 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
34713 * config/i386/i386.md (define_insn "rdpid"): New.
34714 * config/i386/i386.opt Add -mrdpid.
34715 * config/i386/immintrin.h (_rdpid_u32): New.
34717 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
34719 PR rtl-optimization/79541
34720 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
34721 instead of transforming it into USE.
34723 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
34725 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
34726 If HONOR_SNANS (SFmode) force the input to a register.
34727 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
34728 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
34729 an frsp or similar insn.
34731 2017-02-17 Martin Liska <mliska@suse.cz>
34733 PR rtl-optimization/79577
34734 * params.def (selsched-max-sched-times): Increase minimum to 1.
34736 2017-02-17 Martin Liska <mliska@suse.cz>
34738 PR rtl-optimization/79574
34739 * gcse.c (want_to_gcse_p): Prevent integer overflow.
34741 2017-02-17 Martin Liska <mliska@suse.cz>
34743 PR tree-optimization/79529
34744 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
34745 ssa_defined_default_def_p to handle cases which are implicitly
34747 * tree-ssa.c (ssa_defined_default_def_p): New function.
34748 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
34749 which are implicitly defined.
34750 * tree-ssa.h (ssa_defined_default_def_p): Declare.
34752 2017-02-17 Richard Biener <rguenther@suse.de>
34754 PR middle-end/79576
34755 * params.def (max-ssa-name-query-depth): Limit to 10.
34757 2017-02-17 Richard Biener <rguenther@suse.de>
34759 PR tree-optimization/79552
34760 * tree-ssa-structalias.c (visit_loadstore): Properly verify
34763 2017-02-17 Richard Biener <rguenther@suse.de>
34766 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
34768 2017-02-17 Marek Polacek <polacek@redhat.com>
34770 PR middle-end/79536
34771 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
34772 (fold_negate_expr): New wrapper.
34774 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
34776 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
34777 Correct terminology and de-emphasize pre-standard behavior.
34779 2017-02-16 Alan Modra <amodra@gmail.com>
34781 PR rtl-optimization/79286
34782 * ira.c (def_dominates_uses): New function.
34783 (update_equiv_regs): Don't create an equivalence for insns that
34784 may trap where the register def does not dominate the use.
34786 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
34788 PR rtl-optimization/78127
34789 * lra.c (lra): Call lra_eliminate before finish the loop after
34792 2017-02-16 Richard Biener <rguenther@suse.de>
34794 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
34796 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
34797 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
34798 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
34799 (isl_val_int_from_wi): New function.
34800 (extract_affine_gmp): Rename to ...
34801 (extract_affine_wi): ... this, take a widest_int.
34802 (extract_affine_int): Just wrap extract_affine_wi.
34803 (add_param_constraints): Use isl_val_int_from_wi.
34804 (add_loop_constraints): Likewise, and extract_affine_wi.
34806 2017-02-15 Jeff Law <law@redhat.com>
34808 PR middle-end/79521
34809 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
34810 ira_init_register_move_cost_if_necessary.
34812 2017-02-15 Martin Sebor <msebor@redhat.com>
34814 PR middle-end/32003
34815 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
34816 removed in a prior commit.
34818 2017-02-15 Bin Cheng <bin.cheng@arm.com>
34820 PR tree-optimization/79347
34821 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
34822 counters during peeling.
34824 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
34826 * Makefile.in (site.exp): Remove "set ISLVER".
34828 2017-02-15 Jakub Jelinek <jakub@redhat.com>
34831 * real.c (real_from_integer): Call real_convert even for decimal.
34833 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
34836 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
34838 2017-02-14 Andrew Pinski <apinski@cavium.com>
34840 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
34841 cores and change the partno/implementer to be correct.
34842 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
34843 the 'B" as the implementer.
34844 * config/aarch64/aarch64-tune.md: Regenerate.
34846 2017-02-14 Carl Love <cel@us.ibm.com>
34848 * config/rs6000/rs6000.c: Add case statement entry to make the
34849 xvcvuxdsp built-in argument unsigned.
34850 * config/rs6000/vsx.md: Fix the source and return operand types so they
34851 match the instruction definitions from the ISA document. Fix typo
34852 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
34855 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
34858 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
34859 member early_clobber_alts.
34860 * lra-lives.c (reg_early_clobber_p): New.
34861 (process_bb_lives): Use it.
34862 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
34863 (debug_operand_data): Initialize early_clobber_alts.
34864 (setup_operand_alternative): Set up early_clobber_alts.
34865 (collect_non_operand_hard_regs): Ditto. Pass early clobber
34866 alternatives to new_insn_reg.
34867 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
34869 (lra_update_insn_regno_info): Pass the new arg.
34871 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34873 PR middle-end/79505
34874 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
34875 (new_oacc_loop_raw): Don't clear already cleared fields.
34878 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
34879 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
34880 _mm512_prefetch_i64gather_ps): New inline functions and macros.
34882 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
34885 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34887 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34890 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34891 the extra instruction to the right place to store 128-bit constant
34894 2017-02-14 Martin Sebor <msebor@redhat.com>
34896 PR middle-end/79448
34897 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34898 warning for strings of unknown length.
34900 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34902 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34904 2017-02-14 Jeff Law <law@redhat.com>
34907 * ira-costs.c (scan_one_insn): Initialize register move costs
34908 for pseudos seen in USE/CLOBBER insns.
34910 PR tree-optimization/79095
34911 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34912 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34913 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34914 if the operands are known to be not equal, then the resulting range
34916 (intersect_ranges): If the new range is ~[0,0] and the old range is
34917 wide, then prefer ~[0,0].
34918 * tree-vrp.c (overflow_comparison_p_1): New function.
34919 (overflow_comparison_p): New function.
34920 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34921 if NAME is used in an overflow test.
34922 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34923 overflow check that can be expressed as an equality test, then adjust
34924 ops to be that equality test.
34926 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34928 * config/s390/s390-builtin-types.def: Remove flags argument.
34929 * config/s390/s390.c (s390_init_builtins): Likewise.
34931 2017-02-14 Martin Liska <mliska@suse.cz>
34933 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34934 vector. Fix trailing white spaces.
34936 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34938 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34941 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34943 PR rtl-optimization/68664
34944 * config/arm/arm.c (arm_sched_can_speculate_insn):
34945 New function. Declare prototype.
34946 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34948 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34950 PR rtl-optimization/68664
34951 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34953 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34955 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34957 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34958 max skip bytes for function, loop and jump.
34960 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34962 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34963 ABS_EXPR for gimple dump.
34965 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34968 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34970 PR tree-optimization/79408
34971 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34972 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34973 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34974 also if rhs1 is INTEGER_CST.
34976 2017-02-14 Richard Biener <rguenther@suse.de>
34978 PR middle-end/79432
34979 * tree-into-ssa.c (insert_phi_nodes): When the function can
34980 have abnormal edges rewrite SSA names with broken use-def
34981 dominance out of SSA and register them for PHI insertion.
34983 2017-02-13 Martin Sebor <msebor@redhat.com>
34985 PR middle-end/79496
34986 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34987 clearing info.nowrite flag when snprintf size argument is a range.
34989 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34991 * cprop.c (cprop_jump): Add missing space in string literal.
34992 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34993 (get_constraint_for_component_ref): Likewise.
34994 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34995 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34996 * lra-constraints.c (process_alt_operands): Likewise.
34997 * ipa-inline.c (inline_small_functions): Likewise.
34998 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34999 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35000 * trans-mem.c (diagnose_tm_1_op): Likewise.
35001 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35002 (grid_parallel_clauses_gridifiable): Likewise.
35004 * config/nvptx/mkoffload.c (process): Add space in between
35007 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35008 "MOD4_SSE_REGS" and "ALL_REGS".
35010 * spellcheck.c (test_data): Add , in between "foo" and "food".
35012 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35015 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35016 boundary crossing check and subsequent code generation agree.
35018 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35020 * config/aarch64/aarch64.c (has_memory_op): Delete.
35021 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35024 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35026 PR rtl-optimization/79388
35027 PR rtl-optimization/79450
35028 * combine.c (distribute_notes): When removing TEM_INSN for which
35029 corresponding dest has last value recorded, invalidate that last
35032 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35034 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35035 of explicit '@'. Add missing assembly comment marker on branch costs
35038 2017-02-13 Nathan Sidwell <nathan@acm.org>
35040 * gengtype-lex.l (<in_struct>): Add '/'.
35042 2017-02-13 Martin Liska <mliska@suse.cz>
35045 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35047 2017-02-13 Richard Biener <rguenther@suse.de>
35049 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35051 * configure: Re-generate.
35052 * config.in: Likewise.
35053 * graphite-dependences.c: Simplify as if
35054 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35055 * graphite-isl-ast-to-gimple.c: Likewise.
35056 * graphite-optimize-isl.c: Likewise.
35057 * graphite-poly.c: Likewise.
35058 * graphite-sese-to-poly.c: Likewise.
35059 * graphite.h: Likewise.
35060 * toplev.c: Include isl/version.h and use isl_version () for
35061 printing the ISL version.
35062 * doc/install.texi: Update ISL requirement.
35064 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35066 * doc/standards.texi (Standards): Update reference to
35069 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35071 * doc/extend.texi (Named Address Spaces): sourceware.org now
35073 * doc/install.texi (Binaries): Ditto.
35076 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35078 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35079 standard terminology "stringize"/"stringizing" throughout.
35080 * doc/cppinternals.texi: Likewise.
35082 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35084 * doc/extend.texi: Fix some spelling mistakes and typos.
35085 * doc/invoke.texi: Likewise.
35087 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35090 * params.def (inline-min-speedup) Change from 10 to 8.
35092 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35094 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35097 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35100 * ipa-inline-analysis.c (get_minimal_bb): New function.
35101 (record_modified): Use it.
35102 (remap_edge_change_prob): Handle also ancestor functions.
35104 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35106 * doc/contrib.texi (Contributors): Remove broken link into
35107 the Mauve CVS repository.
35109 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35111 PR middle-end/79454
35112 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35113 result computation whenever lhs doesn't have vector mode, not
35114 just when it has BLKmode.
35116 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35118 * doc/makefile.texi (profiledbootstrap): Refer to the
35119 installation instructions only in textual form.
35121 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35124 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35126 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35128 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35129 (Specific): Update mingw-w64 reference.
35131 (Specific): Remove broken link to Renesas RX processor.
35133 2017-02-10 Richard Biener <rguenther@suse.de>
35135 * toplev.c (process_options): Do not mention obsolete graphite
35136 options when printing sorry message about missing graphite support.
35137 Mention -floop-nest-optimize.
35139 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35141 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35142 (vtst_p16): Likewise.
35143 (vtstq_p8): Likewise.
35144 (vtstq_p16): Likewise.
35146 (vtstq_p64): Likewise.
35147 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35148 (vset_lane_p64): New.
35149 (vsetq_lane_p64): New.
35151 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35153 PR tree-optimization/79411
35154 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35155 stmt operands are SSA_NAMEs used in abnormal phis.
35156 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35159 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35162 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35165 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35167 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35169 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35171 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35175 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35176 not arbitrary TREE_CONSTANT.
35179 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35180 "omp declare target link" attribute unless is_global_var.
35181 * omp-offload.c (find_link_var_op): Likewise.
35183 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35184 Chung-Lin Tang <cltang@codesourcery.com>
35186 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35188 (gimplify_adjust_omp_clauses): Don't delete TILE.
35189 (gimplify_omp_for): Deal with TILE.
35190 * internal-fn.c (expand_GOACC_TILE): New function.
35191 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35193 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35194 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35196 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35197 avoid DIV for outermost collapse var.
35198 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35199 Remove out of date comments, fix whitespace.
35200 * omp-general.c (omp_extract_for_data): Deal with tiling.
35201 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35202 adjust OLF_DIM_BASE value.
35203 (struct omp_for_data): Add tiling field.
35204 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35205 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35206 for auto loops. Remove default auto determining, moved to
35207 oacc_loop_fixed_partitions.
35208 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35209 stmts, add e_mask field.
35210 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35211 (oacc_thread_numbers): Use oacc_dim_call.
35212 (oacc_xform_tile): New.
35213 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35214 (finish_oacc_loop): Adjust for ifns vector.
35215 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35216 add case for GOACC_TILE fns.
35217 (oacc_loop_xform_loop): Delete.
35218 (oacc_loop_process): Iterate over call list directly, and add
35219 handling for GOACC_TILE fns.
35220 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35222 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35223 vector partitioning to outer loops. Assign 2 partitions to loops
35224 when available. Add TILE handling.
35225 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35226 (execite_oacc_device_lower): Process GOACC_TILE fns,
35227 ignore unknown specs.
35228 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35229 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35230 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35232 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35234 * configure.ac (ACX_BUGURL): Update.
35235 * configure: Regenerate.
35237 2017-02-09 Richard Biener <rguenther@suse.de>
35239 PR tree-optimization/69823
35240 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35241 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35243 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35245 * config/arc/arc-c.def: Add __NPS400__ definition.
35246 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35247 (TARGET_NPS400): Define.
35249 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35251 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
35253 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
35254 pointer, arch_info.
35255 (arc_cpu_types): Fill the arch_info field with a pointer into the
35256 arc_arch_types table.
35257 (arc_selected_cpu): Declare.
35258 * config/arc/arc.c (arc_selected_cpu): Make global.
35259 (arc_selected_arch): Delete.
35260 (arc_base_cpu): Delete.
35261 (arc_override_options): Remove references to deleted variables,
35262 update access to arch information.
35263 (ARC_OPT): Update access to arch information.
35264 (ARC_OPTX): Likewise.
35265 * config/arc/arc.h (arc_base_cpu): Remove declaration.
35266 (TARGET_ARC600): Update access to arch information.
35267 (TARGET_ARC601): Likewise.
35268 (TARGET_ARC700): Likewise.
35269 (TARGET_EM): Likewise.
35270 (TARGET_HS): Likewise.
35271 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
35274 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
35277 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
35278 condition/operands for integer GE/LE/GEU/LEU operations.
35280 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
35282 PR translation/79397
35283 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
35286 2017-02-08 Martin Jambor <mjambor@suse.cz>
35289 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
35290 whether allocation happened.
35291 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
35292 nothing was allocated.
35294 2017-02-08 Jakub Jelinek <jakub@redhat.com>
35296 PR tree-optimization/79408
35297 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
35298 constant, but SSA_NAME with a known integer range, use the minimum
35299 of that range instead of op1 to determine if modulo can be replaced
35300 with its first operand.
35302 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35304 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
35306 2017-02-08 Richard Biener <rguenther@suse.de>
35308 PR tree-optimization/71824
35309 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35310 Check all loops contained in the merged region.
35312 2017-02-07 Andrew Pinski <apinski@cavium.com>
35314 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
35316 2017-02-07 Andrew Pinski <apinski@cavium.com>
35318 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
35319 (thunderxt88): Likewise.
35320 (thunderxt81): Disable LSE and change v8.1 to v8.
35321 (thunderxt83): Likewise.
35323 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35324 Richard Biener <rguenther@suse.de>
35326 PR middle-end/79399
35327 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
35328 type from int to size_t.
35329 * ira-costs.c (struct_costs_size): Change type from int to size_t.
35331 2017-02-07 Jakub Jelinek <jakub@redhat.com>
35333 PR rtl-optimization/79386
35334 * cprop.c (bypass_conditional_jumps): Initialize
35335 bypass_last_basic_block already before splitting bbs after
35336 unconditional traps...
35337 (bypass_conditional_jumps): ... rather than here.
35340 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
35341 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
35342 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
35343 fix -masm=intel patterns.
35345 2017-02-07 Richard Biener <rguenther@suse.de>
35347 PR tree-optimization/79256
35348 PR middle-end/79278
35349 * builtins.c (get_object_alignment_2): Use min_align_of_type
35350 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
35351 and ADJUST_FIELD_ALIGN.
35353 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
35355 * doc/tm.texi: Regenerate.
35356 * stor-layout.c (layout_decl): Adjust.
35357 (update_alignment_for_field): Likewise.
35358 (place_field): Likewise.
35359 (min_align_of_type): Likewise.
35360 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
35361 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
35362 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
35363 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
35364 * config/frv/frv.c (frv_adjust_field_align): Likewise.
35365 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
35366 * config/i386/i386.c (x86_field_alignment): Likewise.
35367 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
35368 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
35369 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
35370 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
35371 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
35372 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
35376 2017-01-30 Richard Biener <rguenther@suse.de>
35378 PR tree-optimization/79256
35379 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35380 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35383 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
35385 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
35386 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
35387 builtins to SImode and emit a zero-extend, if necessary.
35389 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35391 * docs/invoke.texi (RISC-V Options): Alphabetize.
35393 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35395 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
35398 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
35400 * config/riscv/riscv.c: New file.
35401 * common/config/riscv/riscv-common.c: Likewise.
35402 * config.gcc: Likewise.
35403 * config/riscv/constraints.md: Likewise.
35404 * config/riscv/elf.h: Likewise.
35405 * config/riscv/generic.md: Likewise.
35406 * config/riscv/linux.h: Likewise.
35407 * config/riscv/multilib-generator: Likewise.
35408 * config/riscv/peephole.md: Likewise.
35409 * config/riscv/pic.md: Likewise.
35410 * config/riscv/predicates.md: Likewise.
35411 * config/riscv/riscv-builtins.c: Likewise.
35412 * config/riscv/riscv-c.c: Likewise.
35413 * config/riscv/riscv-ftypes.def: Likewise.
35414 * config/riscv/riscv-modes.def: Likewise.
35415 * config/riscv/riscv-opts.h: Likewise.
35416 * config/riscv/riscv-protos.h: Likewise.
35417 * config/riscv/riscv.h: Likewise.
35418 * config/riscv/riscv.md: Likewise.
35419 * config/riscv/riscv.opt: Likewise.
35420 * config/riscv/sync.md: Likewise.
35421 * config/riscv/t-elf-multilib: Likewise.
35422 * config/riscv/t-linux: Likewise.
35423 * config/riscv/t-linux-multilib: Likewise.
35424 * config/riscv/t-riscv: Likewise.
35425 * configure.ac: Likewise.
35426 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
35427 Waterman as RISC-V maintainers.
35428 * doc/install.texi: Add RISC-V entries.
35429 * doc/invoke.texi: Add RISC-V options section.
35430 * doc/md.texi: Add RISC-V constraints section.
35431 * configure: Regenerated.
35433 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
35436 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
35437 false values to be constant vectors with all 0 or all 1 bits set.
35438 (vcondu<mode><mode>): Likewise.
35439 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
35441 (fpmask_comparison_operator): Update comment.
35442 (vecint_comparison_operator): New predicate.
35443 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
35444 vector conditionals when the true and false values are constant
35445 vectors with all 0 bits or all 1 bits set.
35447 2017-02-06 Martin Sebor <msebor@redhat.com>
35449 PR tree-optimization/79376
35450 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
35452 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
35454 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
35455 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
35456 to simplify split condition.
35458 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35460 * omp-expand.c (oxpand_omp_atomic_fetch_op,
35461 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
35464 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
35466 PR rtl-optimization/68664
35467 * target.def (can_speculate_insn): New hook.
35468 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
35469 * doc/tm.texi: Regenerate.
35470 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
35471 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
35472 (rs6000_sched_can_speculate_insn): New function.
35474 2017-02-06 Jakub Jelinek <jakub@redhat.com>
35476 PR tree-optimization/79284
35477 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
35478 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
35479 vectorizable_mask_load_store, vectorizable_operation,
35480 vect_is_simple_cond, get_same_sized_vectype): Use it instead
35481 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
35482 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
35483 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
35484 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
35485 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
35486 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
35487 is_gimple_assign (stmt). Replace another such test with
35488 is_gimple_assign (stmt).
35490 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35493 * config/avr/avr.c (rtl-iter.h): Include it.
35494 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
35495 (avr_legitimate_combined_insn): ...and implementation.
35497 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
35499 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
35500 * config/s390/s390.c (s390_const_operand_ok)
35501 (s390_canonicalize_comparison, s390_extract_part)
35502 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
35503 (s390_contiguous_bitmask_p, s390_rtx_costs)
35504 (legitimize_pic_address): Likewise.
35505 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
35506 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
35507 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
35508 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
35509 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
35511 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
35513 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
35514 REGNO($0) == REGNO($1).
35516 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35518 * config/s390/linux.h(SIZE_TYPE): Add comment.
35520 2017-02-06 Julian Brown <julian@codesourcery.com>
35521 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35522 Virendra Pathak <virendra.pathak@broadcom.com>
35524 * config/aarch64/aarch64-cores.def: Change the scheduler
35526 * config/aarch64/aarch64.md: Include thunderx2t99.md.
35527 * config/aarch64/thunderx2t99.md: New file.
35529 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35531 * doc/standards.texi (Go Language): Update link to language
35534 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35536 * tree-eh.c (lower_resx): Sanitize profile.
35537 (cleanup_empty_eh_move_lp): Likewise.
35539 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
35542 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
35544 * cfgloopmanip.h (loop_version): Update prototype.
35545 * modulo-sched.c (sms_schedule): Update call of loop_version.
35546 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
35547 * tree-parloops.c (gen_parallel_loop): Likewise.
35548 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
35549 * tree-ssa-loop-split.c (split_loop): Likewise.
35550 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
35551 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
35553 2017-02-05 Martin Liska <mliska@suse.cz>
35556 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
35558 (print_operand_address): Initialize a struct to zero.
35560 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35562 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
35563 garbage collector only in textual form.
35565 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
35567 * doc/extend.texi (x86 specific memory model extensions for
35568 transactional memory): Simplify a phrase.
35570 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
35573 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
35574 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
35575 (atomic_storedi_1): Likewise.
35577 2017-02-04 Jakub Jelinek <jakub@redhat.com>
35579 PR tree-optimization/79338
35580 * tree-parloops.c (gather_scalar_reductions): Don't call
35581 vect_analyze_loop_form for loop->inner before destroying loop's
35584 2017-02-03 Martin Sebor <msebor@redhat.com>
35586 PR tree-optimization/79327
35587 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
35588 when precision has resulted in leading zeros.
35589 (format_integer): Adjust the likely counter to assume an unknown
35590 argument that may be zero is non-zero.
35592 2017-02-03 Jason Merrill <jason@redhat.com>
35595 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
35596 avoid copying non-taken branch.
35598 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35600 PR tree-optimization/79340
35601 * tree-vect-loop.c (vectorizable_reduction): Release
35602 vec_defs elements after safe_splicing them into other vectors.
35605 PR tree-optimization/79327
35606 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
35607 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
35609 (format_integer): Use wide_int_to_tree instead of build_int_cst
35610 + to_?hwi. If argmin is NULL, just set argmin and argmax to
35611 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
35612 of shortest and longest sequence.
35614 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
35616 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
35617 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
35619 2017-02-03 Walter Lee <walt@tilera.com>
35622 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
35623 after initial stackframe link reg save.
35624 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
35626 2017-02-03 Jakub Jelinek <jakub@redhat.com>
35629 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
35630 wu for stxssp alternative.
35632 2017-02-03 Martin Sebor <msebor@redhat.com>
35634 PR tree-optimization/79352
35635 * gimple-fold.c (get_range_strlen): Add argument.
35636 (get_range_strlen): Change return type to bool.
35637 (get_maxval_strlen): Pass in a dummy argument.
35638 * gimple-fold.h (get_range_strlen): Change return type to bool.
35639 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
35640 * tree.h (array_at_struct_end_p): Add argument.
35641 * tree.c (array_at_struct_end_p): Handle it.
35643 2017-02-03 Martin Liska <mliska@suse.cz>
35646 * multiple_target.c (create_dispatcher_calls): Redirect edge
35647 from a caller of a dispatcher.
35648 (expand_target_clones): Make the clones local.
35649 (ipa_target_clone): Do both target clones and resolvers.
35650 (ipa_dispatcher_calls): Remove the pass.
35651 (pass_dispatcher_calls::gate): Likewise.
35652 (make_pass_dispatcher_calls): Likewise.
35653 * passes.def (pass_target_clone): Put as very first IPA early
35656 2017-02-03 Martin Liska <mliska@suse.cz>
35658 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
35659 in case of a function with ifunc attribute.
35661 2017-02-03 Martin Liska <mliska@suse.cz>
35663 * cgraph.c (cgraph_node::dump): Dump function version info.
35664 * symtab.c (symtab_node::dump_base): Add missing new line.
35666 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35668 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
35669 (ifcombine_ifandif): Use it.
35671 2017-02-03 Martin Liska <mliska@suse.cz>
35673 * doc/invoke.texi: Document default value for
35674 use-after-scope-direct-emission-threshold.
35676 2017-02-03 Martin Liska <mliska@suse.cz>
35678 PR tree-optimization/79339
35679 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
35680 (format_floating): Likewise.
35682 2017-02-03 Martin Liska <mliska@suse.cz>
35685 * ipa-prop.c (ipa_node_params_t::insert): Remove current
35687 (ipa_node_params_t::remove): Likewise.
35688 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
35689 initialization from removed ipa_node_params_t::insert.
35690 (ipa_node_params::~ipa_node_params): Move from removed
35691 ipa_node_params_t::release.
35692 * symbol-summary.h (symbol_summary::m_released): New member.
35693 Do not release a summary twice. Do not allow to call finalizer
35694 for types of a summary that live in GGC memory.
35696 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
35698 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
35701 2017-02-02 Martin Sebor <msebor@redhat.com>
35703 PR middle-end/79275
35704 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
35705 (format_string): Tighten up the range of output for non-constant
35706 strings and correct the expected range for wide non-constant strings.
35708 2017-02-02 Martin Sebor <msebor@redhat.com>
35710 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
35712 PR middle-end/32003
35713 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
35715 (-fdump-tree-@var): Add to index and document how to come up
35716 with pass-specific option and dump file names.
35717 (-fdump-passes): Clarify where to look for output.
35719 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
35721 PR middle-end/77445
35722 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
35723 statistics of the analyzed path; allow threading for speed when
35724 any of BBs along the path are optimized for speed.
35726 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
35728 PR middle-end/78468
35729 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
35730 settings of the virtual registers.
35733 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
35735 * explow.c (get_dynamic_stack_size): Take known alignment of stack
35736 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
35739 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35741 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
35742 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
35744 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35746 * config/s390/s390.md: Add missing comments with the expanded
35748 * config/s390/vector.md: Likewise.
35749 * config/s390/vx-builtins.md: Likewise.
35751 2017-02-02 Jakub Jelinek <jakub@redhat.com>
35754 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
35755 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
35756 conditions on a single line.
35758 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35760 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35761 __S390_VX__ to __VX__.
35763 2017-02-01 Andrew Pinski <apinski@cavium.com>
35765 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
35766 stmt_info to record_stmt_cost.
35767 (vect_get_known_peeling_cost): Pass stmt_info if known to
35769 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
35770 cpu_vector_cost field into
35771 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
35772 field into vec_int_stmt_cost and vec_fp_stmt_cost.
35773 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
35774 splitting of scalar_stmt_cost and vec_stmt_cost.
35775 (thunderx_vector_cost): Likewise.
35776 (cortexa57_vector_cost): LIkewise.
35777 (exynosm1_vector_cost): Likewise.
35778 (xgene1_vector_cost): Likewise.
35779 (thunderx2t99_vector_cost): Improve after the splitting of the two
35781 (aarch64_builtin_vectorization_cost): Update for the splitting of
35782 scalar_stmt_cost and vec_stmt_cost.
35784 2017-02-01 Torvald Riegel <triegel@redhat.com>
35785 Richard Henderson <rth@redhat.com>
35787 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
35788 conditional on existance of a fast atomic load.
35789 * optabs-query.c (can_atomic_load_p): New function.
35790 * optabs-query.h (can_atomic_load_p): Declare it.
35791 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
35792 no fast atomic load is available for the particular size of access.
35793 (expand_atomic_compare_and_swap): Likewise.
35794 (expand_atomic_load): Likewise.
35795 (expand_atomic_store): Likewise.
35796 (expand_atomic_fetch_op): Likewise.
35797 * testsuite/lib/target-supports.exp
35798 (check_effective_target_sync_int_128): Remove x86 because it provides
35799 no fast atomic load.
35800 (check_effective_target_sync_int_128_runtime): Likewise.
35802 2017-02-01 Richard Biener <rguenther@suse.de>
35804 * graphite.c: Include tree-vectorizer.h for find_loop_location.
35805 (graphite_transform_loops): Provide opt-info for optimized nests.
35806 * tree-parloop.c (parallelize_loops): Provide opt-info for
35807 parallelized loops.
35809 2017-02-01 Richard Biener <rguenther@suse.de>
35811 PR middle-end/79315
35812 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
35813 was not set before.
35815 2017-02-01 Richard Biener <rguenther@suse.de>
35817 PR tree-optimization/71824
35818 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
35819 Verify the loops are valid in the merged SESE region.
35820 (scop_detection::can_represent_loop_1): Check analyzing the
35821 evolution of the number of iterations in the region succeeds.
35823 2017-01-31 Ian Lance Taylor <iant@golang.org>
35825 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
35826 REG_ARGS_SIZE note to 32-bit push insns and call insn.
35828 2017-01-31 David Malcolm <dmalcolm@redhat.com>
35830 PR preprocessor/79210
35831 * input.c (get_substring_ranges_for_loc): Replace line_width
35832 assertion with error-handling.
35834 2017-01-31 Richard Biener <rguenther@suse.de>
35836 PR tree-optimization/77318
35837 * graphite-sese-to-poly.c (extract_affine): Fix assert.
35838 (create_pw_aff_from_tree): Take loop parameter.
35839 (add_condition_to_pbb): Pass loop of the condition to
35840 create_pw_aff_from_tree.
35842 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35844 * config/s390/s390.c (s390_asan_shadow_offset): New function.
35845 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
35847 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
35851 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
35853 (convert_int_to_float128): Likewise.
35854 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
35855 (convert_int_to_float128): Likewise.
35856 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
35857 (UNSPEC_IEEE128_CONVERT): Likewise.
35858 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
35859 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
35860 Use local variables for IBM extended format.
35861 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
35862 (fix_trunc<mode>si2_fprs): Likewise.
35863 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
35864 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
35865 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
35866 to know that we can now have integers of all sizes in vector
35868 (fix<uns>_<mode>di2_hw): Likewise.
35869 (float<uns>_<mode>si2_hw): Likewise.
35870 (fix_<mode>si2_hw): Likewise.
35871 (fixuns_<mode>si2_hw): Likewise.
35872 (float<uns>_<mode>di2_hw): Likewise.
35873 (float_<mode>di2_hw): Likewise.
35874 (float_<mode>si2_hw): Likewise.
35875 (floatuns_<mode>di2_hw): Likewise.
35876 (floatuns_<mode>si2_hw): Likewise.
35877 (xscvqp<su>wz_<mode>): Delete, no longer used.
35878 (xscvqp<su>dz_<mode>): Likewise.
35879 (xscv<su>dqp_<mode>): Likewise.
35880 (ieee128_mfvsrd_64bit): Likewise.
35881 (ieee128_mfvsrd_32bit): Likewise.
35882 (ieee128_mfvsrwz): Likewise.
35883 (ieee128_mtvsrw): Likewise.
35884 (ieee128_mtvsrd_64bit): Likewise.
35885 (ieee128_mtvsrd_32bit): Likewise.
35887 2017-01-31 Martin Liska <mliska@suse.cz>
35890 * ipa-prop.c (ipa_free_all_node_params): Call release method
35891 instead of ~sumbol_summary to not to trigger double times
35894 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35896 PR tree-optimization/71691
35897 * bitmap.h (class auto_bitmap): New.
35898 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35899 is_maybe_undefined instead of ssa_undefined_value_p.
35901 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35903 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35904 __S390_ARCH_LEVEL__ to __ARCH__.
35906 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35908 PR tree-optimization/79267
35909 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35910 if should_remove_lhs_p is true.
35912 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35915 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35916 (add_alignment_attribute): New.
35917 (base_type_die): Add alignment attribute.
35918 (subrange_type_die): Likewise.
35919 (modified_type_die): Likewise.
35920 (gen_array_type_die): Likewise.
35921 (gen_descr_array_type_die: Likewise.
35922 (gen_enumeration_type_die): Likewise.
35923 (gen_subprogram_die): Likewise.
35924 (gen_variable_die): Likewise.
35925 (gen_field_die): Likewise.
35926 (gen_ptr_to_mbr_type_die): Likewise.
35927 (gen_struct_or_union_type_die): Likewise.
35928 (gen_subroutine_type_die): Likewise.
35929 (gen_typedef_die): Likewise.
35930 (base_type_cmp): Compare alignment attribute.
35932 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35935 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35936 (setb_unsigned) New pattern for setb with CCUNS.
35937 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35938 subfc./subfe sequence to avoid overflow problems. Generate a
35939 shorter sequence with cmpld/setb for power9.
35940 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35941 for generating subfc. instruction.
35942 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35943 now uses this instruction.
35945 2017-01-30 Ian Lance Taylor <iant@google.com>
35948 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35949 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35951 2017-01-30 Martin Sebor <msebor@redhat.com>
35953 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35954 Move constant to the right of a relational operator.
35955 (get_mpfr_format_length, format_character, format_string): Ditto.
35956 (should_warn_p, maybe_warn): Same.
35958 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35960 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35963 * asan.c (get_translation_unit_decl): Remove function.
35964 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35966 2017-01-30 Martin Liska <mliska@suse.cz>
35968 PR gcov-profile/79259
35969 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35970 -fprofile-generate.
35972 2017-01-30 Martin Liska <mliska@suse.cz>
35975 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35976 Initialize variables with NULL value.
35978 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35981 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35983 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35985 2017-01-30 Richard Biener <rguenther@suse.de>
35987 PR tree-optimization/79276
35988 * tree-vrp.c (process_assert_insertions): Properly adjust common
35989 when removing a duplicate.
35991 2017-01-30 Richard Biener <rguenther@suse.de>
35993 PR tree-optimization/79256
35994 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35995 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35997 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35999 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36002 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36003 ("*r<noxa>sbg_<mode>_sll_bitmask")
36004 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36005 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36006 Use contiguous_bitmask_nowrap_operand.
36008 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36011 * config/rs6000/altivec.h (vec_xl): Revise #define.
36012 (vec_xst): Likewise.
36014 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36016 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36018 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36020 PR rtl-optimization/79194
36021 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36022 traps before call to bypass_conditional_jumps.
36024 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36026 PR tree-optimization/71374
36027 * lra-constraints.c (check_conflict_input_operands): New.
36028 (match_reload): Use it.
36030 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36033 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36034 account to calculate conflict_set.
36036 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36038 PR rtl-optimization/78559
36039 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36040 other_insn in combine.
36042 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36044 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36045 uint16_type_node for BT_UINT16.
36047 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36049 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36050 "RTL Tests" to menu.
36051 (GIMPLE Tests): New node.
36052 (RTL Tests): New node.
36054 2017-01-27 Richard Biener <rguenther@suse.de>
36056 PR tree-optimization/79245
36057 * tree-loop-distribution.c (distribute_loop): Apply cost
36058 modeling also to detected patterns.
36060 2017-01-27 Richard Biener <rguenther@suse.de>
36062 PR tree-optimization/71433
36063 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36064 (compare_assert_loc): New function.
36065 (process_assert_insertions): Sort and optimize assert locations
36066 to remove duplicates and push down identical assertions on
36067 edges to their destination block.
36069 2017-01-27 Richard Biener <rguenther@suse.de>
36071 PR tree-optimization/79244
36072 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36073 out SSA names even if abnormal.
36075 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36077 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36078 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36079 instead of MPFR_RNDN.
36081 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36084 * arm.c (arm_option_override): Don't call build_target_option_node
36085 until after doing all option overrides.
36086 (arm_valid_target_attribute_tree): Likewise.
36088 2017-01-27 Martin Liska <mliska@suse.cz>
36090 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36091 for {cd}tors and C++ {cd}tors.
36093 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36095 * config/s390/s390.md ("*setmem_long_and")
36096 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36098 2017-01-26 Martin Sebor <msebor@redhat.com>
36100 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36103 2017-01-26 Martin Sebor <msebor@redhat.com>
36105 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36106 HAVE_DFmode before using XFmode or DFmode.
36107 (parse_directive): Avoid using the z length modifier to avoid
36108 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36110 PR middle-end/78703
36111 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36112 to accept adjustment as an array.
36113 (get_int_range): New function.
36114 (struct directive): Make width and prec arrays.
36115 (directive::set_width, directive::set_precision): Call get_int_range.
36116 (format_integer, format_floating): Handle width and precision ranges.
36117 (format_string, parse_directive): Same.
36119 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36122 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36123 comdat_type_p set, just clone them, but keep the children in the
36127 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36128 which have direct callers with -fvar-tracking-assignments enabled
36130 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36131 inside of type units.
36133 2017-01-26 Martin Sebor <msebor@redhat.com>
36135 PR middle-end/78703
36136 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36138 (struct format_result): Replace number_chars, number_chars_min,
36139 and number_chars_max with a single member of struct result_range.
36141 (format_result::operator+=): Adjust.
36142 (struct fmtresult): Remove bounded. Handle likely and unlikely
36144 (fmtresult::adjust_for_width_or_precision): New function.
36145 (fmtresult:type_max_digits): New function.
36146 (bytes_remaining): Handle likely and unlikely counters.
36147 (min_bytes_remaining): Remove.
36148 (format_percent): Simplify.
36149 (format_integer, format_floating): Set likely and unlikely counters.
36150 (get_string_length, format_character, format_string): Same.
36151 (format_plain, should_warn_p): New function.
36152 (maybe_warn): Call should_warn_p. Update diagnostic messages
36153 and handle those for all directives, including plain strings.
36154 (format_directive): Handle likely and unlikely counters.
36155 Remove unnecessary quoting from diagnostics. Add an informational
36157 (add_bytes): Remove.
36158 (pass_sprintf_length::compute_format_length): Simplify.
36159 (try_substitute_return_value): Handle likely and unlikely counters.
36161 2017-01-26 Carl Love <cel@us.ibm.com>
36163 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36164 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36166 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36169 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36170 endianess for subregs into account.
36171 * lra-constraints.c (lra_constraints): Do risky transformations
36172 always on the first iteration.
36173 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36174 last_call_used_reg_set.
36175 (process_bb_lives): Define and use last_call_used_reg_set.
36176 * lra.c (lra): Always continue after lra_constraints on the first
36179 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36181 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36183 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36185 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36187 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36188 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36189 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36190 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36191 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36192 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36193 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36194 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36195 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36197 2017-01-26 Marek Polacek <polacek@redhat.com>
36200 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36201 for the third operand.
36203 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36205 PR middle-end/79236
36206 * omp-low.c (struct omp_context): Add simt_stmt field.
36207 (scan_omp_for): Return omp_context *.
36208 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36209 context to the _simt_ SIMD stmt.
36210 (lower_omp_for): For combined SIMD with sibling _simt_
36211 SIMD, make sure to use the same decls in _looptemp_
36212 clauses as in the sibling.
36214 2017-01-26 David Sherwood <david.sherwood@arm.com>
36216 PR middle-end/79212
36217 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36220 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36223 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36224 emit fld b; fld a; if possible.
36226 * brig-builtins.def: Update copyright years.
36227 * config/arm/arm_acle_builtins.def: Update copyright years.
36229 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36232 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36233 constraint instead of o for the stxsd instruction.
36235 2017-01-25 Carl Love <cel@us.ibm.com>
36237 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36238 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36240 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36242 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36244 2017-01-25 Richard Biener <rguenther@suse.de>
36246 PR tree-optimization/69264
36247 * target.def (vector_alignment_reachable): Improve documentation.
36248 * doc/tm.texi: Regenerate.
36249 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
36251 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
36252 earlier changes with respect to TYPE_USER_ALIGN.
36253 (vector_alignment_reachable_p): Likewise. Improve dumping.
36255 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36258 * config/arm/arm.md (xordi3): Force constant operand into a register
36261 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36263 * doc/invoke.texi (-fstore-merging): Correct default optimization
36264 levels at which it is enabled.
36265 (-O): Move -fstore-merging from list to...
36268 2017-01-25 Richard Biener <rguenther@suse.de>
36271 * omp-expand.c: Include debug.h.
36272 (expand_omp_taskreg): Make sure to generate early debug before
36273 outlining anything from a function.
36274 (expand_omp_target): Likewise.
36275 (grid_expand_target_grid_body): Likewise.
36277 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
36280 * asan.c (get_translation_unit_decl): New function.
36281 (asan_add_global): Extract modules file name from globals
36282 TRANSLATION_UNIT_DECL name.
36284 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
36287 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
36288 for long calls with APCS frame and VFP.
36290 2017-01-24 David Malcolm <dmalcolm@redhat.com>
36292 * cfg.c (original_copy_tables_initialized_p): New function.
36293 * cfg.h (original_copy_tables_initialized_p): New decl.
36294 * cfgrtl.c (relink_block_chain): Guard the call to
36295 free_original_copy_tables with a call to
36296 original_copy_tables_initialized_p.
36297 * cgraph.h (symtab_node::native_rtl_p): New decl.
36298 * cgraphunit.c (symtab_node::native_rtl_p): New function.
36299 (symtab_node::needed_p): Don't assert for early assembly output
36300 for __RTL functions.
36301 (cgraph_node::finalize_function): Set "force_output" for __RTL
36303 (cgraph_node::analyze): Bail out early for __RTL functions.
36304 (analyze_functions): Update assertion to support __RTL functions.
36305 (cgraph_node::expand): Bail out early for __RTL functions.
36306 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
36308 * function.h (struct function): Update comment for field
36310 * gimple-expr.c: Include "tree-pass.h".
36311 (gimple_has_body_p): Return false for __RTL functions.
36312 * Makefile.in (OBJS): Add run-rtl-passes.o.
36313 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
36315 (gcc::pass_manager::get_clean_slate): New accessor.
36316 * passes.c: Include "insn-addr.h".
36317 (should_skip_pass_p): Add logging. Update logic for running
36318 "expand" to be compatible with both __GIMPLE and __RTL. Guard
36319 property-provider override so it is only done for gimple passes.
36321 (skip_pass): New function.
36322 (execute_one_pass): Call skip_pass when skipping passes.
36323 * read-md.c (md_reader::read_char): Support filtering
36324 the input to a subset of line numbers.
36325 (md_reader::md_reader): Initialize fields
36326 m_first_line and m_last_line.
36327 (md_reader::read_file_fragment): New function.
36328 * read-md.h (md_reader::read_file_fragment): New decl.
36329 (md_reader::m_first_line): New field.
36330 (md_reader::m_last_line): New field.
36331 * read-rtl-function.c (function_reader::create_function): Only
36332 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
36333 curr_properties. Set DECL_INITIAL to a dummy block.
36334 (read_rtl_function_body_from_file_range): New function.
36335 * read-rtl-function.h (read_rtl_function_body_from_file_range):
36337 * run-rtl-passes.c: New file.
36338 * run-rtl-passes.h: New file.
36340 2017-01-24 Jeff Law <law@redhat.com>
36342 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
36345 2017-01-24 Bin Cheng <bin.cheng@arm.com>
36347 PR tree-optimization/79159
36348 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
36349 (record_nonwrapping_iv): Improve boundary using above function if no
36350 value range information.
36352 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
36353 Martin Jambor <mjambor@suse.cz>
36355 * brig-builtins.def: New file.
36356 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
36357 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
36358 (DEF_HSAIL_SAT_BUILTIN): Likewise.
36359 (DEF_HSAIL_INTR_BUILTIN): Likewise.
36360 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
36361 * builtin-types.def (BT_INT8): New.
36362 (BT_INT16): Likewise.
36363 (BT_UINT8): Likewise.
36364 (BT_UINT16): Likewise.
36365 (BT_FN_ULONG): Likewise.
36366 (BT_FN_UINT_INT): Likewise.
36367 (BT_FN_UINT_ULONG): Likewise.
36368 (BT_FN_UINT_LONG): Likewise.
36369 (BT_FN_UINT_PTR): Likewise.
36370 (BT_FN_ULONG_PTR): Likewise.
36371 (BT_FN_INT8_FLOAT): Likewise.
36372 (BT_FN_INT16_FLOAT): Likewise.
36373 (BT_FN_UINT32_FLOAT): Likewise.
36374 (BT_FN_UINT16_FLOAT): Likewise.
36375 (BT_FN_UINT8_FLOAT): Likewise.
36376 (BT_FN_UINT64_FLOAT): Likewise.
36377 (BT_FN_UINT16_UINT32): Likewise.
36378 (BT_FN_UINT32_UINT16): Likewise.
36379 (BT_FN_UINT16_UINT16_UINT16): Likewise.
36380 (BT_FN_INT_PTR_INT): Likewise.
36381 (BT_FN_UINT_PTR_UINT): Likewise.
36382 (BT_FN_LONG_PTR_LONG): Likewise.
36383 (BT_FN_ULONG_PTR_ULONG): Likewise.
36384 (BT_FN_VOID_UINT64_UINT64): Likewise.
36385 (BT_FN_UINT8_UINT8_UINT8): Likewise.
36386 (BT_FN_INT8_INT8_INT8): Likewise.
36387 (BT_FN_INT16_INT16_INT16): Likewise.
36388 (BT_FN_INT_INT_INT): Likewise.
36389 (BT_FN_UINT_FLOAT_UINT): Likewise.
36390 (BT_FN_FLOAT_UINT_UINT): Likewise.
36391 (BT_FN_ULONG_UINT_UINT): Likewise.
36392 (BT_FN_ULONG_UINT_PTR): Likewise.
36393 (BT_FN_ULONG_ULONG_ULONG): Likewise.
36394 (BT_FN_UINT_UINT_UINT): Likewise.
36395 (BT_FN_VOID_UINT_PTR): Likewise.
36396 (BT_FN_UINT_UINT_PTR: Likewise.
36397 (BT_FN_UINT32_UINT64_PTR): Likewise.
36398 (BT_FN_INT_INT_UINT_UINT): Likewise.
36399 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
36400 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
36401 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
36402 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
36403 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
36404 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
36405 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
36406 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
36407 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
36408 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
36409 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
36410 * doc/frontends.texi: List BRIG FE.
36411 * doc/install.texi (Testing): Add BRIG tesring requirements.
36412 * doc/invoke.texi (Overall Options): Mention BRIG.
36413 * doc/standards.texi (Standards): Doucment BRIG HSA version.
36415 2017-01-24 Richard Biener <rguenther@suse.de>
36417 PR translation/79208
36418 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
36420 2017-01-24 Martin Jambor <mjambor@suse.cz>
36423 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
36424 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
36425 and known_contexts.
36427 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
36429 PR middle-end/79123
36430 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
36431 casts from signed to unsigned really don't have a range.
36433 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
36435 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
36436 GMP_RNDx for compatiblity.
36438 2017-01-24 Martin Liska <mliska@suse.cz>
36441 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
36442 that would prevent us to call alloca with -1 as argument.
36444 2017-01-24 Jakub Jelinek <jakub@redhat.com>
36446 * dwarf2out.c (output_compilation_unit_header, output_file_names):
36447 Avoid -Wformat-security warning.
36449 2017-01-23 Andrew Pinski <apinski@cavium.com>
36451 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
36454 2017-01-23 Martin Sebor <msebor@redhat.com>
36456 PR middle-end/78703
36457 * gimple-ssa-sprintf.c (warn_level): New global.
36458 (format_integer): Use it here and throughout the rest of the file.
36459 Use the same switch to compute sign as base.
36460 (maybe_warn): New function.
36461 (format_directive): Factor out warnings into maybe_warn.
36462 Add debugging output. Use warn_level.
36463 (add_bytes): Use warn_level.
36464 (pass_sprintf_length::compute_format_length): Add debugging output.
36465 (try_substitute_return_value): Same.
36466 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
36468 PR middle-end/78703
36469 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
36470 (struct fmtresult, format_integer, format_floating): Adjust.
36471 (fmtresult::fmtresult): Set max correctly in two argument ctor.
36472 (get_string_length, format_string,format_directive): Same.
36473 (pass_sprintf_length::compute_format_length): Same.
36474 (try_substitute_return_value): Simplify slightly.
36476 PR middle-end/78703
36477 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
36478 (fmtresult::operator+=): Outlined.
36479 (struct fmtresult): Add ctors.
36480 (struct conversion_spec): Rename...
36481 (struct directive): ...to this. Add and remove data members.
36482 (directive::set_width, directive::set_precision): New functions.
36483 (format_percent): Use fmtresult ctor.
36484 (get_width_and_precision): Remove.
36485 (format_integer): Make naming changes. Avoid computing width and
36487 (format_floating): Same. Adjust indentation.
36488 (format_character, format_none): New functions.
36489 (format_string): Moved character handling to format_character.
36490 (format_directive): Remove arguments, change return type.
36491 (parse_directive): New function.
36492 (pass_sprintf_length::compute_format_length): Move directive
36493 parsing to parse_directive.
36495 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36497 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
36498 (assign_assembler_name_if_needed): ... this.
36499 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
36500 (assign_assembler_name_if_needed): ... this.
36501 (free_lang_data_in_cgraph): Adjust callers.
36502 * cgraphunit.c (cgraph_node::analyze): Likewise.
36503 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
36506 2017-01-23 Richard Biener <rguenther@suse.de>
36508 PR tree-optimization/79088
36509 PR tree-optimization/79188
36510 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
36511 resetting loop bounds after last path deletion. Reset loop
36512 bounds of the target loop, make code match the comments.
36513 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
36514 Make sure loops need no fixups.
36516 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36518 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
36519 exponent support with double type for first argument.
36520 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
36521 type returned by __builtin_vec_extract_sig,
36522 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
36523 functions from "vector int" to "vector unsigned int" or from
36524 "vector long long int" to "vector unsigned long long int".
36525 Changed type returned by __builtin_vec_extract_exp,
36526 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
36527 functions from "vector int" to "vector unsigned int" or from
36528 "vector long long int" to "vector unsigned long long int".
36529 Changed return type of __builtin_vec_test_data_class,
36530 __builtin_vec_test_data_class_sp, and
36531 __builtin_vec_test_data_class_dp from "vector int" to
36532 "vector bool int" or from "vector long long int" to "vector bool
36533 long long int" and changed second argument type from "unsigned
36534 int" to "int". Added new overloaded function forms "vector float
36535 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
36536 "vector float __builtin_vec_insert_exp_sp (vector float, vector
36537 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
36538 double, vector unsigned long long int)" and "vector double
36539 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
36540 long int)". Changed return type of
36541 __builtin_scalar_test_data_class and
36542 __builtin_scalar_test_data_class_sp and
36543 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
36544 int" and changed second argument from "unsigned int" to "int".
36545 Changed type returned by __builtin_scalar_test_neg,
36546 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
36547 from "int" to "bool int". Added new overloaded function form
36548 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
36549 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
36550 exponent double-precision with floating point first argument.
36551 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
36552 documentation of scalar_test_data_class, scalar_test_neg,
36553 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
36554 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
36555 vec_test_data_class built-in functions to reflect refinements in
36556 their type signatures.
36558 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
36560 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
36562 (aarch64_elf_asm_destructor): Likewise.
36564 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
36566 PR rtl-optimization/78634
36567 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
36568 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
36569 * ifcvt.c (noce_try_cmove): Add missing cost check.
36571 PR rtl-optimization/71724
36572 * combine.c (if_then_else_cond): Look for situations where it is
36573 beneficial to undo the work of one of the recursive calls.
36575 2017-01-23 Bin Cheng <bin.cheng@arm.com>
36577 PR tree-optimization/70754
36578 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
36579 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
36580 combined stmt before it if not NULL.
36581 (combine_chains): Process refs reversely and compute dominance point
36584 2017-01-23 Martin Liska <mliska@suse.cz>
36586 PR tree-optimization/79196
36587 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
36588 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
36590 (strlen_optimize_stmt): Call the renamed function.
36592 2017-01-23 Michael Matz <matz@suse.de>
36594 PR tree-optimization/78384
36595 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
36597 2017-01-23 Richard Biener <rguenther@suse.de>
36599 PR tree-optimization/79186
36600 * tree-vrp.c (register_new_assert_for): Make sure we've seen
36601 both incoming edges before moving an assert.
36603 2017-01-23 Martin Jambor <mjambor@suse.cz>
36605 * ipa-prop.c (load_from_param_1): Removed.
36606 (load_from_unmodified_param): Bits from load_from_param_1 put back
36608 (load_from_param): Removed.
36609 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
36610 with stmt. Reverted back to use of load_from_unmodified_param.
36612 2017-01-23 Martin Jambor <mjambor@suse.cz>
36615 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
36616 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
36617 field a pointer to garbage collected vector, mark lattices and
36618 ipcp_orig_node with GTY((skip)).
36619 (ipa_get_param_count): Adjust to descriptors being a pointer.
36620 (ipa_get_param): Likewise.
36621 (ipa_get_type): Likewise.
36622 (ipa_get_param_move_cost): Likewise.
36623 (ipa_set_param_used): Likewise.
36624 (ipa_get_controlled_uses): Likewise.
36625 (ipa_set_controlled_uses): Likewise.
36626 (ipa_is_param_used): Likewise.
36627 (ipa_node_params_t): Move into garbage collector. New methods insert
36629 (ipa_node_params_sum): Annotate wth GTY(()).
36630 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
36632 (ipa_load_from_parm_agg): Adjust declaration.
36633 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
36634 * ipa-profile.c (ipa_profile): Likewise.
36635 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
36636 (ipa_populate_param_decls): Make descriptors parameter garbage
36638 (ipa_dump_param): Adjust to descriptors being a pointer.
36639 (ipa_alloc_node_params): Likewise.
36640 (ipa_initialize_node_params): Likewise.
36641 (load_from_param_1): Make descriptors parameter garbage collected.
36642 (load_from_unmodified_param): Likewise.
36643 (load_from_param): Likewise.
36644 (ipa_load_from_parm_agg): Likewise.
36645 (ipa_node_params::~ipa_node_params): Removed.
36646 (ipa_free_all_node_params): Remove call to delete operator.
36647 (ipa_node_params_t::insert): New.
36648 (ipa_node_params_t::remove): Likewise.
36649 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
36650 copy known_csts and known_contexts vectors.
36651 (ipa_read_node_info): Adjust to descriptors being a pointer.
36652 (ipcp_modif_dom_walker): Make m_descriptors field garbage
36654 (ipcp_transform_function): Make descriptors variable garbage
36657 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
36659 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
36660 * config/i386/avx512dqintrin.h: Ditto.
36661 * config/i386/avx512fintrin.h: Ditto.
36662 * config/i386/i386.c: Handle new builtins.
36663 * config/i386/i386-builtin.def: Add new builtins.
36664 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
36665 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
36667 2017-01-23 Jakub Jelinek <jakub@redhat.com>
36668 Martin Liska <mliska@suse.cz>
36670 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
36671 * asan.c (asan_expand_poison_ifn): Support stores and use
36672 appropriate ASAN report function.
36673 * internal-fn.c (expand_ASAN_POISON_USE): New function.
36674 * internal-fn.def (ASAN_POISON_USE): Declare.
36675 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
36676 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
36677 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
36678 ASAN_POISON calls w/o LHS.
36679 * tree-ssa.c (execute_update_addresses_taken): Create clobber
36680 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
36681 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
36682 * gimplify.c (asan_poison_variables): Add attribute
36683 use_after_scope_memory to variables that really needs to live
36685 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
36686 having the attribute.
36688 2017-01-23 Martin Liska <mliska@suse.cz>
36690 * asan.c (create_asan_shadow_var): New function.
36691 (asan_expand_poison_ifn): Likewise.
36692 * asan.h (asan_expand_poison_ifn): New declaration.
36693 * internal-fn.c (expand_ASAN_POISON): Likewise.
36694 * internal-fn.def (ASAN_POISON): New builtin.
36695 * sanopt.c (pass_sanopt::execute): Expand
36696 asan_expand_poison_ifn.
36697 * tree-inline.c (copy_decl_for_dup_finish): Make function
36699 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
36700 * tree-ssa.c (is_asan_mark_p): New function.
36701 (execute_update_addresses_taken): Rewrite local variables
36702 (identified just by use-after-scope as addressable) into SSA.
36704 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36706 * doc/install.texi (Specific): opensource.apple.com uses https
36707 now. Remove trailing slash.
36709 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
36711 * README.Portability: Remove note on an Irix compatibility issue.
36713 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
36715 * gcov.c (INCLUDE_ALGORITHM): Define.
36716 (INCLUDE_VECTOR): Define.
36717 No longer include <vector> and <algorithm> directly.
36719 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
36721 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
36723 * doc/invoke.texi (Code Gen Options): Ditto.
36725 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
36728 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
36730 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
36732 rtl-optimization/79125
36733 * cprop.c (local_cprop_pass): Handle cases where we make an
36734 unconditional trap.
36736 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
36740 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
36741 read from, for big endian.
36743 2017-01-20 Jiong Wang <jiong.wang@arm.com>
36745 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
36746 register pauth builtins for LP64 only.
36748 2017-01-20 Marek Polacek <polacek@redhat.com>
36751 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
36754 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36756 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
36758 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
36759 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
36760 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
36762 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36765 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
36766 in big-endian mode when they are not a single duplicated value.
36768 2017-01-20 Richard Biener <rguenther@suse.de>
36770 * BASE-VER: Bump to 7.0.1.
36772 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
36774 * omp-low.c (omplow_simd_context): New struct. Use it...
36775 (lower_rec_simd_input_clauses): ...here and...
36776 (lower_rec_input_clauses): ...here to hold common data. Adjust all
36777 references to idx, lane, max_vf, is_simt.
36779 2017-01-20 Graham Markall <graham.markall@embecosm.com>
36781 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
36784 2017-01-20 Martin Jambor <mjambor@suse.cz>
36786 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
36787 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
36789 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
36790 (GTFILES): Rename hsa.c to hsa-common.c.
36791 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
36792 * hsa-dump.c: Likewise.
36793 * hsa-gen.c: Likewise.
36794 * hsa-regalloc.c: Likewise.
36795 * ipa-hsa.c: Likewise.
36796 * omp-expand.c: Likewise.
36797 * omp-low.c: Likewise.
36798 * toplev.c: Likewise.
36800 2017-01-20 Marek Polacek <polacek@redhat.com>
36803 * doc/invoke.texi: Document -Wduplicated-branches.
36804 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
36805 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
36806 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
36807 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
36808 return 0 only when not OEP_LEXICOGRAPHIC.
36809 (fold_build_cleanup_point_expr): Use the expression
36810 location when building CLEANUP_POINT_EXPR.
36811 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
36812 * tree.c (add_expr): Handle error_mark_node.
36814 2017-01-20 Martin Liska <mliska@suse.cz>
36817 * tree-profile.c (init_ic_make_global_vars): Do not call
36819 (gimple_init_gcov_profiler): Likewise.
36821 2017-01-20 Martin Liska <mliska@suse.cz>
36824 * cgraph.h (maybe_create_reference): Remove argument and
36826 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
36828 * ipa-cp.c (create_specialized_node): Likewise.
36829 * symtab.c (symtab_node::maybe_create_reference): Handle
36830 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
36832 2017-01-20 Martin Liska <mliska@suse.cz>
36834 * read-rtl-function.c (function_reader::create_function): Use
36835 build_decl instread of build_decl_stat.
36837 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
36839 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
36840 * config/i386/avx512dqintrin.h: Ditto.
36841 * config/i386/avx512fintrin.h: Ditto.
36842 * config/i386/i386-builtin-types.def: Add new types.
36843 * config/i386/i386.c: Handle new types.
36844 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
36845 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
36846 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
36847 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
36848 (__builtin_ia32_kshiftridi): New.
36849 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
36851 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
36855 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
36856 define to rs6000_init_stack_protect_guard.
36857 (rs6000_init_stack_protect_guard): New function.
36859 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36860 Yunqiang Su <yunqiang.su@imgtec.com>
36862 * config.gcc (supported_defaults): Add madd4.
36863 (with_madd4): Add validation.
36864 (all_defaults): Add madd4.
36865 * config/mips/mips.opt (mmadd4): New option.
36866 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36868 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36870 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
36871 (ISA_HAS_FUSED_MADD4): Likewise.
36872 * doc/invoke.texi (-mmadd4): Document the new option.
36873 * doc/install.texi (--with-madd4): Document the new option.
36875 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36877 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
36878 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
36879 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
36880 (aarch64_init_pauth_hint_builtins): New.
36881 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
36882 (aarch64_expand_builtin): Expand new builtins.
36884 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36886 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36887 * combine-stack-adj.c (no_unhandled_cfa): Handle
36888 REG_CFA_TOGGLE_RA_MANGLE.
36889 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36890 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36891 info for return address signing.
36892 (aarch64_expand_epilogue): Likewise.
36894 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36896 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36897 * config/aarch64/aarch64-protos.h
36898 (aarch64_return_address_signing_enabled): New declaration.
36899 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36901 (aarch64_expand_prologue): Sign return address before it's pushed onto
36903 (aarch64_expand_epilogue): Authenticate return address fetched from
36905 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36906 (aarch64_attributes): New function attributes for "sign-return-address".
36907 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36908 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36909 ("*do_return"): Generate combined instructions according to key index.
36910 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36911 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36913 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36914 * config/aarch64/aarch64.opt (msign-return-address=): New.
36915 * doc/extend.texi (AArch64 Function Attributes): Documents
36916 "sign-return-address=".
36917 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36919 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36921 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36922 overall option summary.
36924 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36926 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36927 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36928 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36929 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36931 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36933 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36934 -mpower9-minmax by default for -mcpu=power9.
36935 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36936 128-bit floating point.
36938 2017-01-20 Alan Modra <amodra@gmail.com>
36940 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36941 optimizing for size.
36943 2017-01-20 Alan Modra <amodra@gmail.com>
36946 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36947 for strcmp and strncmp from corresponding builtin decl.
36949 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36951 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36952 instead of i386/rtems-64.h.
36953 * config/i386/rtems-64.h: Remove.
36955 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36959 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36961 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36963 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36965 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36966 Change int to HOST_WIDE_INT.
36967 * config/aarch64/aarch64-protos.h
36968 (aarch64_simd_gen_const_vector_dup): Likewise.
36969 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36971 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36973 * langhooks-def.h (lhd_type_for_size): New decl.
36974 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36975 * langhooks.c (lhd_type_for_size): New function, taken from
36978 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36980 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36981 define_bypass for CR latency.
36982 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36983 (power9-alu2): Add define_bypass for CR latency.
36985 (power9-mul): Update insn latency.
36986 (power9-mul-compare): Update insn latency, bypass latency and remove
36989 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36991 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36993 * config/aarch64/aarch64.md
36994 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36995 aarch64_nopcrelative_literal_loads.
36996 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36998 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37000 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37001 TARGET_LOONGSON_3A.
37002 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37004 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37007 * config.gcc (supported_defaults): Add lxc1-sxc1.
37008 (with_lxc1_sxc1): Add validation.
37009 (all_defaults): Add lxc1-sxc1.
37010 * config/mips/mips.opt (mlxc1-sxc1): New option.
37011 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37013 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37014 __mips_no_lxc1_sxc1.
37015 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37016 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37017 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37019 2017-01-19 Richard Biener <rguenther@suse.de>
37021 PR tree-optimization/72488
37022 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37023 sure to restore SSA info.
37024 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37026 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37028 PR rtl-optimization/79121
37029 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37030 of the inner type when shifting an extended value.
37032 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37035 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37036 interposable aliases.
37038 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37041 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37042 Use the evmergelohi instruction.
37043 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37044 (mov_sitf_e500_subreg8_2_be): Likewise.
37045 (mov_sitf_e500_subreg12_2_le): Likewise.
37046 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37047 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37048 (mov_sitf_e500_subreg8_2_le): Likewise.
37049 (mov_sitf_e500_subreg12_2_be): Likewise.
37051 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37053 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37054 attribute from vecsimple to vecperm.
37055 (altivec_vbpermq2): Likewise.
37057 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37060 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37062 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37063 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37064 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37065 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37066 case where N arg is SIZE_MAX.
37067 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37068 (cmpstrsi): Add pattern.
37070 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37072 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37073 __builtin_vec_revb builtins.
37074 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37075 built-in functions to support generation of the ISA 3.0 XXBR<x>
37076 vector byte reverse instructions.
37077 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37078 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37079 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37080 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37081 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37082 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37083 (P9V_BUILTIN_VEC_REVB): Likewise.
37084 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37085 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37086 (p9_xxbrq_v16qi): Likewise.
37087 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37088 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37089 (p9_xxbrh_v8hi): Likewise.
37090 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37091 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37092 vec_revb built-in functions.
37094 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37096 PR rtl-optimization/78952
37097 * config/i386/i386.md (any_extract): New code iterator.
37098 (*insvqi_2): Use any_extract for source operand.
37099 (*insvqi_3): Use any_shiftrt for source operand.
37101 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37103 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37105 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37107 2017-01-18 Matthias Klose <doko@ubuntu.com>
37109 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37111 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37113 * config/rs6000/altivec.h (vec_bperm): Change #define.
37114 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37115 (altivec_vbpermq2): New define_insn.
37116 (altivec_vbpermd): Likewise.
37117 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37118 function interface.
37119 (VBPERMD): Likewise.
37120 (VBPERM): New polymorphic function interface.
37121 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37122 Add entries for P9V_BUILTIN_VEC_VBPERM.
37123 * doc/extend.texi: Add interfaces for vec_bperm.
37125 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37127 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37128 first letter of error messages.
37129 (s390_resolve_overloaded_builtin): Likewise.
37130 * config/s390/s390.c (s390_expand_builtin): Likewise.
37131 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37132 (s390_valid_target_attribute_inner_p): Likewise.
37133 * config/s390/s390.md ("tabort"): Likewise.
37135 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37137 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37138 (ISA_AVOID_DIV_HILO): New macro.
37139 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37140 (ISA_HAS_DDIV): Likewise.
37142 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37144 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37146 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37148 * doc/invoke.texi (fabi-version): Spelling fix.
37150 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37153 * doc/invoke.texi (fabi-version): Mention mangling fix for
37156 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37159 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37161 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37164 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37165 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37167 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37169 (stack_protect_test): Ditto.
37170 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37171 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37173 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37174 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37175 -mstack-protector-guard-offset=.
37176 (RS/6000 and PowerPC Options): Ditto.
37178 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37180 * config/i386/i386.h (MASK_CLASS_P): New define.
37181 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37182 there are no registers from different register sets also when
37183 mask registers are used. Update function comment.
37184 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37185 to (*k/*r) and (*k/*km) alternatives.
37187 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37189 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37190 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37191 (EH_RETURN_HANDLER_RTX): New define.
37192 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37193 Force frame pointer in EH return functions.
37194 (aarch64_expand_epilogue): Add barrier for eh_return.
37195 (aarch64_final_eh_return_addr): Remove.
37196 (aarch64_eh_return_handler_rtx): New function.
37197 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37199 (aarch64_eh_return_handler_rtx): New prototype.
37201 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37203 * config/rs6000/altivec.h (vec_rlmi): New #define.
37204 (vec_vrlnm): Likewise.
37205 (vec_rlnm): Likewise.
37206 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37207 (UNSPEC_VRLNM): Likewise.
37208 (VIlong): New mode iterator.
37209 (altivec_vrl<VI_char>mi): New define_insn.
37210 (altivec_vrl<VI_char>nm): Likewise.
37211 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37213 (VRLDNM): Likewise.
37214 (RLNM): New polymorphic function entry.
37215 (VRLWMI): New monomorphic function entry.
37216 (VRLDMI): Likewise.
37217 (RLMI): New polymorphic function entry.
37218 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37219 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37220 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37223 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37226 * dwarf2out.c (field_byte_offset): Restore the
37227 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37228 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37229 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37232 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37235 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37237 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37240 * dwarf2out.c (add_data_member_location_attribute): For constant
37241 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37242 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37243 DW_AT_byte_size attributes.
37245 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37247 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37248 after forcing to constant memory when the code model is medium.
37250 2017-01-17 Julia Koval <julia.koval@intel.com>
37253 * config/i386/avx512fintrin.h
37254 (_mm512_i32gather_ps): Change __addr type to void const*.
37255 (_mm512_mask_i32gather_ps): Ditto.
37256 (_mm512_i32gather_pd): Ditto.
37257 (_mm512_mask_i32gather_pd): Ditto.
37258 (_mm512_i64gather_ps): Ditto.
37259 (_mm512_mask_i64gather_ps): Ditto.
37260 (_mm512_i64gather_pd): Ditto.
37261 (_mm512_mask_i64gather_pd): Ditto.
37262 (_mm512_i32gather_epi32): Ditto.
37263 (_mm512_mask_i32gather_epi32): Ditto.
37264 (_mm512_i32gather_epi64): Ditto.
37265 (_mm512_mask_i32gather_epi64): Ditto.
37266 (_mm512_i64gather_epi32): Ditto.
37267 (_mm512_mask_i64gather_epi32): Ditto.
37268 (_mm512_i64gather_epi64): Ditto.
37269 (_mm512_mask_i64gather_epi64): Ditto.
37270 (_mm512_i32scatter_ps): Change __addr type to void*.
37271 (_mm512_mask_i32scatter_ps): Ditto.
37272 (_mm512_i32scatter_pd): Ditto.
37273 (_mm512_mask_i32scatter_pd): Ditto.
37274 (_mm512_i64scatter_ps): Ditto.
37275 (_mm512_mask_i64scatter_ps): Ditto.
37276 (_mm512_i64scatter_pd): Ditto.
37277 (_mm512_mask_i64scatter_pd): Ditto.
37278 (_mm512_i32scatter_epi32): Ditto.
37279 (_mm512_mask_i32scatter_epi32): Ditto.
37280 (_mm512_i32scatter_epi64): Ditto.
37281 (_mm512_mask_i32scatter_epi64): Ditto.
37282 (_mm512_i64scatter_epi32): Ditto.
37283 (_mm512_mask_i64scatter_epi32): Ditto.
37284 (_mm512_i64scatter_epi64): Ditto.
37285 (_mm512_mask_i64scatter_epi64): Ditto.
37286 * config/i386/avx512pfintrin.h
37287 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
37288 (_mm512_mask_prefetch_i32gather_ps): Ditto.
37289 (_mm512_mask_prefetch_i64gather_pd): Ditto.
37290 (_mm512_mask_prefetch_i64gather_ps): Ditto.
37291 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
37292 (_mm512_prefetch_i32scatter_ps): Ditto.
37293 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
37294 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
37295 (_mm512_prefetch_i64scatter_pd): Ditto.
37296 (_mm512_prefetch_i64scatter_ps): Ditto.
37297 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
37298 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
37299 * config/i386/avx512vlintrin.h
37300 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
37301 (_mm_mmask_i32gather_ps): Ditto.
37302 (_mm256_mmask_i32gather_pd): Ditto.
37303 (_mm_mmask_i32gather_pd): Ditto.
37304 (_mm256_mmask_i64gather_ps): Ditto.
37305 (_mm_mmask_i64gather_ps): Ditto.
37306 (_mm256_mmask_i64gather_pd): Ditto.
37307 (_mm_mmask_i64gather_pd): Ditto.
37308 (_mm256_mmask_i32gather_epi32): Ditto.
37309 (_mm_mmask_i32gather_epi32): Ditto.
37310 (_mm256_mmask_i32gather_epi64): Ditto.
37311 (_mm_mmask_i32gather_epi64): Ditto.
37312 (_mm256_mmask_i64gather_epi32): Ditto.
37313 (_mm_mmask_i64gather_epi32): Ditto.
37314 (_mm256_mmask_i64gather_epi64): Ditto.
37315 (_mm_mmask_i64gather_epi64): Ditto.
37316 (_mm256_i32scatter_ps): Change __addr type to void*.
37317 (_mm256_mask_i32scatter_ps): Ditto.
37318 (_mm_i32scatter_ps): Ditto.
37319 (_mm_mask_i32scatter_ps): Ditto.
37320 (_mm256_i32scatter_pd): Ditto.
37321 (_mm256_mask_i32scatter_pd): Ditto.
37322 (_mm_i32scatter_pd): Ditto.
37323 (_mm_mask_i32scatter_pd): Ditto.
37324 (_mm256_i64scatter_ps): Ditto.
37325 (_mm256_mask_i64scatter_ps): Ditto.
37326 (_mm_i64scatter_ps): Ditto.
37327 (_mm_mask_i64scatter_ps): Ditto.
37328 (_mm256_i64scatter_pd): Ditto.
37329 (_mm256_mask_i64scatter_pd): Ditto.
37330 (_mm_i64scatter_pd): Ditto.
37331 (_mm_mask_i64scatter_pd): Ditto.
37332 (_mm256_i32scatter_epi32): Ditto.
37333 (_mm256_mask_i32scatter_epi32): Ditto.
37334 (_mm_i32scatter_epi32): Ditto.
37335 (_mm_mask_i32scatter_epi32): Ditto.
37336 (_mm256_i32scatter_epi64): Ditto.
37337 (_mm256_mask_i32scatter_epi64): Ditto.
37338 (_mm_i32scatter_epi64): Ditto.
37339 (_mm_mask_i32scatter_epi64): Ditto.
37340 (_mm256_i64scatter_epi32): Ditto.
37341 (_mm256_mask_i64scatter_epi32): Ditto.
37342 (_mm_i64scatter_epi32): Ditto.
37343 (_mm_mask_i64scatter_epi32): Ditto.
37344 (_mm256_i64scatter_epi64): Ditto.
37345 (_mm256_mask_i64scatter_epi64): Ditto.
37346 (_mm_i64scatter_epi64): Ditto.
37347 (_mm_mask_i64scatter_epi64): Ditto.
37348 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
37349 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
37350 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
37351 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
37352 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
37353 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
37354 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
37355 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
37356 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
37357 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
37358 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
37359 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
37360 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
37361 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
37362 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
37363 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
37364 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
37365 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
37366 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
37367 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
37368 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
37369 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
37370 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
37371 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
37372 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
37373 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
37374 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
37375 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
37376 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
37377 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
37378 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
37379 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
37380 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
37381 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
37382 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
37383 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
37384 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
37385 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
37386 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
37387 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
37388 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
37389 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
37390 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
37391 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
37392 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
37393 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
37394 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
37395 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
37396 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
37397 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
37398 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
37399 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
37400 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
37401 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
37402 definitions accordingly.
37404 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
37405 Kuan-Lin Chen <kuanlinchentw@gmail.com>
37408 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
37411 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
37414 * ira-conflicts.c (ira_build_conflicts): Update total conflict
37415 hard regs for inner regno.
37417 2017-01-17 Martin Liska <mliska@suse.cz>
37420 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
37421 assumption and add comment.
37423 2017-01-17 Nathan Sidwell <nathan@acm.org>
37425 * ipa-visibility.c (localize_node): New function, broken out of ...
37426 (function_and_variable_visibility): ... here. Call it.
37428 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37430 PR middle-end/77445
37431 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
37432 correctly set frequency of oudgoing edge.
37433 (duplicate_thread_path): Fix profile updating.
37435 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37438 * configure.ac: Add GCC_BASE_VER.
37439 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
37440 version from BASE-VER file.
37441 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
37442 (gcc.o): Depend on $(BASEVER).
37443 * common.opt (dumpfullversion): New option.
37444 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
37445 * doc/invoke.texi: Document -dumpfullversion.
37446 * doc/install.texi: Document --with-gcc-major-version-only.
37447 * configure: Regenerated.
37449 2017-01-17 Richard Biener <rguenther@suse.de>
37451 PR tree-optimization/71433
37452 * tree-vrp.c (register_new_assert_for): Merge same asserts
37453 on all incoming edges.
37454 (process_assert_insertions_for): Handle insertions at the
37457 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
37459 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
37460 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
37462 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
37465 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
37468 2017-01-17 Alan Modra <amodra@gmail.com>
37471 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
37472 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
37473 symbolic stack limit when pic.
37475 2017-01-16 Martin Sebor <msebor@redhat.com>
37477 PR tree-optimization/78608
37478 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
37480 2017-01-16 Jeff Law <law@redhat.com>
37483 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
37484 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
37485 for several include directories that may be relative to sysroot.
37486 * config/i386/x-mingw32 (gplus_includedir): Define.
37487 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
37488 (native_system_includedir): Likewise.
37489 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
37490 override if TARGET_SYSTEM_ROOT is defined.
37491 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
37493 PR tree-optimization/79090
37494 PR tree-optimization/33562
37495 PR tree-optimization/61912
37496 PR tree-optimization/77485
37497 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
37498 and computed trims into the dump file.
37500 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37502 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
37504 2017-01-16 Jakub Jelinek <jakub@redhat.com>
37507 * gimplify.c (gimplify_init_constructor): If want_value and
37508 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
37512 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
37513 sequence. Formatting fixes.
37514 (doloop_optimize): Formatting fixes.
37517 * gcc.c (debug_level_greater_than_spec_func): New function.
37518 (static_spec_functions): Add debug-level-gt spec function.
37519 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
37521 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37522 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
37523 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
37524 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
37525 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
37526 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
37528 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
37530 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
37531 QImode fixups to general and mask registers only.
37533 2017-01-16 Carl Love <cel@us.ibm.com>
37535 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
37536 for built-in functions
37537 vector signed char vec_nabs (vector signed char)
37538 vector signed short vec_nabs (vector signed short)
37539 vector signed int vec_nabs (vector signed int)
37540 vector signed long long vec_nabs (vector signed long long)
37541 vector float vec_nabs (vector float)
37542 vector double vec_nabs (vector double)
37543 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
37545 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
37546 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
37547 * doc/extend.texi: Update the documentation file for the new built-in
37550 2017-01-16 Martin Sebor <msebor@redhat.com>
37552 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
37555 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37557 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
37558 UNSPEC_VSX__XXSPLTD to require special splat handling.
37560 2017-01-16 David Malcolm <dmalcolm@redhat.com>
37563 * system.h: Poison strndup.
37565 2017-01-16 Alan Modra <amodra@gmail.com>
37568 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
37571 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
37573 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
37575 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
37577 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
37578 call recog here. Assert that INSN_CODE (insn) is non-negative.
37580 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37583 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
37585 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
37586 in the currently scheduled RTL fragment.
37588 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
37590 PR rtl-optimization/78751
37591 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
37594 2017-01-14 Jeff Law <law@redhat.com>
37596 PR tree-optimization/79090
37597 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
37598 variable length stores.
37599 (compute_trims): Delete dead assignment to *trim_tail.
37600 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
37603 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
37605 PR rtl-optimization/78626
37606 PR rtl-optimization/78727
37607 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
37608 of a block, and split such blocks after everything else is finished.
37610 2017-01-14 Alan Modra <amodra@gmail.com>
37613 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
37614 target legitimate_combined_insn.
37615 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
37616 (rs6000_legitimate_combined_insn): New function.
37617 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
37619 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
37620 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
37621 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
37623 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
37625 * doc/frontends.texi (G++ and GCC): Remove references to Java.
37627 2017-01-13 Jeff Law <law@redhat.com>
37629 PR tree-optimization/33562
37630 PR tree-optimization/61912
37631 PR tree-optimization/77485
37632 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
37634 (delete_dead_assignment): Likewise.
37635 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
37636 statement to delete_dead_call and delete_dead_assignment.
37638 2017-01-13 David Malcolm <dmalcolm@redhat.com>
37641 * substring-locations.c (format_warning_va): Strengthen case 1 so
37642 that both endpoints of the substring must be within the format
37643 range for just the substring to be printed.
37645 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
37647 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
37648 * config/i386/i386.c (ix86_target_string): Add missing options
37649 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
37650 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
37651 flags_other and ix86_target_other to flags2_other. Display unknown
37653 (ix86_valid_target_attribute_inner_p): Add missing options and
37654 reorder options by implied ISAs, as in ix86_target_string.
37656 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37658 * hash-table.h (hash_table::too_empty_p): New function.
37659 (hash_table::expand): Use it.
37660 (hash_table::traverse): Likewise.
37661 (hash_table::empty_slot): Use sizeof (value_type) instead of
37662 sizeof (PTR) to convert bytes to elements. Shrink the table
37663 if the current size is excessive for the current number of
37666 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
37668 * ira-costs.c (record_reg_classes): Break from the inner loop
37669 early once alt_fail is known to be true. Update outer loop
37670 handling accordingly.
37672 2017-01-13 Jeff Law <law@redhat.com>
37674 * tree-ssa-dse.c (decrement_count): New function.
37675 (increment_start_addr, maybe_trim_memstar_call): Likewise.
37676 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
37677 when we know the partially dead statement is a mem* function.
37679 PR tree-optimization/61912
37680 PR tree-optimization/77485
37681 * tree-ssa-dse.c: Include expr.h.
37682 (maybe_trim_constructor_store): New function.
37683 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
37685 PR tree-optimization/33562
37686 PR tree-optimization/61912
37687 PR tree-optimization/77485
37688 * doc/invoke.texi: Document new dse-max-object-size param.
37689 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
37690 * tree-ssa-dse.c: Include params.h.
37691 (dse_store_status): New enum.
37692 (initialize_ao_ref_for_dse): New, partially extracted from
37694 (valid_ao_ref_for_dse, normalize_ref): New.
37695 (setup_live_bytes_from_ref, compute_trims): Likewise.
37696 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
37697 (maybe_trim_partially_dead_store): Likewise.
37698 (maybe_trim_complex_store): Likewise.
37699 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
37700 Track what bytes live from the original store. Return tri-state
37701 for dead, partially dead or live.
37702 (dse_dom_walker): Add constructor, destructor and new private members.
37703 (delete_dead_call, delete_dead_assignment): New extracted from
37705 (dse_optimize_stmt): Make a member of dse_dom_walker.
37706 Use initialize_ao_ref_for_dse.
37708 PR tree-optimization/33562
37709 PR tree-optimization/61912
37710 PR tree-optimization/77485
37711 * sbitmap.h (bitmap_count_bits): Prototype.
37712 (bitmap_clear_range, bitmap_set_range): Likewise.
37713 * sbitmap.c (bitmap_clear_range): New function.
37714 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
37716 2017-01-13 Martin Liska <mliska@suse.cz>
37719 * function.c (set_cfun): Add new argument force.
37720 * function.h (set_cfun): Likewise.
37721 * ipa-inline-transform.c (inline_call): Use the function when
37722 strict alising from is dropped for function we inline to.
37724 2017-01-13 Richard Biener <rguenther@suse.de>
37726 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
37727 for dumping GIMPLE INTEGER_CSTs.
37729 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
37731 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
37732 to 201112L since C++17.
37734 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
37737 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
37738 if -fsanitize=kernel-address is present.
37740 2017-01-13 Richard Biener <rguenther@suse.de>
37742 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
37743 as _Literal ( type ) number in case usual suffixes do not
37744 preserve all information.
37746 2017-01-13 Richard Biener <rguenther@suse.de>
37748 PR tree-optimization/77283
37749 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
37750 and ssa-iterators.h.
37751 (is_feasible_trace): Implement a cost model based on joiner
37754 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
37757 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
37758 char or short to __float128/_Float128 directly.
37760 2017-01-12 Martin Sebor <msebor@redhat.com>
37762 to -Wformat-overflow.
37763 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
37764 (min_bytes_remaining): Same.
37765 (get_string_length): Same.
37766 (format_string): Same.
37767 (format_directive): Same.
37769 (pass_sprintf_length::handle_gimple_call): Same.
37771 2017-01-12 Jakub Jelinek <jakub@redhat.com>
37773 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
37774 info.nowrite calls with no lhs that can't throw. Return bool
37775 whether gsi_remove has been called or not.
37776 (pass_sprintf_length::handle_gimple_call): Return bool whether
37777 try_substitute_return_value called gsi_remove. Formatting fix.
37778 (pass_sprintf_length::execute): Don't use gsi_remove if
37779 handle_gimple_call returned true.
37782 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
37783 be removed due to side-effects, don't remove following barrier nor
37784 turn the successor edge into fallthru edge.
37786 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37789 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
37790 element-reversing loads and stores as not swappable.
37792 2017-01-12 Nathan Sidwell <nathan@acm.org>
37793 Nicolai Stange <nicstange@gmail.com>
37795 * combine.c (try_combine): Don't ignore result of overlap checking
37796 loop. Combine overlap & asm check into single loop.
37798 2017-01-12 Richard Biener <rguenther@suse.de>
37800 * tree-pretty-print.c (dump_generic_node): Provide -gimple
37801 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
37803 2017-01-12 Richard Biener <rguenther@suse.de>
37805 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
37806 and TS_TARGET_OPTION directly derive from TS_BASE.
37807 * tree-core.h (tree_optimization_option): Derive from tree_base.
37808 (tree_target_option): Likewise.
37810 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
37812 * config/i386/i386.c (memory_address_length): Increase len
37813 only when rip_relative_addr_p returns false.
37815 2017-01-11 Julia Koval <julia.koval@intel.com>
37817 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
37818 (OPTION_MASK_ISA_SGX_SET): New.
37819 (ix86_handle_option): Handle OPT_msgx.
37820 * config.gcc: Added sgxintrin.h.
37821 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
37822 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
37823 * config/i386/i386.c (ix86_target_string): Add -msgx.
37825 (ix86_option_override_internal): Handle new options.
37826 (ix86_valid_target_attribute_inner_p): Add sgx.
37827 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
37828 * config/i386/i386.opt: Add msgx.
37829 * config/i386/sgxintrin.h: New file.
37830 * config/i386/x86intrin.h: Add sgxintrin.h.
37832 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37835 * fold-const.c (maybe_nonzero_address): Return 1 for function
37837 (tree_single_nonzero_warnv_p): Don't handle function local objects
37841 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
37844 2017-01-11 David Malcolm <dmalcolm@redhat.com>
37847 * opts.c: Include "spellcheck.h"
37848 (struct string_fragment): New struct.
37849 (struct edit_distance_traits<const string_fragment &>): New
37851 (get_closest_sanitizer_option): New function.
37852 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
37854 2017-01-11 Jakub Jelinek <jakub@redhat.com>
37856 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
37858 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
37859 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
37860 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
37861 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
37862 for initial die_offset if dwarf_split_debug_info.
37863 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
37864 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
37866 (output_skeleton_debug_sections): Formatting fix. Use
37867 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
37868 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
37870 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
37872 * config/arm/cortex-a53.md: Add bypasses for
37873 cortex_a53_r2f_cvt.
37874 (cortex_a53_r2f): Only use for transfers.
37875 (cortex_a53_f2r): Likewise.
37876 (cortex_a53_r2f_cvt): Add reservation for conversions.
37877 (cortex_a53_f2r_cvt): Likewise.
37879 2017-01-11 Tamar Christina <tamar.christina@arm.com>
37881 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
37882 to all inlined functions, change static to extern.
37884 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37887 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37889 (arm_assemble_integer): Likewise.
37891 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37893 * config.gcc: Use new awk script to check CPU, FPU and architecture
37894 parameters for --with-... options.
37895 * config/arm/parsecpu.awk: New file
37896 * config/arm/arm-cpus.in: New file.
37897 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37899 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37901 * config/arm/t-arm: Update dependency rules.
37902 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37903 of processing .def files.
37904 * config/arm/genopt.sh: Deleted.
37905 * config/arm/gentune.sh: Deleted.
37906 * config/arm/arm-cores.def: Deleted.
37907 * config/arm/arm-arches.def: Deleted.
37908 * config/arm/arm-fpus.def: Deleted.
37909 * config/arm/arm-tune.md: Regenerated.
37910 * config/arm/arm-tables.opt: Regenerated.
37911 * config/arm/arm-cpu.h: New generated file.
37912 * config/arm/arm-cpu-data.h: New generated file.
37913 * config/arm/arm-cpu-cdata.h: New generated file.
37915 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37918 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37920 (input_varpool_node): Unpack dynamically_initialized bit.
37922 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37924 PR rtl-optimization/79032
37925 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37926 the alignment of the adjusted memory reference against that of MODE,
37927 instead of the alignment of the original memory reference.
37929 2017-01-11 Martin Jambor <mjambor@suse.cz>
37931 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37933 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37934 decorated functions.
37936 2017-01-11 Richard Biener <rguenther@suse.de>
37938 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37939 set range/nonnull info for PHI results. Do not set it on
37940 stmts marked for removal.
37942 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37944 * expr.c (store_field): In the bitfield case, fetch the return value
37945 from the registers before applying a single big-endian adjustment.
37946 Always do a final load for a BLKmode value not larger than a word.
37948 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37951 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37952 that we correctly handle column numbers greater than
37953 LINE_MAP_MAX_COLUMN_NUMBER.
37955 2017-01-10 Martin Sebor <msebor@redhat.com>
37957 PR middle-end/78245
37958 * gimple-ssa-sprintf.c (get_destination_size): Call
37959 {init,fini}object_sizes.
37960 * tree-object-size.c (addr_object_size): Adjust.
37961 (pass_through_call): Adjust.
37962 (pass_object_sizes::execute): Adjust.
37963 * tree-object-size.h (fini_object_sizes): Declare.
37965 2017-01-10 Martin Sebor <msebor@redhat.com>
37967 PR tree-optimization/78775
37968 * builtins.c (get_size_range): Move...
37969 * calls.c: ...to here.
37970 (alloc_max_size): Accept zero argument.
37971 (operand_signed_p): Remove.
37972 (maybe_warn_alloc_args_overflow): Call get_size_range.
37973 * calls.h (get_size_range): Declare.
37975 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37977 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37978 from TI's devices.csv file as of September 2016.
37979 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37981 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37983 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37984 * doc/invoke.texi: Likewise.
37985 * doc/md.texi: Likewise.
37986 * doc/objc.texi: Likewise.
37988 2017-01-10 Joshua Conner <joshconner@google.com>
37990 * config/arm/fuchsia-elf.h: New file.
37991 * config/fuchsia.h: New file.
37992 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37993 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37995 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37997 2016-01-10 Richard Biener <rguenther@suse.de>
37999 PR tree-optimization/79034
38000 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38001 Propagate out degenerate PHIs in the joiner.
38003 2017-01-10 Martin Liska <mliska@suse.cz>
38005 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38006 (sort_congruence_classes_by_decl_uid): Likewise.
38007 (sort_congruence_class_groups_by_decl_uid): Likewise.
38008 (sem_item_optimizer::merge_classes): Sort class, groups in these
38009 classes and members in the groups by DECL_UID of declarations.
38010 This would make merge operations stable.
38012 2017-01-10 Martin Liska <mliska@suse.cz>
38014 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38015 usage of m_classes_vec.
38016 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38017 (sem_item_optimizer::get_group_by_hash): Likewise.
38018 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38019 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38020 (sem_item_optimizer::verify_classes): Likewise.
38021 (sem_item_optimizer::process_cong_reduction): Likewise.
38022 (sem_item_optimizer::dump_cong_classes): Likewise.
38023 (sem_item_optimizer::merge_classes): Likewise.
38024 * ipa-icf.h (congruence_class_hash): Rename from
38025 congruence_class_group_hash. Remove declaration of m_classes_vec.
38027 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38029 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38030 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38031 * config.gcc: Add avx512vpopcntdqintrin.h.
38032 * config/i386/avx512vpopcntdqintrin.h: New.
38033 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38034 * config/i386/i386-builtin-types.def: Add new types.
38035 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38036 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38037 __builtin_ia32_vpopcountq_v8di_mask): New.
38038 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38039 __AVX512VPOPCNTDQ__.
38040 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38041 (PTA_AVX512VPOPCNTDQ): Define.
38042 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38043 TARGET_AVX512VPOPCNTDQ_P): Define.
38044 * config/i386/i386.opt: Add mavx512vpopcntdq.
38045 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38046 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38048 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38050 PR middle-end/77484
38051 * predict.def (PRED_CALL): Set to 67.
38053 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38055 * expr.c (store_field): In the bitfield case, if the value comes from
38056 a function call and is of an aggregate type returned in registers, do
38057 not modify the field mode; extract the value in all cases if the mode
38058 is BLKmode and the size is not larger than a word.
38060 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38063 * config/i386/cpuid.h: Fix undefined behavior.
38065 2017-01-04 Jeff Law <law@redhat.com>
38067 PR tree-optimization/79007
38068 PR tree-optimization/67955
38069 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38070 conservative for pt.null when flag_non_call_exceptions is on.
38072 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38074 PR translation/79019
38075 PR translation/79020
38076 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38077 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38078 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38080 * config/avr/avr.opt (maccumulate-args): Likewise.
38081 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38082 * common.opt (freport-bug): Likewise.
38083 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38084 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38085 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38086 translatable string.
38087 * config/i386/i386.c (function_value_32): Likewise.
38088 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38089 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38091 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38092 * common/config/msp430/msp430-common.c (msp430_handle_option):
38094 * symtab.c (symtab_node::verify_base): Likewise.
38095 * opts.c (set_debug_level): Likewise.
38096 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38097 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38098 missing whitespace to translatable strings.
38099 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38100 * config/sh/superh.h: Likewise.
38101 * config/i386/xopintrin.h: Likewise.
38102 * config/i386/znver1.md: Likewise.
38103 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38104 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38105 * double-int.h (struct double_int): Likewise.
38106 * double-int.c (div_and_round_double): Likewise.
38107 * wide-int.cc: Likewise.
38108 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38109 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38110 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38111 (crossjumps_occurred): ... this.
38112 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38115 PR tree-optimization/78899
38116 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38117 returning bool return struct loop *, NULL for failure and the new
38119 (versionable_outer_loop_p): Don't version outer loop if it has
38120 dont_vectorized bit set.
38121 (tree_if_conversion): When versioning outer loop, ensure
38122 tree_if_conversion is performed also on the inner loop of the
38123 non-vectorizable outer loop copy.
38124 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38125 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38126 prevent vectorization of it.
38127 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38128 the outer loop vectorization of the non-scalar version is attempted
38129 before vectorization of the inner loop in scalar version. If
38130 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38131 vectorization of its inner loop.
38132 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38133 has 2 inner loops, rename also on edges from bb whose single pred
38134 is outer_loop->header. Fix typo in function comment.
38136 2017-01-09 Martin Sebor <msebor@redhat.com>
38139 * asan.c (asan_emit_stack_protection): Increase local buffer size
38140 to avoid snprintf truncation warning.
38142 2017-01-09 Andrew Pinski <apinski@cavium.com>
38144 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38145 to reference thunderx2t99 for the tuning structure
38146 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38148 (thunderx2t99_extra_costs): This.
38149 * config/aarch64/aarch64-tune.md: Regenerate.
38150 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38151 (vulcan_addrcost_table): This.
38152 (vulcan_regmove_cost): Rename to ...
38153 (thunderx2t99_regmove_cost): This.
38154 (vulcan_vector_cost): Rename to ...
38155 (thunderx2t99_vector_cost): this.
38156 (vulcan_branch_cost): Rename to ...
38157 (thunderx2t99_branch_cost): This.
38158 (vulcan_tunings): Rename to ...
38159 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38160 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38162 2017-01-09 Martin Jambor <mjambor@suse.cz>
38166 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38167 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38168 (propagate_vr_accross_jump_function): Use the above function for all
38169 value range computations for pass-through jump functions and type
38170 converasion from explicit value range values.
38171 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38172 parameters from TYPE_ARG_TYPES.
38173 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38174 (ipa_write_node_info): Stream type of the actual argument.
38175 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38177 2017-01-09 Martin Liska <mliska@suse.cz>
38180 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38181 (lookup_compiler): Do not show error message with have_E.
38183 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38185 PR tree-optimization/78938
38186 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38187 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38188 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38189 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38192 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38194 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38197 2017-01-09 Richard Biener <rguenther@suse.de>
38199 PR tree-optimization/78997
38200 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38201 name condition properly.
38203 2017-01-09 Richard Biener <rguenther@suse.de>
38206 * dwarf2out.c (is_cxx): New overload with context.
38207 (is_naming_typedef_decl): Use it.
38209 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38211 * invoke.texi (Option Summary): Correct spacing in option lists
38212 and add line breaks to fix over-long lines.
38214 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38216 PR middle-end/17660
38218 * extend.texi (Common Variable Attributes): Add xref to GCC
38219 Internals manual to explain mode attribute keywords.
38221 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38224 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38225 and Preprocessor Options.
38226 (Options for Linking): Document -pthread here....
38227 (RS/6000 and PowerPC Options): ...not here.
38228 (Solaris 2 Options): ...or here.
38229 * doc/cppopts.texi: Document -pthread.
38231 2017-01-08 Martin Sebor <msebor@redhat.com>
38233 PR middle-end/77708
38234 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38235 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38236 New member functions.
38237 (format_directive): Used them.
38239 (pass_sprintf_length::handle_gimple_call): Same.
38240 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38241 to avoid truncation for any argument.
38242 (extract_affine_mul): Same.
38243 * tree.c (get_file_function_name): Same.
38245 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38247 PR middle-end/77484
38248 * predict.def (PRED_INDIR_CALL): Set to 86.
38250 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38252 PR preprocessor/54124
38253 * doc/cppopts.texi: Reformat -d subtable to list the full name
38254 of the options. Add cross-reference to the docs for the general
38255 compiler -d options.
38256 * doc/invoke.texi (Developer Options): Add cross-reference to the
38257 preprocessor-specific -d option documentation.
38259 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38261 PR preprocessor/13498
38262 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
38263 redudant material, and reflect new command-line options.
38264 (System Headers): Likewise.
38266 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
38268 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
38269 -isystem, and -idirafter. Copy-edit.
38270 * doc/cppopts.texi: Copy-edit. Remove contradiction about
38271 default for -ftrack-macro-expansion. Delete obsolete and
38272 badly-formatted implementation details about -fdebug-cpp output.
38273 * doc/cppwarnopts.texi: Copy-edit.
38275 2017-01-07 David Malcolm <dmalcolm@redhat.com>
38278 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38279 that the transition from a max line width >= 1<<10 to narrower
38280 lines works correctly.
38282 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
38284 * doc/options.texi (PerFunction): New.
38285 * opt-functions.awk (switch_flags): Map both Optimization and
38286 PerFunction to CL_OPTIMIZATION.
38287 * opth-gen.awk: Test for PerFunction flag along with
38289 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
38290 it only when the latter is present. Skip those that don't in
38291 the hash function generator.
38292 * common.opt (fvar-tracking): Mark as PerFunction instead of
38294 (fvar-tracking-assignments): Likewise.
38295 (fvar-tracking-assignments-toggle): Likewise.
38296 (fvar-tracking-uninit): Likewise.
38298 2017-01-07 Jakub Jelinek <jakub@redhat.com>
38300 PR translation/79018
38301 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
38304 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
38307 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
38308 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
38309 TARGET_LONG_JUMP_TABLE_OFFSETS.
38310 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
38311 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
38312 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
38313 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
38314 * config/m68k/m68k.md (tablejump expander): Likewise.
38315 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
38316 TARGET_LONG_JUMP_TABLE_OFFSETS.
38317 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
38318 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
38320 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
38321 David Holsgrove <david.holsgrove@xilinx.com>
38323 * common/config/microblaze/microblaze-common.c
38324 (TARGET_EXCEPT_UNWIND_INFO): Remove.
38325 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
38327 * config/microblaze/microblaze.c (microblaze_must_save_register)
38328 (microblaze_expand_epilogue, microblaze_return_addr): Handle
38330 (microblaze_eh_return): New function.
38331 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
38332 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
38333 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
38334 * config/microblaze/microblaze.md (eh_return): New pattern.
38336 2017-01-06 Jakub Jelinek <jakub@redhat.com>
38338 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
38339 GCC_DIAGNOSTIC_STRINGIFY): Define.
38341 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
38343 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38345 * config/arm/arm.md (<mcrr>): New.
38347 * config/arm/arm.c (arm_arch5te): New.
38348 (arm_option_override): Set arm_arch5te.
38349 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
38351 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
38352 (arm_mcrr_qualifiers): ... this. New.
38353 (MRRC_QUALIFIERS): Define to...
38354 (arm_mrrc_qualifiers): ... this. New.
38355 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
38357 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
38358 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
38359 (MRRCI, mrrc, MRRC): New.
38360 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
38361 VUNSPEC_MRRC2): New.
38363 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38365 * config/arm/arm.md (<mcr>): New.
38367 * config/arm/arm.c (arm_coproc_builtin_available): Add
38368 support for mcr, mrc, mcr2 and mrc2.
38369 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
38370 (arm_mcr_qualifiers): ... this. New.
38371 (MRC_QUALIFIERS): Define to ...
38372 (arm_mrc_qualifiers): ... this. New.
38373 (MCR_QUALIFIERS): Define to ...
38374 (arm_mcr_qualifiers): ... this. New.
38375 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
38377 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
38378 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
38379 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
38380 VUNSPEC_MRC2): New.
38382 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38384 * config/arm/arm.md (*ldc): New.
38388 * config/arm/arm.c (arm_coproc_builtin_available): Add
38389 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
38390 (arm_coproc_ldc_stc_legitimate_address): New.
38391 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
38392 'qualifier_const_pointer'.
38393 (LDC_QUALIFIERS): Define to...
38394 (arm_ldc_qualifiers): ... this. New.
38395 (STC_QUALIFIERS): Define to...
38396 (arm_stc_qualifiers): ... this. New.
38397 * config/arm/arm-protos.h
38398 (arm_coproc_ldc_stc_legitimate_address): New.
38399 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
38400 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
38401 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
38402 stc2, stcl, stc2l): New.
38403 * config/arm/constraints.md (Uz): New.
38404 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
38405 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
38406 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
38407 VUNSPEC_STC2L): New.
38409 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38411 * config/arm/arm.md (<cdp>): New.
38412 * config/arm/arm.c (neon_const_bounds): Rename this ...
38413 (arm_const_bounds): ... this.
38414 (arm_coproc_builtin_available): New.
38415 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
38416 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
38417 (CDP_QUALIFIERS): Define to...
38418 (arm_cdp_qualifiers): ... this. New.
38420 (arm_expand_builtin_args): Add case for 6 arguments.
38421 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
38422 (arm_const_bounds): ... this.
38423 (arm_coproc_builtin_available): New.
38424 * config/arm/arm_acle.h (__arm_cdp): New.
38426 * config/arm/arm_acle_builtins.def (cdp): New.
38428 * config/arm/iterators.md (CDPI,CDP,cdp): New.
38429 * config/arm/neon.md: Rename all 'neon_const_bounds' to
38430 'arm_const_bounds'.
38431 * config/arm/types.md (coproc): New.
38432 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
38433 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
38434 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
38435 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
38437 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38439 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
38440 (UBINOP_QUALIFIERS): New.
38442 (acle_builtin_data): New. Change comment.
38443 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
38444 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
38445 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
38446 arm_acle_builtins.def.
38447 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
38448 (arm_init_acle_builtins): New.
38449 (CRC32_BUILTIN): Remove.
38450 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
38451 crc32cb, crc32ch and crc32cw.
38452 (arm_init_crc32_builtins): Remove.
38453 (arm_init_builtins): Use arm_init_acle_builtins rather
38454 than arm_init_crc32_builtins.
38455 (arm_expand_acle_builtin): New.
38456 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
38457 * config/arm/arm_acle_builtins.def: New.
38459 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
38461 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
38462 (arm_builtin_datum): ... this.
38463 (arm_init_neon_builtin): Rename to ...
38464 (arm_init_builtin): ... this. Add a new parameters PREFIX
38465 and USE_SIG_IN_NAME.
38466 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
38467 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
38468 'arm_builtin_datum'.
38469 (arm_init_vfp_builtins): Likewise.
38470 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
38471 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
38472 (arm_expand_neon_args): Rename to ...
38473 (arm_expand_builtin_args): ... this. Rename builtin_arg
38474 enum values and differentiate between ARG_BUILTIN_MEMORY
38475 and ARG_BUILTIN_NEON_MEMORY.
38476 (arm_expand_neon_builtin_1): Rename to ...
38477 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
38478 values, arm_expand_builtin_args and add bool parameter NEON.
38479 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
38480 (arm_expand_vfp_builtin): Likewise.
38481 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
38483 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38485 PR middle-end/77484
38486 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
38487 * predict.c (tree_estimate_probability_bb): Reverse direction of
38488 polymorphic call predictor.
38490 2017-01-06 David Malcolm <dmalcolm@redhat.com>
38492 * passes.c (execute_one_pass): Split out pass-skipping logic into...
38493 (determine_pass_name_match): ...this new function and...
38494 (should_skip_pass_p): ...this new function.
38496 2017-01-06 Nathan Sidwell <nathan@acm.org>
38498 * ipa-visibility.c (function_and_variable_visibility): Reformat
38499 comments and long lines. Remove extrneous if.
38500 * symtab.c (symtab_node::make_decl_local): Fix code format.
38501 (symtab_node::set_section_for_node): Fix comment typo.
38503 2017-01-06 Martin Liska <mliska@suse.cz>
38506 * lra-constraints.c: Rename invariant to lra_invariant.
38507 * predict.c (set_even_probabilities): Initialize e to NULL.
38509 2017-01-05 Martin Sebor <msebor@redhat.com>
38511 PR tree-optimization/78910
38512 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
38513 (format_integer): Correct off-by-one error in the handling
38514 of precision with negative numbers in signed conversions..
38516 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
38518 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
38520 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38522 PR tree-optimization/71016
38523 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
38524 factor_out_conditional_conversion. Formatting fix.
38525 (factor_out_conditional_conversion): Add cond_stmt argument.
38526 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
38527 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
38530 2017-01-05 David Malcolm <dmalcolm@redhat.com>
38532 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
38533 read-rtl-function.o, and selftest-rtl.o.
38534 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
38535 (selftest::aarch64_test_loading_full_dump): New function.
38536 (selftest::aarch64_run_selftests): New function.
38537 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
38538 selftest::aarch64_run_selftests.
38539 * config/i386/i386.c
38540 (selftest::ix86_test_loading_dump_fragment_1): New function.
38541 (selftest::ix86_test_loading_call_insn): New function.
38542 (selftest::ix86_test_loading_full_dump): New function.
38543 (selftest::ix86_test_loading_unspec): New function.
38544 (selftest::ix86_run_selftests): Call the new functions.
38545 * emit-rtl.c (maybe_set_max_label_num): New function.
38546 * emit-rtl.h (maybe_set_max_label_num): New decl.
38547 * function.c (instantiate_decls): Guard call to
38548 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
38549 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
38551 * gensupport.c (gen_reader::gen_reader): Pass "false"
38552 for new "compact" param of rtx_reader.
38553 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
38554 rather than an empty string for NULL strings.
38555 * read-md.c: Potentially include config.h rather than bconfig.h.
38556 Wrap include of errors.h with #ifdef GENERATOR_FILE.
38557 (have_error): New global, copied from errors.c.
38558 (md_reader::read_name): Rename to...
38559 (md_reader::read_name_1): ...this, adding "out_loc" param,
38560 and converting "missing name or number" to returning false, rather
38562 (md_reader::read_name): Reimplement in terms of read_name_1.
38563 (md_reader::read_name_or_nil): New function.
38564 (md_reader::read_string): Handle "(nil)" by returning NULL.
38565 (md_reader::md_reader): Add new param "compact".
38566 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
38567 (md_reader::read_file): New method.
38568 * read-md.h (md_reader::md_reader): Add new param "compact".
38569 (md_reader::read_file): New method.
38570 (md_reader::is_compact): New accessor.
38571 (md_reader::read_name): Convert return type from void to file_location.
38572 (md_reader::read_name_or_nil): New decl.
38573 (md_reader::read_name_1): New decl.
38574 (md_reader::m_compact): New field.
38575 (noop_reader::noop_reader): Pass "false" for new "compact" param
38577 (rtx_reader::rtx_reader): Add new "compact" param.
38578 (rtx_reader::read_rtx_operand): Make virtual and convert return
38579 type from void to rtx.
38580 (rtx_reader::read_until): New decl.
38581 (rtx_reader::handle_any_trailing_information): New virtual function.
38582 (rtx_reader::postprocess): New virtual function.
38583 (rtx_reader::finalize_string): New virtual function.
38584 (rtx_reader::m_in_call_function_usage): New field.
38585 (rtx_reader::m_reuse_rtx_by_id): New field.
38586 * read-rtl-function.c: New file.
38587 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
38588 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
38589 (selftest::verify_three_block_rtl_cfg): New decl.
38590 * read-rtl-function.h: New file.
38591 * read-rtl.c: Potentially include config.h rather than bconfig.h.
38592 For host, include function.h, memmodel.h, and emit-rtl.h.
38593 (one_time_initialization): New function.
38594 (struct compact_insn_name): New struct.
38595 (compact_insn_names): New array.
38596 (find_code): Handle insn codes in compact dumps.
38597 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
38598 (bind_subst_iter_and_attr): Likewise.
38599 (add_condition_to_string): Likewise.
38600 (add_condition_to_rtx): Likewise.
38601 (apply_attribute_uses): Likewise.
38602 (add_current_iterators): Likewise.
38603 (apply_iterators): Likewise.
38604 (initialize_iterators): Guard usage of apply_subst_iterator with
38605 #ifdef GENERATOR_FILE.
38606 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
38607 (md_reader::read_mapping): Likewise.
38608 (add_define_attr_for_define_subst): Likewise.
38609 (add_define_subst_attr): Likewise.
38610 (read_subst_mapping): Likewise.
38611 (check_code_iterator): Likewise.
38612 (rtx_reader::read_rtx): Likewise. Move one-time initialization
38614 (one_time_initialization): New function.
38615 (rtx_reader::read_until): New method.
38616 (read_flags): New function.
38617 (parse_reg_note_name): New function.
38618 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
38619 Handle reuse_rtx ids.
38620 Wrap iterator lookup within #ifdef GENERATOR_FILE.
38621 Add parsing support for RTL dumps, mirroring the special-cases in
38622 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
38623 values, and calling handle_any_trailing_information.
38624 (rtx_reader::read_rtx_operand): Convert return type from void
38625 to rtx, returning return_rtx. Handle case 'e'. Call
38626 finalize_string on XSTR and XTMPL fields.
38627 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
38628 "(nil)" values were omitted. Call the postprocess vfunc on the
38630 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
38631 class ctor. Initialize m_in_call_function_usage. Call
38632 one_time_initialization.
38633 * rtl-tests.c (selftest::test_uncond_jump): Call
38634 set_new_first_and_last_insn.
38635 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
38636 * selftest-rtl.c: New file.
38637 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
38638 (selftest::get_insn_by_uid): New decl.
38639 * selftest-run-tests.c (selftest::run_tests): Call
38640 read_rtl_function_c_tests.
38641 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
38642 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
38645 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
38647 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
38648 operands in a special way. Assert that pos+len <= mode precision.
38650 2017-01-05 Jakub Jelinek <jakub@redhat.com>
38652 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
38653 3 argument Alias with unlimited for the negative form.
38654 (fno-vect-cost-model): Removed.
38656 2017-01-05 Martin Liska <mliska@suse.cz>
38658 * hsa-gen.c (gen_hsa_divmod): New function.
38659 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
38661 2017-01-05 Martin Liska <mliska@suse.cz>
38664 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
38667 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38669 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
38670 small constant length operands.
38672 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38674 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
38675 between loop iterations.
38677 2017-01-05 Martin Liska <mliska@suse.cz>
38680 * gimplify.c (gimplify_decl_expr): Compare to
38681 asan_poisoned_variables instread of checking flags.
38682 (gimplify_target_expr): Likewise.
38683 (gimplify_expr): Likewise.
38684 (gimplify_function_tree): Conditionally initialize
38685 asan_poisoned_variables.
38687 2017-01-04 Jeff Law <law@redhat.com>
38689 PR tree-optimizatin/78812
38690 * rtl.h (contains_mem_rtx_p): Prototype.
38691 * ifcvt.c (containts_mem_rtx_p): Move from here to...
38692 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
38693 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
38694 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
38695 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
38697 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38699 * input.c (assert_char_at_range): Default-initialize actual_range.
38701 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38703 * df-scan.c (df_ref_create_structure): Make regno unsigned,
38704 to match the caller.
38706 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38708 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
38709 insns after final jump in test to emit dummy move.
38711 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38713 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
38714 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
38716 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
38718 * multiple_target.c (create_dispatcher_calls): Init e_next.
38719 * tree-ssa-loop-split.c (split_loop): Init border.
38720 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
38723 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
38728 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
38729 (altivec_register_operand): Do not return true if the operand
38730 contains a SUBREG mixing SImode and SFmode.
38731 (vsx_register_operand): Likewise.
38732 (vsx_reg_sfsubreg_ok): New predicate.
38733 (vfloat_operand): Do not return true if the operand contains a
38734 SUBREG mixing SImode and SFmode.
38735 (vint_operand): Likewise.
38736 (vlogical_operand): Likewise.
38737 (gpc_reg_operand): Likewise.
38738 (int_reg_operand): Likewise.
38739 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
38740 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
38741 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
38743 (rs6000_emit_move_si_sf_subreg): New helper function.
38744 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
38745 fixup SUBREGs involving SImode and SFmode.
38746 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
38747 numbers for the new peephole2 optimization.
38748 (peephole2 for SFmode unions): New peephole2 to optimize cases in
38749 the GLIBC math library that do AND/IOR/XOR operations on single
38750 precision floating point.
38751 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
38752 target macros to say whether we need to avoid SUBREGs mixing
38754 (TARGET_ALLOW_SF_SUBREG): Likewise.
38755 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
38756 (UNSPEC_SI_FROM_SF): Likewise.
38757 (iorxor): Change spacing.
38758 (and_ior_xor): New iterator for AND, IOR, and XOR.
38759 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
38760 (movdi_from_sf_zero_ext): Likewise.
38761 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
38762 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
38763 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
38764 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
38765 (fms<mode>4): Likewise.
38766 (fnma<mode>4): Likewise.
38767 (fnms<mode>4): Likewise.
38768 (nfma<mode>4): Likewise.
38769 (nfms<mode>4): Likewise.
38771 2017-01-04 Marek Polacek <polacek@redhat.com>
38774 * doc/invoke.texi: Document -Wpointer-compare.
38776 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38778 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
38781 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
38782 descriptions for -gdwarf-5 and emit them as uleb128 instead of
38785 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
38788 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
38789 documentation of the powerpc_popcntb_ok attribute.
38790 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
38791 code to issue warning messages if a requested CPU configuration is
38792 not supported by the binary (assembler and loader) toolchain.
38793 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
38794 made to define a built-in function that has been disabled.
38795 (paired_init_builtins): Add assertion to prevent ICE if attempt is
38796 made to define a built-in function that has been disabled.
38797 (altivec_init_builtins): Add comment explaining why definition
38798 of the DST built-in functions is not preceded by an assertion
38799 check. Add assertions to prevent ICE if attempts are made to
38800 define an altivec predicate or an abs* built-in function that has
38802 (htm_init_builtins): Add comment explaining why definition of the
38803 htm built-in functions is not preceded by an assertion check.
38805 2017-01-04 Jeff Law <law@redhat.com>
38807 PR tree-optimizatin/67955
38808 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
38809 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
38810 the points-to solution does not include pt_null. Use DECL_PT_UID
38813 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
38815 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
38816 Use gen_int_mode instead of gen_lopwart for const_int operands.
38818 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38820 PR tree-optimization/71563
38821 * match.pd: Simplify X << Y into X if Y is known to be 0 or
38822 out of range value - has low bits known to be zero.
38824 2017-01-04 Alan Modra <amodra@gmail.com>
38826 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
38827 * configure: Regenerate.
38828 * config.in: Regenerate.
38830 2017-01-04 Jakub Jelinek <jakub@redhat.com>
38833 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
38834 a substring of the message, but strcmp with the whole message. Ifdef
38835 ENABLE_NLS, translate the message first using dgettext.
38837 2017-01-03 Jeff Law <law@redhat.com>
38839 PR tree-optimizatin/78856
38840 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
38841 (mark_threaded_blocks): Remove code to truncate thread paths that
38842 cross multiple loop headers. Instead invalidate the cached loop
38843 iteration information and handle case of a thread path walking
38844 into an irreducible region.
38846 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
38849 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
38850 assertions. Add support for doing the signbit if the IEEE 128-bit
38851 floating point value is in a GPR.
38852 * config/rs6000/rs6000.md (Fsignbit): Delete.
38853 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
38854 Update the length attribute if the value is in a GPR.
38855 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
38856 the sign or zero extension instruction, since the value is always 0/1.
38857 (signbit<mode>2_dm2): Delete using <Fsignbit>.
38860 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
38861 extracting SImode to a GPR register so that we can generate a
38862 store, limit the vector to be in a traditional Altivec register
38863 for the vextuwrx instruction.
38865 2017-01-03 Ian Lance Taylor <iant@google.com>
38867 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
38869 2017-01-03 Martin Sebor <msebor@redhat.com>
38871 PR tree-optimization/78696
38872 * gimple-ssa-sprintf.c (format_floating): Correct handling of
38873 precision. Use MPFR for %f for greater fidelity. Correct handling
38875 (pass_sprintf_length::compute_format_length): Set width and precision
38876 specified by asrerisk to void_node for vararg functions.
38877 (try_substitute_return_value): Adjust dump output.
38879 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
38881 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
38883 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
38885 * doc/invoke.texi (SPARC options): Document -mlra as the default.
38886 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38887 -mlra/-mno-lra was passed to the compiler.
38889 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38891 PR rtl-optimization/65618
38892 * emit-rtl.c (try_split): Move initialization of "before" and
38893 "after" to just before the call to emit_insn_after_setloc.
38895 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38897 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38899 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38901 * dwarf2out.c (gen_enumeration_type_die): When
38902 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38904 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38906 PR tree-optimization/78965
38907 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38908 Change first argument from const call_info & to call_info &. For %n
38909 set info.nowrite to false.
38911 PR middle-end/78901
38912 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38913 possibly throwing calls.
38915 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38916 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38917 and fns handling, rather than in a separate case SSA_NAME.
38919 2017-01-02 Jeff Law <law@redhat.com>
38921 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38922 fixes for first_period and second_period variables.
38924 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38927 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38928 (*insvqi_1): New insn pattern.
38929 (*insvqi_1_mem_rex64): Ditto.
38930 (*insvqi_2): Ditto.
38931 (*insvqi_3): Rename from *insvqi.
38933 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38935 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38937 * doc/cfg.texi (Edges): Remove reference to Java.
38938 (Maintaining the CFG): Ditto.
38940 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38942 PR middle-end/77674
38943 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38944 transparent aliases.
38946 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38948 PR middle-end/77484
38949 * predict.def (PRED_CALL): Update hitrate.
38950 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38951 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38952 into direct/indirect/polymorphic variants.
38954 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38956 Update copyright years.
38958 * gcc.c (process_command): Update copyright notice dates.
38959 * gcov-dump.c (print_version): Ditto.
38960 * gcov.c (print_version): Ditto.
38961 * gcov-tool.c (print_version): Ditto.
38962 * gengtype.c (create_file): Ditto.
38963 * doc/cpp.texi: Bump @copying's copyright year.
38964 * doc/cppinternals.texi: Ditto.
38965 * doc/gcc.texi: Ditto.
38966 * doc/gccint.texi: Ditto.
38967 * doc/gcov.texi: Ditto.
38968 * doc/install.texi: Ditto.
38969 * doc/invoke.texi: Ditto.
38971 Copyright (C) 2017 Free Software Foundation, Inc.
38973 Copying and distribution of this file, with or without modification,
38974 are permitted in any medium without royalty provided the copyright
38975 notice and this notice are preserved.